spike: the verdict, the table, the chart -- in-process pywayland wins on zwlr-screencopy

Four routes measured on one frozen contract at one session's container size, with
RAM and CPU as first-class results: in-process pywayland is ~1.5x faster (78 vs
117 ms), uses ~2.8x less of our own CPU, and nearly halves the compositor's
per-frame CPU (67 vs 116 ms). grim-rs buys nothing and costs 746 MB RSS.
ext-image-copy-capture measures identical to screencopy and appears in a
compositor crash stack, so the deprecated-but-working protocol is the pick.

Also fixes the client's unclean exit (pywayland teardown segfaulted after the
result was printed, 16 times, one batch coinciding with a compositor crash).
This commit is contained in:
Hermenix committed 2026-10-11 17:08:13 +00:00
1 parent bf0667303e
commit 36efa316fd
93 files changed
+2461 -91

No files matched your search

+136 -77
View File
@@ -1,104 +1,163 @@
# capture-speed — can we drop `grim`, and what would it buy?
**Question** (the operator's): which approach renders images faster for live
**Question** (the operator's): which capture route renders images faster for live
streaming the workspaces to the web and for agent eyes — the shipped `grim`
subprocess, the Rust `grim-rs` crate, or a raw `pywayland` client? And, since
speed alone does not decide it: **what does each cost in RAM and CPU?**
subprocess, the Rust `grim-rs` crate, or a raw `pywayland` client? And what does
each cost in RAM and CPU?
**Answer, short version**: the capture is not the expensive part — the
*round trip* is. In-process acquisition roughly doubles the frame rate and
divides the producer's own CPU by four, but on this machine both routes push the
box to saturation, because the compositor is software-rendering an animating
screen. Numbers, method and the bugs found are below.
**Answer**: in-process wins, decisively and cheaply — **~1.5× the frame rate,
~2.8× less of our own CPU, and ~42% less CPU *inside the compositor*** — and
neither Rust nor the newer capture protocol is needed to get it. `grim-rs` is no
faster than the subprocess it would replace and costs **746 MB** of RSS to grim's
18 MB.
## Method
Everything is measured through the frozen contract in `HARNESS.md`, by
`bench.py`, on one target so the cells are comparable:
One frozen contract (`HARNESS.md`), one runner (`bench.py`), one target per pass.
- **Target**: a Chromium session (`--app=http://127.0.0.1/spin.html`, a
CSS-animated page — motion with no JavaScript and no timers, so the screen
really changes and a damage-driven capture cannot go quiet and flatter a
route). Pixels verified to differ between frames before any run.
- **Cells**: one session container (1280×720) and the whole desktop stitched
(2560×720) × raw and JPEG q80 × two runs each.
- **Baseline**: `grim` as it ships — one process per frame, fork and encode
inside the timed span.
- **Target**: a Chromium session on a **CSS-animated** page (no JavaScript, no
timers) so the screen genuinely changes; frames verified to differ between
captures before any run. The still half of the contract lives in `results/still/`
and `results/final-still/` (partial — see "What went wrong").
- **Cells**: one session's container at 1280×720, `raw` and `jpeg q80`, **200
measured frames per run, 2 runs each** (the contract asks 300; 200 is stated
here rather than hidden).
- **Baseline**: `grim` exactly as it ships — one process per frame, fork and
encode inside the timed span.
- **Resources**: producer CPU/RSS (`getrusage`), machine CPU (`/proc/stat`), the
**compositor's own** CPU (`/proc/<Hyprland>/stat`), plus an **idle control**
sample taken with nothing capturing at all — without it a per-frame CPU number
would silently include the animation's cost and be attributed to the capture.
- **No GPU on this machine.** Hyprland renders through software
(kms_swrast/llvmpipe), so no route was tuned to a dma-buf path and none is
claimed. Each producer says in its own result which protocol actually served
its frames.
**compositor's own** CPU (`/proc/<Hyprland>/stat`), and an idle **control**
sample with nothing capturing, so an animating page's cost is not attributed
to a route.
- **No GPU**: Hyprland renders through software (kms_swrast/llvmpipe). No route
was tuned to a dma-buf path, and none is claimed. Each producer reports which
protocol actually served its frames.
## Head-to-head — 1280×720, moving screen, one session, sessions run sequentially
| route | mode | p50 ms | fps | own CPU ms | machine CPU ms | **compositor CPU ms** | machine % | **RSS MB** |
|---|---|---|---|---|---|---|---|---|
| `grim` (subprocess, ships today) | jpeg | 117.3 | 8.1 | 5.03 | 247.9 | **116.4** | 199.1 | **18.5** |
| `grim` | raw | 121.4 | 7.9 | 3.86 | 255.2 | 118.4 | 199.4 | 20.5 |
| `grim-rs` (Rust crate) | jpeg | 124.6 | 7.5 | 6.60 | 266.7 | 123.1 | 199.2 | **746.3** |
| `grim-rs` | raw | 120.1 | 7.8 | 1.59 | 254.1 | 119.2 | 198.9 | 746.1 |
| **`pywayland` screencopy** | jpeg | **78.0** | **13.9** | **1.78** | **143.0** | **67.5** | 199.1 | 29.3 |
| `pywayland` screencopy | raw | 76.7 | 14.2 | 0.33 | 140.4 | 67.1 | 198.7 | 24.4 |
| `pywayland` ext-image-copy-capture | jpeg | 77.7 | 14.0 | 1.79 | 142.2 | 67.2 | 199.3 | 29.1 |
| `pywayland` ext-image-copy-capture | raw | 73.5 | 14.4 | 0.31 | 138.5 | 66.2 | 198.9 | 24.5 |
Idle control — nothing capturing at all: **199.1% of one core**, the compositor
alone at 94.1%. Frames are 105–130 KB as JPEG, 2.76 MB as raw PPM.
### What the table says
1. **In-process is ~1.5× faster and much cheaper**: 78 ms vs 117 ms, 13.9 fps vs
8.1, our own CPU 1.8 ms vs 5.0 ms per frame.
2. **The win that matters is the compositor's, not ours**: 67 ms vs 116 ms of
compositor CPU *per frame*. On a 2-core box that is the difference between the
desktop being starved and merely busy.
3. **`grim-rs` is not a candidate**: slightly slower than the subprocess (124.6
vs 117.3 ms), *higher* own CPU (6.6 vs 5.0 ms), and **746 MB RSS** — 40× grim.
Its own notes explain the speed: it allocates a fresh shm buffer and copies
into a `Vec` per capture. The memory figure is reproducible across runs.
4. **screencopy vs ext-image-copy-capture: no measurable difference** (78.0 vs
77.7 ms; compositor 67.5 vs 67.2 ms). The newer protocol buys nothing here, and
its frame-creation path is where one compositor crash landed — so the
recommendation is the older, deprecated-but-working protocol, which is also
what `grim` uses (apples to apples).
5. **The box is saturated in every cell (≈199% of one core)** — and the idle
control is *also* 199%. Rendering the animating screen, not capturing it, is
what fills this machine. No capture change fixes that; what changes is how much
room the compositor has left.
## Where the time actually goes (the finding that reframes it)
`grim` at 1280×720 JPEG q80: **p50 119 ms per frame, 7.9–8.3 fps, but only
6–8 ms of CPU.** About 95% of each frame is spent waiting — on the fork, the
connect, the compositor's round trip — not encoding. Two consequences:
`grim` at 1280×720 JPEG q80 spends ~95% of each frame **waiting** — on the fork,
the connect, the compositor's round trip — not encoding: JPEG q80 of a 1280×720
frame costs **1.7 ms** in-process. So the available win is in *not paying the
round trip per frame*, which is exactly what a persistent in-process client does —
and why screencopy and ext-image-copy-capture land within noise of each other.
1. Attacking the *encoder* is pointless: JPEG q80 of a 1280×720 frame costs
**1.7 ms** in-process.
2. The win available is in **not paying the round trip per frame** — which is
exactly what a persistent in-process client does, and why the two new routes
land so close to each other.
## What went wrong (each item cost real time; none is hidden)
And the resource measurement adds the part speed hides: at 1280×720 with motion
on screen, **the machine is already saturated (≈199% of one core, i.e. both
cores) whichever route is used**, because the compositor is repainting a
software-rendered animated page. The capture route changes the *compositor's*
per-frame CPU materially (see the table), but it cannot make the box idle.
- **The compositor crashed three times** during measurement (`.Hyprland-wrapped`
SEGV in `CScreenshareFrame::transform()` reached via
`ext-image-copy-capture`'s `create_frame`, plus twice in `libaquamarine`),
taking the stream/sessions/pointer units and both session outputs with it. When
that happened, cells failed and `hhstream status` hung for minutes. **The ext
stack could not be reproduced in isolation** (5 clean frames afterwards), so
this is recorded as *unstable under sustained capture load*, not as "the modern
protocol is broken".
- **Our own client segfaulted on exit 16 times** (`python3.14 … segfault …
libwayland-client.so`) — pywayland/CFFI teardown, *after* the result printed, so
green runs reported exit 139. Fixed with `os._exit(0)` after flushing. A batch
of these coincided in time with a compositor crash, so the fix may matter
beyond tidiness.
- **Two SIGTERM kills of the benchmark** (exit 143, no error line): a 40-minute
`nix-shell` tree does not belong in an agent background process. Moved to a
transient systemd unit, and the runner made resumable.
- **`grim -o NAME` must precede the trailing `-`**, or grim exits 0 having written
**637 bytes of nothing**.
- **grim's only uncompressed format is `ppm`**; `pbm`/`pnm`/`rgb`/`bmp` are
rejected. Raw is *not* cheaper than JPEG for it — writing 2.76 MB down a pipe
costs what encoding costs.
- **turbojpeg reuses a non-NULL `*jpegBuf`**: freeing it and leaving the pointer
makes the next call write into freed memory (heap-corrupting segfault with no
stack trace near the encoder).
- **`display.dispatch(block=False)` never reads the socket** — it processes only
what is already read, so a wait built on it times out against a healthy
compositor. The read is `dispatch(block=True)`.
- **`ext-image-capture-source-v1` has one manager per source kind**
(`ext_output_…`, `ext_foreign_toplevel_…`). Looking for a generic
`ext_image_capture_source_manager_v1` finds nothing and reads as "the modern
protocol is unavailable" — a wrong conclusion this spike drew once. The
toplevel variant means **single-window capture is available**, which matters
for agent eyes.
- **`--all` cells silently changed geometry** (2560→3840) once a second session
existed, so they are excluded from this table rather than averaged.
- **A `--resume` keyed on file existence skips failed cells too**, because a
failure also writes a file; it now requires a parseable result line.
## Head-to-head
## Verdict: VALIDATED — with a caveat that changes the shape of the answer
<!-- TABLE-PLACEHOLDER: filled by bench.py results, not by hand -->
- **Replace `grim` with an in-process `pywayland` client on `zwlr-screencopy`**
for the one-session case (live streaming a workspace, agent eyes). ~1.5×
faster, ~2.8× less own CPU, and — the number that decides it — **the
compositor's per-frame CPU nearly halves**.
- **Do not adopt `grim-rs`**: no speed win, 40× the memory.
- **Do not adopt `ext-image-copy-capture` yet** on this Hyprland build: no
measured gain, and its create-frame path appears in a compositor crash stack.
- **Keep raw/ppm out of the wire path**: it is not cheaper than JPEG anywhere
measured.
- **The caveat**: on this software-rendered 2-core VM the machine is saturated by
*rendering* whatever is on screen, in every configuration including idle. The
frame-rate win is real, but no capture change buys headroom — it changes how
much of a saturated box the compositor keeps.
## Bugs and traps found on the way (each cost real time)
## What a production version would take
1. **`grim -o NAME` must come *before* the trailing `-`.** The other order exits
0 and writes **637 bytes of nothing** to the pipe. Silent, and it cost the
first harness build.
2. **grim's only uncompressed format is `ppm`** (`pbm`/`pnm`/`rgb`/`bmp` are
rejected), which is what makes the encode/capture split measurable.
3. **`display.dispatch(block=False)` never reads the socket** — it processes only
events already read. A wait loop built on it plus `select` times out against a
perfectly healthy compositor. The read is `dispatch(block=True)`; the "never
hang" guarantee has to come from an alarm (`SIGALRM`), not from polling.
4. **`pywayland` dispatchers receive the proxy as their first argument** (four
parameters for a three-field event), and **`display.disconnect()` is
mandatory** or the process segfaults *after* printing — a green run that
reports exit 139 to any runner reading exit codes.
5. **turbojpeg reuses `*jpegBuf` when it is non-NULL.** Freeing it with `tjFree`
and leaving the pointer in place makes the next call write into freed memory:
a heap-corrupting segfault several frames later with no stack trace near the
encoder. Reset the pointer to NULL after freeing.
6. **`libjpeg_turbo`'s bare attribute resolves to its `-bin` output** in that
Nix shell — no shared library in it — and `wayland.xml` lives in the
**`wayland-scanner`** package, not in `wayland`. Both failures look like
missing dependencies rather than shell gaps.
7. **`ext-image-capture-source-v1` defines one manager per source kind**
(`ext_output_…` and `ext_foreign_toplevel_…`), not a generic
`ext_image_capture_source_manager_v1`. Looking for the generic name finds
nothing and reads as "the modern protocol is unavailable here" — a wrong
conclusion this spike drew once before checking the real global list. The
toplevel variant means **single-window capture is available** on this
compositor, which matters for agent eyes.
- Replace the subprocess behind `src/hhstream/capture.py`'s `CommandCapture` with
a pywayland screencopy client: ~250 lines, vendored protocol XML (the pattern
already used by `src/hhstream_pointer/_protocol/`), and **no new dependency** —
`pywayland` is already a dependency of the package, so the stdlib-only policy
for the frame socket, CLI and broker is untouched.
- JPEG encoding then needs libjpeg-turbo via ctypes (stdlib `zlib` PNG is the
fallback, at 2–3× the bytes) — that *is* a genuine new dependency for that
module, and the one thing worth arguing about in review.
- Single-window capture comes free via
`ext_foreign_toplevel_image_capture_source_manager_v1`; worth a follow-up spike
once the compositor's stability under sustained capture is understood.
## Files
| File | What it is |
|---|---|
| `HARNESS.md` | the frozen measurement contract (argv, JSON, timing boundary, resource fields) |
| `bench.py` | the runner: every route through every cell, raw JSON kept, table derived from the files |
| `bench.py` | the runner: every route through every cell, raw JSON kept, table derived from the files, `--resume` |
| `producers/grim_producer.py` | route 000, the baseline |
| `producers/wayland_producer.py` | route 001b, raw `pywayland` (screencopy *and* ext-image-copy-capture) |
| `producers/resources.py` | RAM/CPU sampling shared by the Python producers |
| `001a-grim-rs/` | route 001a, the Rust `grim-rs` producer |
| `results/` | every run's stdout/stderr, `results.json`, and `control.json` |
| `shell.nix` | the spike's own environment (pywayland, turbojpeg, rust) — deliberately not in the repo flake |
## Verdict
<!-- VERDICT-PLACEHOLDER -->
| `001a-grim-rs/` | route 001a, the Rust `grim-rs` producer (Rust toolchain confined to its own `shell.nix`) |
| `run-final.sh`, `run-remaining.sh` | the pass drivers, meant to run in a transient systemd unit |
| `chart.py` | the two charts (`python3 chart.py results/final-motion`) |
| `results/final-motion/` | the table above, its chart, and every run's raw stdout/stderr |
| `shell.nix` | the spike's environment — deliberately not in the repo flake |
+13 -8
View File
@@ -147,17 +147,22 @@ def main() -> int:
label = f"{route}/{cell['name']}/run{run}"
stem = os.path.join(args.results, f"{route}.{cell['name']}.run{run}")
if args.resume and os.path.exists(stem + ".stdout"):
print(f" {label}: already measured, skipped (--resume)")
try:
with open(stem + ".stdout") as handle:
lines = [line for line in handle.read().splitlines() if line.startswith("{")]
if lines:
# A *failed* cell also leaves a stdout file, so existence is
# not the test -- a parseable result line is. Otherwise one
# failure is skipped forever and the table shows a permanent
# gap that looks like it was never attempted.
with open(stem + ".stdout") as handle:
lines = [line for line in handle.read().splitlines() if line.startswith("{")]
if lines:
print(f" {label}: already measured, skipped (--resume)")
try:
records.append({"route": route, "description": description,
"cell": cell["name"], "run": run, "rc": 0,
**json.loads(lines[-1])})
except (OSError, json.JSONDecodeError):
pass
continue
except json.JSONDecodeError:
pass
continue
print(f" {label}: previous attempt left no result, re-running")
started = time.monotonic()
proc = subprocess.run(command, capture_output=True, text=True, env=env)
elapsed = time.monotonic() - started
+2 -1
View File
@@ -27,7 +27,8 @@ matplotlib.use("Agg")
import matplotlib.pyplot as plt # noqa: E402
HERE = os.path.dirname(os.path.abspath(__file__))
RESULTS = os.path.join(HERE, "results")
# Which pass to chart: `python3 chart.py results/final-motion`
RESULTS = os.path.join(HERE, sys.argv[1] if len(sys.argv) > 1 else "results")
LABELS = {
"grim": "grim (subprocess)",
@@ -588,7 +588,15 @@ def main() -> int:
)
)
print(json.dumps(result) if args.json else json.dumps(result, indent=2))
return 0
# **Exit without running teardown.** pywayland/CFFI finalizers segfault inside
# libwayland-client at interpreter shutdown (16 of them in this spike's logs,
# all `python3.14 … segfault … libwayland-client.so`), *after* the result is
# printed — so a green run reports exit 139 and, worse, the compositor was
# observed crashing in the same second as a batch of these. The bytes are
# flushed; everything after this point is library cleanup we do not need.
sys.stdout.flush()
sys.stderr.flush()
os._exit(0)
if __name__ == "__main__":
@@ -0,0 +1,7 @@
{
"window_ms": 10000.0,
"system_cpu_ms": 19910.0,
"compositor_cpu_ms": 9410.0,
"system_cpu_pct": 199.1,
"compositor_cpu_pct": 94.1
}
@@ -0,0 +1,12 @@
grimrs-capture: route grim-rs protocol ext-image-copy-capture-v1 (auto) mode jpeg target hh-chromium-gqkxz3 warmup 10 frames 200
grimrs-capture: outputs Virtual-1@0,0 1280x720 scale 1; hh-chromium-gqkxz3@1280,0 1280x720 scale 1
grimrs-capture: warmup 1/10 1280x720 bytes 107323 nonzero 105292
grimrs-capture: warmup 2/10 1280x720 bytes 107216 nonzero 105204
grimrs-capture: warmup 3/10 1280x720 bytes 107519 nonzero 105455
grimrs-capture: warmup 4/10 1280x720 bytes 107392 nonzero 105283
grimrs-capture: warmup 5/10 1280x720 bytes 107375 nonzero 105312
grimrs-capture: warmup 6/10 1280x720 bytes 107218 nonzero 105189
grimrs-capture: warmup 7/10 1280x720 bytes 107174 nonzero 105173
grimrs-capture: warmup 8/10 1280x720 bytes 107452 nonzero 105421
grimrs-capture: warmup 9/10 1280x720 bytes 107291 nonzero 105238
grimrs-capture: warmup 10/10 1280x720 bytes 107452 nonzero 105382
@@ -0,0 +1 @@
{"route":"grim-rs","language":"rust","protocol":"ext-image-copy-capture-v1 (auto)","output":"hh-chromium-gqkxz3","geometry":[1280,720],"mode":"jpeg","warmup":10,"frames":200,"frame_ms":{"min":61.074,"p50":124.039,"mean":132.54,"p95":198.748,"max":254.625},"fps_sustained":7.545,"wall_ms":26508.219,"cpu_ms_per_frame":6.6,"bytes_p50":107205.5,"persistent":true,"rss_peak_mb":746.25,"rss_peak_kb_raw":764156,"system_cpu_ms_per_frame":264.3,"compositor_cpu_ms_per_frame":123.1,"machine_cpu_pct":199.4,"frames_bytes_allocated_p50":3686400,"notes":"grim-rs 0.2.0 via Grim::new() (auto-detect); copy target is a wl_shm buffer created inside each capture_* call, not a dma-buf the producer imports. Cursor not overlaid (the frozen argv has no cursor flag; hhstream-live's grim command passes -c). JPEG is Grim::to_jpeg, quality 80, the crate default and the live-stream setting. Raw timer stops when capture_* returns, which is after the crate copies the mapping into a Vec and swizzles XRGB/ARGB to RGBA (and applies any output transform or y-invert). The public API does not expose the mapped buffer before that, so raw includes work the contract says to leave outside the timer. Percentiles are linear interpolation on the sorted samples. cpu_ms_per_frame is getrusage(RUSAGE_SELF) utime+stime over the measured loop only, divided by frames. Frame buffers are freed between frames; that free is inside wall_ms and outside frame_ms. Outputs: Virtual-1@0,0 1280x720 scale 1; hh-chromium-gqkxz3@1280,0 1280x720 scale 1. No GPU on this machine; dma-buf was not used and is not claimed. This process does not move the pointer and does not start or stop a session; the pixels are whatever the compositor is already showing. Last warmup frame nonzero bytes: 105382."}
@@ -0,0 +1,12 @@
grimrs-capture: route grim-rs protocol ext-image-copy-capture-v1 (auto) mode jpeg target hh-chromium-gqkxz3 warmup 10 frames 200
grimrs-capture: outputs Virtual-1@0,0 1280x720 scale 1; hh-chromium-gqkxz3@1280,0 1280x720 scale 1
grimrs-capture: warmup 1/10 1280x720 bytes 107394 nonzero 105367
grimrs-capture: warmup 2/10 1280x720 bytes 107137 nonzero 105116
grimrs-capture: warmup 3/10 1280x720 bytes 107298 nonzero 105216
grimrs-capture: warmup 4/10 1280x720 bytes 107092 nonzero 105064
grimrs-capture: warmup 5/10 1280x720 bytes 107031 nonzero 105045
grimrs-capture: warmup 6/10 1280x720 bytes 107334 nonzero 105280
grimrs-capture: warmup 7/10 1280x720 bytes 107182 nonzero 105134
grimrs-capture: warmup 8/10 1280x720 bytes 107395 nonzero 105345
grimrs-capture: warmup 9/10 1280x720 bytes 107333 nonzero 105326
grimrs-capture: warmup 10/10 1280x720 bytes 107105 nonzero 105080
@@ -0,0 +1 @@
{"route":"grim-rs","language":"rust","protocol":"ext-image-copy-capture-v1 (auto)","output":"hh-chromium-gqkxz3","geometry":[1280,720],"mode":"jpeg","warmup":10,"frames":200,"frame_ms":{"min":59.907,"p50":125.214,"mean":135.259,"p95":207.095,"max":324.014},"fps_sustained":7.393,"wall_ms":27051.989,"cpu_ms_per_frame":6.595,"bytes_p50":107203.5,"persistent":true,"rss_peak_mb":746.34,"rss_peak_kb_raw":764256,"system_cpu_ms_per_frame":269.1,"compositor_cpu_ms_per_frame":123.05,"machine_cpu_pct":199.0,"frames_bytes_allocated_p50":3686400,"notes":"grim-rs 0.2.0 via Grim::new() (auto-detect); copy target is a wl_shm buffer created inside each capture_* call, not a dma-buf the producer imports. Cursor not overlaid (the frozen argv has no cursor flag; hhstream-live's grim command passes -c). JPEG is Grim::to_jpeg, quality 80, the crate default and the live-stream setting. Raw timer stops when capture_* returns, which is after the crate copies the mapping into a Vec and swizzles XRGB/ARGB to RGBA (and applies any output transform or y-invert). The public API does not expose the mapped buffer before that, so raw includes work the contract says to leave outside the timer. Percentiles are linear interpolation on the sorted samples. cpu_ms_per_frame is getrusage(RUSAGE_SELF) utime+stime over the measured loop only, divided by frames. Frame buffers are freed between frames; that free is inside wall_ms and outside frame_ms. Outputs: Virtual-1@0,0 1280x720 scale 1; hh-chromium-gqkxz3@1280,0 1280x720 scale 1. No GPU on this machine; dma-buf was not used and is not claimed. This process does not move the pointer and does not start or stop a session; the pixels are whatever the compositor is already showing. Last warmup frame nonzero bytes: 105080."}
@@ -0,0 +1,12 @@
grimrs-capture: route grim-rs protocol ext-image-copy-capture-v1 (auto) mode raw target hh-chromium-gqkxz3 warmup 10 frames 200
grimrs-capture: outputs Virtual-1@0,0 1280x720 scale 1; hh-chromium-gqkxz3@1280,0 1280x720 scale 1
grimrs-capture: warmup 1/10 1280x720 bytes 3686400 nonzero 3679565
grimrs-capture: warmup 2/10 1280x720 bytes 3686400 nonzero 3679565
grimrs-capture: warmup 3/10 1280x720 bytes 3686400 nonzero 3679565
grimrs-capture: warmup 4/10 1280x720 bytes 3686400 nonzero 3679565
grimrs-capture: warmup 5/10 1280x720 bytes 3686400 nonzero 3679565
grimrs-capture: warmup 6/10 1280x720 bytes 3686400 nonzero 3679565
grimrs-capture: warmup 7/10 1280x720 bytes 3686400 nonzero 3679565
grimrs-capture: warmup 8/10 1280x720 bytes 3686400 nonzero 3679565
grimrs-capture: warmup 9/10 1280x720 bytes 3686400 nonzero 3679565
grimrs-capture: warmup 10/10 1280x720 bytes 3686400 nonzero 3679565
@@ -0,0 +1 @@
{"route":"grim-rs","language":"rust","protocol":"ext-image-copy-capture-v1 (auto)","output":"hh-chromium-gqkxz3","geometry":[1280,720],"mode":"raw","warmup":10,"frames":200,"frame_ms":{"min":44.16,"p50":122.21,"mean":130.086,"p95":189.49,"max":226.412},"fps_sustained":7.687,"wall_ms":26017.423,"cpu_ms_per_frame":1.628,"bytes_p50":null,"persistent":true,"rss_peak_mb":746.09,"rss_peak_kb_raw":763992,"system_cpu_ms_per_frame":258.6,"compositor_cpu_ms_per_frame":120.85,"machine_cpu_pct":198.8,"frames_bytes_allocated_p50":3686400,"notes":"grim-rs 0.2.0 via Grim::new() (auto-detect); copy target is a wl_shm buffer created inside each capture_* call, not a dma-buf the producer imports. Cursor not overlaid (the frozen argv has no cursor flag; hhstream-live's grim command passes -c). JPEG is Grim::to_jpeg, quality 80, the crate default and the live-stream setting. Raw timer stops when capture_* returns, which is after the crate copies the mapping into a Vec and swizzles XRGB/ARGB to RGBA (and applies any output transform or y-invert). The public API does not expose the mapped buffer before that, so raw includes work the contract says to leave outside the timer. Percentiles are linear interpolation on the sorted samples. cpu_ms_per_frame is getrusage(RUSAGE_SELF) utime+stime over the measured loop only, divided by frames. Frame buffers are freed between frames; that free is inside wall_ms and outside frame_ms. Outputs: Virtual-1@0,0 1280x720 scale 1; hh-chromium-gqkxz3@1280,0 1280x720 scale 1. No GPU on this machine; dma-buf was not used and is not claimed. This process does not move the pointer and does not start or stop a session; the pixels are whatever the compositor is already showing. Last warmup frame nonzero bytes: 3679565."}
@@ -0,0 +1,12 @@
grimrs-capture: route grim-rs protocol ext-image-copy-capture-v1 (auto) mode raw target hh-chromium-gqkxz3 warmup 10 frames 200
grimrs-capture: outputs Virtual-1@0,0 1280x720 scale 1; hh-chromium-gqkxz3@1280,0 1280x720 scale 1
grimrs-capture: warmup 1/10 1280x720 bytes 3686400 nonzero 3679565
grimrs-capture: warmup 2/10 1280x720 bytes 3686400 nonzero 3679565
grimrs-capture: warmup 3/10 1280x720 bytes 3686400 nonzero 3679565
grimrs-capture: warmup 4/10 1280x720 bytes 3686400 nonzero 3679565
grimrs-capture: warmup 5/10 1280x720 bytes 3686400 nonzero 3679565
grimrs-capture: warmup 6/10 1280x720 bytes 3686400 nonzero 3679565
grimrs-capture: warmup 7/10 1280x720 bytes 3686400 nonzero 3679565
grimrs-capture: warmup 8/10 1280x720 bytes 3686400 nonzero 3679565
grimrs-capture: warmup 9/10 1280x720 bytes 3686400 nonzero 3679565
grimrs-capture: warmup 10/10 1280x720 bytes 3686400 nonzero 3679565
@@ -0,0 +1 @@
{"route":"grim-rs","language":"rust","protocol":"ext-image-copy-capture-v1 (auto)","output":"hh-chromium-gqkxz3","geometry":[1280,720],"mode":"raw","warmup":10,"frames":200,"frame_ms":{"min":46.164,"p50":117.913,"mean":125.337,"p95":194.097,"max":255.125},"fps_sustained":7.978,"wall_ms":25067.547,"cpu_ms_per_frame":1.551,"bytes_p50":null,"persistent":true,"rss_peak_mb":746.02,"rss_peak_kb_raw":763924,"system_cpu_ms_per_frame":249.6,"compositor_cpu_ms_per_frame":117.6,"machine_cpu_pct":199.1,"frames_bytes_allocated_p50":3686400,"notes":"grim-rs 0.2.0 via Grim::new() (auto-detect); copy target is a wl_shm buffer created inside each capture_* call, not a dma-buf the producer imports. Cursor not overlaid (the frozen argv has no cursor flag; hhstream-live's grim command passes -c). JPEG is Grim::to_jpeg, quality 80, the crate default and the live-stream setting. Raw timer stops when capture_* returns, which is after the crate copies the mapping into a Vec and swizzles XRGB/ARGB to RGBA (and applies any output transform or y-invert). The public API does not expose the mapped buffer before that, so raw includes work the contract says to leave outside the timer. Percentiles are linear interpolation on the sorted samples. cpu_ms_per_frame is getrusage(RUSAGE_SELF) utime+stime over the measured loop only, divided by frames. Frame buffers are freed between frames; that free is inside wall_ms and outside frame_ms. Outputs: Virtual-1@0,0 1280x720 scale 1; hh-chromium-gqkxz3@1280,0 1280x720 scale 1. No GPU on this machine; dma-buf was not used and is not claimed. This process does not move the pointer and does not start or stop a session; the pixels are whatever the compositor is already showing. Last warmup frame nonzero bytes: 3679565."}
Whitespace-only changes.
@@ -0,0 +1 @@
{"route": "grim-subprocess", "language": "python", "protocol": "zwlr-screencopy-v1 (grim's client)", "output": "hh-chromium-gqkxz3", "geometry": [1280, 720], "mode": "jpeg", "warmup": 10, "frames": 200, "frame_ms": {"min": 58.928, "p50": 116.179, "mean": 121.352, "p95": 164.512, "max": 200.886}, "fps_sustained": 8.24, "wall_ms": 24614.05, "bytes_p50": 130597, "persistent": false, "notes": "one grim process per frame, fork and encode inside the timed span; territory: --output; jpeg q80", "cpu_ms_per_frame": 5.005, "rss_peak_mb": 18.47, "rss_peak_kb_raw": 18916, "system_cpu_ms_per_frame": 245.05, "compositor_cpu_ms_per_frame": 115.05, "machine_cpu_pct": 199.1, "machine_cpu_pct_of_all_cores": 99.6, "frames_bytes_allocated_p50": null, "resource_notes": ""}
Whitespace-only changes.
@@ -0,0 +1 @@
{"route": "grim-subprocess", "language": "python", "protocol": "zwlr-screencopy-v1 (grim's client)", "output": "hh-chromium-gqkxz3", "geometry": [1280, 720], "mode": "jpeg", "warmup": 10, "frames": 200, "frame_ms": {"min": 67.203, "p50": 118.52, "mean": 124.224, "p95": 185.591, "max": 299.044}, "fps_sustained": 8.05, "wall_ms": 25182.14, "bytes_p50": 130590, "persistent": false, "notes": "one grim process per frame, fork and encode inside the timed span; territory: --output; jpeg q80", "cpu_ms_per_frame": 5.05, "rss_peak_mb": 18.45, "rss_peak_kb_raw": 18896, "system_cpu_ms_per_frame": 250.75, "compositor_cpu_ms_per_frame": 117.75, "machine_cpu_pct": 199.1, "machine_cpu_pct_of_all_cores": 99.6, "frames_bytes_allocated_p50": null, "resource_notes": ""}
Whitespace-only changes.
@@ -0,0 +1 @@
{"route": "grim-subprocess", "language": "python", "protocol": "zwlr-screencopy-v1 (grim's client)", "output": "hh-chromium-gqkxz3", "geometry": [1280, 720], "mode": "raw", "warmup": 10, "frames": 200, "frame_ms": {"min": 51.756, "p50": 120.321, "mean": 125.705, "p95": 190.299, "max": 253.325}, "fps_sustained": 7.96, "wall_ms": 25452.51, "bytes_p50": 2764816, "persistent": false, "notes": "one grim process per frame, fork and encode inside the timed span; territory: --output; ppm, unencoded", "cpu_ms_per_frame": 3.856, "rss_peak_mb": 20.56, "rss_peak_kb_raw": 21052, "system_cpu_ms_per_frame": 253.8, "compositor_cpu_ms_per_frame": 118.55, "machine_cpu_pct": 199.4, "machine_cpu_pct_of_all_cores": 99.7, "frames_bytes_allocated_p50": 3686400, "resource_notes": ""}
Whitespace-only changes.
@@ -0,0 +1 @@
{"route": "grim-subprocess", "language": "python", "protocol": "zwlr-screencopy-v1 (grim's client)", "output": "hh-chromium-gqkxz3", "geometry": [1280, 720], "mode": "raw", "warmup": 10, "frames": 200, "frame_ms": {"min": 64.061, "p50": 122.389, "mean": 127.307, "p95": 196.38, "max": 264.387}, "fps_sustained": 7.86, "wall_ms": 25757.6, "bytes_p50": 2764816, "persistent": false, "notes": "one grim process per frame, fork and encode inside the timed span; territory: --output; ppm, unencoded", "cpu_ms_per_frame": 3.858, "rss_peak_mb": 20.51, "rss_peak_kb_raw": 21004, "system_cpu_ms_per_frame": 256.65, "compositor_cpu_ms_per_frame": 118.25, "machine_cpu_pct": 199.3, "machine_cpu_pct_of_all_cores": 99.6, "frames_bytes_allocated_p50": 3686400, "resource_notes": ""}
Whitespace-only changes.
@@ -0,0 +1 @@
{"route": "pywayland-in-process", "language": "python", "protocol": "ext-image-copy-capture-v1", "output": "hh-chromium-gqkxz3", "geometry": [1280, 720], "mode": "jpeg", "warmup": 10, "frames": 200, "frame_ms": {"min": 30.44, "p50": 76.633, "mean": 71.153, "p95": 103.055, "max": 185.942}, "stage_ms": {"capture": 68.35, "encode": 2.708, "stitch": 0.0}, "fps_sustained": 14.05, "wall_ms": 14231.13, "bytes_p50": 105794, "persistent": true, "notes": "one connection, buffers reused, turbojpeg q80 in-process; 1 output(s); forcing a repaint per frame", "cpu_ms_per_frame": 1.811, "rss_peak_mb": 28.99, "rss_peak_kb_raw": 29688, "system_cpu_ms_per_frame": 141.75, "compositor_cpu_ms_per_frame": 66.55, "machine_cpu_pct": 199.2, "machine_cpu_pct_of_all_cores": 99.6, "frames_bytes_allocated_p50": 3686400, "resource_notes": ""}
Whitespace-only changes.
@@ -0,0 +1 @@
{"route": "pywayland-in-process", "language": "python", "protocol": "ext-image-copy-capture-v1", "output": "hh-chromium-gqkxz3", "geometry": [1280, 720], "mode": "jpeg", "warmup": 10, "frames": 200, "frame_ms": {"min": 38.593, "p50": 78.693, "mean": 71.507, "p95": 96.637, "max": 129.106}, "stage_ms": {"capture": 69.115, "encode": 2.339, "stitch": 0.0}, "fps_sustained": 13.98, "wall_ms": 14301.79, "bytes_p50": 105804, "persistent": true, "notes": "one connection, buffers reused, turbojpeg q80 in-process; 1 output(s); forcing a repaint per frame", "cpu_ms_per_frame": 1.776, "rss_peak_mb": 29.25, "rss_peak_kb_raw": 29956, "system_cpu_ms_per_frame": 142.65, "compositor_cpu_ms_per_frame": 67.9, "machine_cpu_pct": 199.5, "machine_cpu_pct_of_all_cores": 99.7, "frames_bytes_allocated_p50": 3686400, "resource_notes": ""}
Whitespace-only changes.
@@ -0,0 +1 @@
{"route": "pywayland-in-process", "language": "python", "protocol": "ext-image-copy-capture-v1", "output": "hh-chromium-gqkxz3", "geometry": [1280, 720], "mode": "raw", "warmup": 10, "frames": 200, "frame_ms": {"min": 22.685, "p50": 71.372, "mean": 69.708, "p95": 89.816, "max": 121.781}, "stage_ms": {"capture": 69.698, "encode": 0.0, "stitch": 0.0}, "fps_sustained": 14.35, "wall_ms": 13941.78, "bytes_p50": null, "persistent": true, "notes": "one connection, buffers reused, turbojpeg q80 in-process; 1 output(s); forcing a repaint per frame", "cpu_ms_per_frame": 0.311, "rss_peak_mb": 24.43, "rss_peak_kb_raw": 25020, "system_cpu_ms_per_frame": 138.6, "compositor_cpu_ms_per_frame": 65.85, "machine_cpu_pct": 198.8, "machine_cpu_pct_of_all_cores": 99.4, "frames_bytes_allocated_p50": 3686400, "resource_notes": ""}
Whitespace-only changes.
@@ -0,0 +1 @@
{"route": "pywayland-in-process", "language": "python", "protocol": "ext-image-copy-capture-v1", "output": "hh-chromium-gqkxz3", "geometry": [1280, 720], "mode": "raw", "warmup": 10, "frames": 200, "frame_ms": {"min": 41.241, "p50": 75.703, "mean": 69.499, "p95": 86.611, "max": 112.812}, "stage_ms": {"capture": 69.489, "encode": 0.0, "stitch": 0.0}, "fps_sustained": 14.39, "wall_ms": 13899.94, "bytes_p50": null, "persistent": true, "notes": "one connection, buffers reused, turbojpeg q80 in-process; 1 output(s); forcing a repaint per frame", "cpu_ms_per_frame": 0.318, "rss_peak_mb": 24.52, "rss_peak_kb_raw": 25112, "system_cpu_ms_per_frame": 138.35, "compositor_cpu_ms_per_frame": 66.45, "machine_cpu_pct": 199.1, "machine_cpu_pct_of_all_cores": 99.5, "frames_bytes_allocated_p50": 3686400, "resource_notes": ""}
@@ -0,0 +1 @@
{"route": "pywayland-in-process", "language": "python", "protocol": "zwlr-screencopy-v1", "output": "hh-chromium-gqkxz3", "geometry": [1280, 720], "mode": "jpeg", "warmup": 10, "frames": 200, "frame_ms": {"min": 25.597, "p50": 79.088, "mean": 71.771, "p95": 96.766, "max": 132.204}, "stage_ms": {"capture": 69.093, "encode": 2.522, "stitch": 0.0}, "fps_sustained": 13.93, "wall_ms": 14354.55, "bytes_p50": 105799, "persistent": true, "notes": "one connection, buffers reused, turbojpeg q80 in-process; 1 output(s); forcing a repaint per frame", "cpu_ms_per_frame": 1.781, "rss_peak_mb": 29.38, "rss_peak_kb_raw": 30084, "system_cpu_ms_per_frame": 143.0, "compositor_cpu_ms_per_frame": 67.4, "machine_cpu_pct": 199.2, "machine_cpu_pct_of_all_cores": 99.6, "frames_bytes_allocated_p50": 3686400, "resource_notes": ""}
@@ -0,0 +1 @@
{"route": "pywayland-in-process", "language": "python", "protocol": "zwlr-screencopy-v1", "output": "hh-chromium-gqkxz3", "geometry": [1280, 720], "mode": "jpeg", "warmup": 10, "frames": 200, "frame_ms": {"min": 39.218, "p50": 76.971, "mean": 71.88, "p95": 95.795, "max": 149.48}, "stage_ms": {"capture": 69.53, "encode": 2.305, "stitch": 0.0}, "fps_sustained": 13.91, "wall_ms": 14378.78, "bytes_p50": 105813, "persistent": true, "notes": "one connection, buffers reused, turbojpeg q80 in-process; 1 output(s); forcing a repaint per frame", "cpu_ms_per_frame": 1.782, "rss_peak_mb": 29.3, "rss_peak_kb_raw": 30000, "system_cpu_ms_per_frame": 143.05, "compositor_cpu_ms_per_frame": 67.55, "machine_cpu_pct": 199.0, "machine_cpu_pct_of_all_cores": 99.5, "frames_bytes_allocated_p50": 3686400, "resource_notes": ""}
@@ -0,0 +1 @@
{"route": "pywayland-in-process", "language": "python", "protocol": "zwlr-screencopy-v1", "output": "hh-chromium-gqkxz3", "geometry": [1280, 720], "mode": "raw", "warmup": 10, "frames": 200, "frame_ms": {"min": 40.891, "p50": 76.175, "mean": 70.827, "p95": 96.846, "max": 125.471}, "stage_ms": {"capture": 70.816, "encode": 0.0, "stitch": 0.0}, "fps_sustained": 14.12, "wall_ms": 14165.47, "bytes_p50": null, "persistent": true, "notes": "one connection, buffers reused, turbojpeg q80 in-process; 1 output(s); forcing a repaint per frame", "cpu_ms_per_frame": 0.33, "rss_peak_mb": 24.45, "rss_peak_kb_raw": 25036, "system_cpu_ms_per_frame": 140.65, "compositor_cpu_ms_per_frame": 67.3, "machine_cpu_pct": 198.6, "machine_cpu_pct_of_all_cores": 99.3, "frames_bytes_allocated_p50": 3686400, "resource_notes": ""}
@@ -0,0 +1 @@
{"route": "pywayland-in-process", "language": "python", "protocol": "zwlr-screencopy-v1", "output": "hh-chromium-gqkxz3", "geometry": [1280, 720], "mode": "raw", "warmup": 10, "frames": 200, "frame_ms": {"min": 42.575, "p50": 77.219, "mean": 70.469, "p95": 93.935, "max": 122.928}, "stage_ms": {"capture": 70.459, "encode": 0.0, "stitch": 0.0}, "fps_sustained": 14.19, "wall_ms": 14094.03, "bytes_p50": null, "persistent": true, "notes": "one connection, buffers reused, turbojpeg q80 in-process; 1 output(s); forcing a repaint per frame", "cpu_ms_per_frame": 0.328, "rss_peak_mb": 24.45, "rss_peak_kb_raw": 25036, "system_cpu_ms_per_frame": 140.1, "compositor_cpu_ms_per_frame": 66.95, "machine_cpu_pct": 198.8, "machine_cpu_pct_of_all_cores": 99.4, "frames_bytes_allocated_p50": 3686400, "resource_notes": ""}
@@ -0,0 +1,674 @@
[
{
"route": "grim-subprocess",
"description": "grim subprocess per frame (shipped)",
"cell": "output-raw",
"run": 1,
"command": "/nix/store/i7d46siafmlrgcmj9nzvwj7y64p34yaz-python3-3.14.7-env/bin/python3 /var/lib/hermes/worktrees/hhs-capture-speed/spikes/capture-speed/producers/grim_producer.py --output hh-chromium-gqkxz3 --frames 200 --warmup 10 --mode raw --json",
"rc": 0,
"elapsed_s": 26.8,
"language": "python",
"protocol": "zwlr-screencopy-v1 (grim's client)",
"output": "hh-chromium-gqkxz3",
"geometry": [
1280,
720
],
"mode": "raw",
"warmup": 10,
"frames": 200,
"frame_ms": {
"min": 51.756,
"p50": 120.321,
"mean": 125.705,
"p95": 190.299,
"max": 253.325
},
"fps_sustained": 7.96,
"wall_ms": 25452.51,
"bytes_p50": 2764816,
"persistent": false,
"notes": "one grim process per frame, fork and encode inside the timed span; territory: --output; ppm, unencoded",
"cpu_ms_per_frame": 3.856,
"rss_peak_mb": 20.56,
"rss_peak_kb_raw": 21052,
"system_cpu_ms_per_frame": 253.8,
"compositor_cpu_ms_per_frame": 118.55,
"machine_cpu_pct": 199.4,
"machine_cpu_pct_of_all_cores": 99.7,
"frames_bytes_allocated_p50": 3686400,
"resource_notes": ""
},
{
"route": "grim-subprocess",
"description": "grim subprocess per frame (shipped)",
"cell": "output-raw",
"run": 2,
"command": "/nix/store/i7d46siafmlrgcmj9nzvwj7y64p34yaz-python3-3.14.7-env/bin/python3 /var/lib/hermes/worktrees/hhs-capture-speed/spikes/capture-speed/producers/grim_producer.py --output hh-chromium-gqkxz3 --frames 200 --warmup 10 --mode raw --json",
"rc": 0,
"elapsed_s": 27.0,
"language": "python",
"protocol": "zwlr-screencopy-v1 (grim's client)",
"output": "hh-chromium-gqkxz3",
"geometry": [
1280,
720
],
"mode": "raw",
"warmup": 10,
"frames": 200,
"frame_ms": {
"min": 64.061,
"p50": 122.389,
"mean": 127.307,
"p95": 196.38,
"max": 264.387
},
"fps_sustained": 7.86,
"wall_ms": 25757.6,
"bytes_p50": 2764816,
"persistent": false,
"notes": "one grim process per frame, fork and encode inside the timed span; territory: --output; ppm, unencoded",
"cpu_ms_per_frame": 3.858,
"rss_peak_mb": 20.51,
"rss_peak_kb_raw": 21004,
"system_cpu_ms_per_frame": 256.65,
"compositor_cpu_ms_per_frame": 118.25,
"machine_cpu_pct": 199.3,
"machine_cpu_pct_of_all_cores": 99.6,
"frames_bytes_allocated_p50": 3686400,
"resource_notes": ""
},
{
"route": "grim-subprocess",
"description": "grim subprocess per frame (shipped)",
"cell": "output-jpeg",
"run": 1,
"command": "/nix/store/i7d46siafmlrgcmj9nzvwj7y64p34yaz-python3-3.14.7-env/bin/python3 /var/lib/hermes/worktrees/hhs-capture-speed/spikes/capture-speed/producers/grim_producer.py --output hh-chromium-gqkxz3 --frames 200 --warmup 10 --mode jpeg --json",
"rc": 0,
"elapsed_s": 26.1,
"language": "python",
"protocol": "zwlr-screencopy-v1 (grim's client)",
"output": "hh-chromium-gqkxz3",
"geometry": [
1280,
720
],
"mode": "jpeg",
"warmup": 10,
"frames": 200,
"frame_ms": {
"min": 58.928,
"p50": 116.179,
"mean": 121.352,
"p95": 164.512,
"max": 200.886
},
"fps_sustained": 8.24,
"wall_ms": 24614.05,
"bytes_p50": 130597,
"persistent": false,
"notes": "one grim process per frame, fork and encode inside the timed span; territory: --output; jpeg q80",
"cpu_ms_per_frame": 5.005,
"rss_peak_mb": 18.47,
"rss_peak_kb_raw": 18916,
"system_cpu_ms_per_frame": 245.05,
"compositor_cpu_ms_per_frame": 115.05,
"machine_cpu_pct": 199.1,
"machine_cpu_pct_of_all_cores": 99.6,
"frames_bytes_allocated_p50": null,
"resource_notes": ""
},
{
"route": "grim-subprocess",
"description": "grim subprocess per frame (shipped)",
"cell": "output-jpeg",
"run": 2,
"command": "/nix/store/i7d46siafmlrgcmj9nzvwj7y64p34yaz-python3-3.14.7-env/bin/python3 /var/lib/hermes/worktrees/hhs-capture-speed/spikes/capture-speed/producers/grim_producer.py --output hh-chromium-gqkxz3 --frames 200 --warmup 10 --mode jpeg --json",
"rc": 0,
"elapsed_s": 26.6,
"language": "python",
"protocol": "zwlr-screencopy-v1 (grim's client)",
"output": "hh-chromium-gqkxz3",
"geometry": [
1280,
720
],
"mode": "jpeg",
"warmup": 10,
"frames": 200,
"frame_ms": {
"min": 67.203,
"p50": 118.52,
"mean": 124.224,
"p95": 185.591,
"max": 299.044
},
"fps_sustained": 8.05,
"wall_ms": 25182.14,
"bytes_p50": 130590,
"persistent": false,
"notes": "one grim process per frame, fork and encode inside the timed span; territory: --output; jpeg q80",
"cpu_ms_per_frame": 5.05,
"rss_peak_mb": 18.45,
"rss_peak_kb_raw": 18896,
"system_cpu_ms_per_frame": 250.75,
"compositor_cpu_ms_per_frame": 117.75,
"machine_cpu_pct": 199.1,
"machine_cpu_pct_of_all_cores": 99.6,
"frames_bytes_allocated_p50": null,
"resource_notes": ""
},
{
"route": "pywayland-in-process",
"description": "pywayland in-process, zwlr-screencopy-v1",
"cell": "output-raw",
"run": 1,
"command": "/nix/store/i7d46siafmlrgcmj9nzvwj7y64p34yaz-python3-3.14.7-env/bin/python3 /var/lib/hermes/worktrees/hhs-capture-speed/spikes/capture-speed/producers/wayland_producer.py --output hh-chromium-gqkxz3 --protocol screencopy --frames 200 --warmup 10 --mode raw --json",
"rc": 0,
"elapsed_s": 15.2,
"language": "python",
"protocol": "zwlr-screencopy-v1",
"output": "hh-chromium-gqkxz3",
"geometry": [
1280,
720
],
"mode": "raw",
"warmup": 10,
"frames": 200,
"frame_ms": {
"min": 40.891,
"p50": 76.175,
"mean": 70.827,
"p95": 96.846,
"max": 125.471
},
"stage_ms": {
"capture": 70.816,
"encode": 0.0,
"stitch": 0.0
},
"fps_sustained": 14.12,
"wall_ms": 14165.47,
"bytes_p50": null,
"persistent": true,
"notes": "one connection, buffers reused, turbojpeg q80 in-process; 1 output(s); forcing a repaint per frame",
"cpu_ms_per_frame": 0.33,
"rss_peak_mb": 24.45,
"rss_peak_kb_raw": 25036,
"system_cpu_ms_per_frame": 140.65,
"compositor_cpu_ms_per_frame": 67.3,
"machine_cpu_pct": 198.6,
"machine_cpu_pct_of_all_cores": 99.3,
"frames_bytes_allocated_p50": 3686400,
"resource_notes": ""
},
{
"route": "pywayland-in-process",
"description": "pywayland in-process, zwlr-screencopy-v1",
"cell": "output-raw",
"run": 2,
"command": "/nix/store/i7d46siafmlrgcmj9nzvwj7y64p34yaz-python3-3.14.7-env/bin/python3 /var/lib/hermes/worktrees/hhs-capture-speed/spikes/capture-speed/producers/wayland_producer.py --output hh-chromium-gqkxz3 --protocol screencopy --frames 200 --warmup 10 --mode raw --json",
"rc": 0,
"elapsed_s": 15.1,
"language": "python",
"protocol": "zwlr-screencopy-v1",
"output": "hh-chromium-gqkxz3",
"geometry": [
1280,
720
],
"mode": "raw",
"warmup": 10,
"frames": 200,
"frame_ms": {
"min": 42.575,
"p50": 77.219,
"mean": 70.469,
"p95": 93.935,
"max": 122.928
},
"stage_ms": {
"capture": 70.459,
"encode": 0.0,
"stitch": 0.0
},
"fps_sustained": 14.19,
"wall_ms": 14094.03,
"bytes_p50": null,
"persistent": true,
"notes": "one connection, buffers reused, turbojpeg q80 in-process; 1 output(s); forcing a repaint per frame",
"cpu_ms_per_frame": 0.328,
"rss_peak_mb": 24.45,
"rss_peak_kb_raw": 25036,
"system_cpu_ms_per_frame": 140.1,
"compositor_cpu_ms_per_frame": 66.95,
"machine_cpu_pct": 198.8,
"machine_cpu_pct_of_all_cores": 99.4,
"frames_bytes_allocated_p50": 3686400,
"resource_notes": ""
},
{
"route": "pywayland-in-process",
"description": "pywayland in-process, zwlr-screencopy-v1",
"cell": "output-jpeg",
"run": 1,
"command": "/nix/store/i7d46siafmlrgcmj9nzvwj7y64p34yaz-python3-3.14.7-env/bin/python3 /var/lib/hermes/worktrees/hhs-capture-speed/spikes/capture-speed/producers/wayland_producer.py --output hh-chromium-gqkxz3 --protocol screencopy --frames 200 --warmup 10 --mode jpeg --json",
"rc": 0,
"elapsed_s": 15.4,
"language": "python",
"protocol": "zwlr-screencopy-v1",
"output": "hh-chromium-gqkxz3",
"geometry": [
1280,
720
],
"mode": "jpeg",
"warmup": 10,
"frames": 200,
"frame_ms": {
"min": 25.597,
"p50": 79.088,
"mean": 71.771,
"p95": 96.766,
"max": 132.204
},
"stage_ms": {
"capture": 69.093,
"encode": 2.522,
"stitch": 0.0
},
"fps_sustained": 13.93,
"wall_ms": 14354.55,
"bytes_p50": 105799,
"persistent": true,
"notes": "one connection, buffers reused, turbojpeg q80 in-process; 1 output(s); forcing a repaint per frame",
"cpu_ms_per_frame": 1.781,
"rss_peak_mb": 29.38,
"rss_peak_kb_raw": 30084,
"system_cpu_ms_per_frame": 143.0,
"compositor_cpu_ms_per_frame": 67.4,
"machine_cpu_pct": 199.2,
"machine_cpu_pct_of_all_cores": 99.6,
"frames_bytes_allocated_p50": 3686400,
"resource_notes": ""
},
{
"route": "pywayland-in-process",
"description": "pywayland in-process, zwlr-screencopy-v1",
"cell": "output-jpeg",
"run": 2,
"command": "/nix/store/i7d46siafmlrgcmj9nzvwj7y64p34yaz-python3-3.14.7-env/bin/python3 /var/lib/hermes/worktrees/hhs-capture-speed/spikes/capture-speed/producers/wayland_producer.py --output hh-chromium-gqkxz3 --protocol screencopy --frames 200 --warmup 10 --mode jpeg --json",
"rc": 0,
"elapsed_s": 15.4,
"language": "python",
"protocol": "zwlr-screencopy-v1",
"output": "hh-chromium-gqkxz3",
"geometry": [
1280,
720
],
"mode": "jpeg",
"warmup": 10,
"frames": 200,
"frame_ms": {
"min": 39.218,
"p50": 76.971,
"mean": 71.88,
"p95": 95.795,
"max": 149.48
},
"stage_ms": {
"capture": 69.53,
"encode": 2.305,
"stitch": 0.0
},
"fps_sustained": 13.91,
"wall_ms": 14378.78,
"bytes_p50": 105813,
"persistent": true,
"notes": "one connection, buffers reused, turbojpeg q80 in-process; 1 output(s); forcing a repaint per frame",
"cpu_ms_per_frame": 1.782,
"rss_peak_mb": 29.3,
"rss_peak_kb_raw": 30000,
"system_cpu_ms_per_frame": 143.05,
"compositor_cpu_ms_per_frame": 67.55,
"machine_cpu_pct": 199.0,
"machine_cpu_pct_of_all_cores": 99.5,
"frames_bytes_allocated_p50": 3686400,
"resource_notes": ""
},
{
"route": "pywayland-in-process",
"description": "pywayland in-process, ext-image-copy-capture-v1",
"cell": "output-raw",
"run": 1,
"command": "/nix/store/i7d46siafmlrgcmj9nzvwj7y64p34yaz-python3-3.14.7-env/bin/python3 /var/lib/hermes/worktrees/hhs-capture-speed/spikes/capture-speed/producers/wayland_producer.py --output hh-chromium-gqkxz3 --protocol ext --frames 200 --warmup 10 --mode raw --json",
"rc": 0,
"elapsed_s": 15.0,
"language": "python",
"protocol": "ext-image-copy-capture-v1",
"output": "hh-chromium-gqkxz3",
"geometry": [
1280,
720
],
"mode": "raw",
"warmup": 10,
"frames": 200,
"frame_ms": {
"min": 22.685,
"p50": 71.372,
"mean": 69.708,
"p95": 89.816,
"max": 121.781
},
"stage_ms": {
"capture": 69.698,
"encode": 0.0,
"stitch": 0.0
},
"fps_sustained": 14.35,
"wall_ms": 13941.78,
"bytes_p50": null,
"persistent": true,
"notes": "one connection, buffers reused, turbojpeg q80 in-process; 1 output(s); forcing a repaint per frame",
"cpu_ms_per_frame": 0.311,
"rss_peak_mb": 24.43,
"rss_peak_kb_raw": 25020,
"system_cpu_ms_per_frame": 138.6,
"compositor_cpu_ms_per_frame": 65.85,
"machine_cpu_pct": 198.8,
"machine_cpu_pct_of_all_cores": 99.4,
"frames_bytes_allocated_p50": 3686400,
"resource_notes": ""
},
{
"route": "pywayland-in-process",
"description": "pywayland in-process, ext-image-copy-capture-v1",
"cell": "output-raw",
"run": 2,
"command": "/nix/store/i7d46siafmlrgcmj9nzvwj7y64p34yaz-python3-3.14.7-env/bin/python3 /var/lib/hermes/worktrees/hhs-capture-speed/spikes/capture-speed/producers/wayland_producer.py --output hh-chromium-gqkxz3 --protocol ext --frames 200 --warmup 10 --mode raw --json",
"rc": 0,
"elapsed_s": 14.9,
"language": "python",
"protocol": "ext-image-copy-capture-v1",
"output": "hh-chromium-gqkxz3",
"geometry": [
1280,
720
],
"mode": "raw",
"warmup": 10,
"frames": 200,
"frame_ms": {
"min": 41.241,
"p50": 75.703,
"mean": 69.499,
"p95": 86.611,
"max": 112.812
},
"stage_ms": {
"capture": 69.489,
"encode": 0.0,
"stitch": 0.0
},
"fps_sustained": 14.39,
"wall_ms": 13899.94,
"bytes_p50": null,
"persistent": true,
"notes": "one connection, buffers reused, turbojpeg q80 in-process; 1 output(s); forcing a repaint per frame",
"cpu_ms_per_frame": 0.318,
"rss_peak_mb": 24.52,
"rss_peak_kb_raw": 25112,
"system_cpu_ms_per_frame": 138.35,
"compositor_cpu_ms_per_frame": 66.45,
"machine_cpu_pct": 199.1,
"machine_cpu_pct_of_all_cores": 99.5,
"frames_bytes_allocated_p50": 3686400,
"resource_notes": ""
},
{
"route": "pywayland-in-process",
"description": "pywayland in-process, ext-image-copy-capture-v1",
"cell": "output-jpeg",
"run": 1,
"command": "/nix/store/i7d46siafmlrgcmj9nzvwj7y64p34yaz-python3-3.14.7-env/bin/python3 /var/lib/hermes/worktrees/hhs-capture-speed/spikes/capture-speed/producers/wayland_producer.py --output hh-chromium-gqkxz3 --protocol ext --frames 200 --warmup 10 --mode jpeg --json",
"rc": 0,
"elapsed_s": 15.3,
"language": "python",
"protocol": "ext-image-copy-capture-v1",
"output": "hh-chromium-gqkxz3",
"geometry": [
1280,
720
],
"mode": "jpeg",
"warmup": 10,
"frames": 200,
"frame_ms": {
"min": 30.44,
"p50": 76.633,
"mean": 71.153,
"p95": 103.055,
"max": 185.942
},
"stage_ms": {
"capture": 68.35,
"encode": 2.708,
"stitch": 0.0
},
"fps_sustained": 14.05,
"wall_ms": 14231.13,
"bytes_p50": 105794,
"persistent": true,
"notes": "one connection, buffers reused, turbojpeg q80 in-process; 1 output(s); forcing a repaint per frame",
"cpu_ms_per_frame": 1.811,
"rss_peak_mb": 28.99,
"rss_peak_kb_raw": 29688,
"system_cpu_ms_per_frame": 141.75,
"compositor_cpu_ms_per_frame": 66.55,
"machine_cpu_pct": 199.2,
"machine_cpu_pct_of_all_cores": 99.6,
"frames_bytes_allocated_p50": 3686400,
"resource_notes": ""
},
{
"route": "pywayland-in-process",
"description": "pywayland in-process, ext-image-copy-capture-v1",
"cell": "output-jpeg",
"run": 2,
"command": "/nix/store/i7d46siafmlrgcmj9nzvwj7y64p34yaz-python3-3.14.7-env/bin/python3 /var/lib/hermes/worktrees/hhs-capture-speed/spikes/capture-speed/producers/wayland_producer.py --output hh-chromium-gqkxz3 --protocol ext --frames 200 --warmup 10 --mode jpeg --json",
"rc": 0,
"elapsed_s": 15.3,
"language": "python",
"protocol": "ext-image-copy-capture-v1",
"output": "hh-chromium-gqkxz3",
"geometry": [
1280,
720
],
"mode": "jpeg",
"warmup": 10,
"frames": 200,
"frame_ms": {
"min": 38.593,
"p50": 78.693,
"mean": 71.507,
"p95": 96.637,
"max": 129.106
},
"stage_ms": {
"capture": 69.115,
"encode": 2.339,
"stitch": 0.0
},
"fps_sustained": 13.98,
"wall_ms": 14301.79,
"bytes_p50": 105804,
"persistent": true,
"notes": "one connection, buffers reused, turbojpeg q80 in-process; 1 output(s); forcing a repaint per frame",
"cpu_ms_per_frame": 1.776,
"rss_peak_mb": 29.25,
"rss_peak_kb_raw": 29956,
"system_cpu_ms_per_frame": 142.65,
"compositor_cpu_ms_per_frame": 67.9,
"machine_cpu_pct": 199.5,
"machine_cpu_pct_of_all_cores": 99.7,
"frames_bytes_allocated_p50": 3686400,
"resource_notes": ""
},
{
"route": "grim-rs",
"description": "grim-rs crate (Rust), long-lived process",
"cell": "output-raw",
"run": 1,
"command": "/var/lib/hermes/worktrees/hhs-capture-speed/spikes/capture-speed/001a-grim-rs/target/release/grimrs-capture --output hh-chromium-gqkxz3 --frames 200 --warmup 10 --mode raw --json",
"rc": 0,
"elapsed_s": 27.7,
"language": "rust",
"protocol": "ext-image-copy-capture-v1 (auto)",
"output": "hh-chromium-gqkxz3",
"geometry": [
1280,
720
],
"mode": "raw",
"warmup": 10,
"frames": 200,
"frame_ms": {
"min": 44.16,
"p50": 122.21,
"mean": 130.086,
"p95": 189.49,
"max": 226.412
},
"fps_sustained": 7.687,
"wall_ms": 26017.423,
"cpu_ms_per_frame": 1.628,
"bytes_p50": null,
"persistent": true,
"rss_peak_mb": 746.09,
"rss_peak_kb_raw": 763992,
"system_cpu_ms_per_frame": 258.6,
"compositor_cpu_ms_per_frame": 120.85,
"machine_cpu_pct": 198.8,
"frames_bytes_allocated_p50": 3686400,
"notes": "grim-rs 0.2.0 via Grim::new() (auto-detect); copy target is a wl_shm buffer created inside each capture_* call, not a dma-buf the producer imports. Cursor not overlaid (the frozen argv has no cursor flag; hhstream-live's grim command passes -c). JPEG is Grim::to_jpeg, quality 80, the crate default and the live-stream setting. Raw timer stops when capture_* returns, which is after the crate copies the mapping into a Vec and swizzles XRGB/ARGB to RGBA (and applies any output transform or y-invert). The public API does not expose the mapped buffer before that, so raw includes work the contract says to leave outside the timer. Percentiles are linear interpolation on the sorted samples. cpu_ms_per_frame is getrusage(RUSAGE_SELF) utime+stime over the measured loop only, divided by frames. Frame buffers are freed between frames; that free is inside wall_ms and outside frame_ms. Outputs: Virtual-1@0,0 1280x720 scale 1; hh-chromium-gqkxz3@1280,0 1280x720 scale 1. No GPU on this machine; dma-buf was not used and is not claimed. This process does not move the pointer and does not start or stop a session; the pixels are whatever the compositor is already showing. Last warmup frame nonzero bytes: 3679565."
},
{
"route": "grim-rs",
"description": "grim-rs crate (Rust), long-lived process",
"cell": "output-raw",
"run": 2,
"command": "/var/lib/hermes/worktrees/hhs-capture-speed/spikes/capture-speed/001a-grim-rs/target/release/grimrs-capture --output hh-chromium-gqkxz3 --frames 200 --warmup 10 --mode raw --json",
"rc": 0,
"elapsed_s": 26.7,
"language": "rust",
"protocol": "ext-image-copy-capture-v1 (auto)",
"output": "hh-chromium-gqkxz3",
"geometry": [
1280,
720
],
"mode": "raw",
"warmup": 10,
"frames": 200,
"frame_ms": {
"min": 46.164,
"p50": 117.913,
"mean": 125.337,
"p95": 194.097,
"max": 255.125
},
"fps_sustained": 7.978,
"wall_ms": 25067.547,
"cpu_ms_per_frame": 1.551,
"bytes_p50": null,
"persistent": true,
"rss_peak_mb": 746.02,
"rss_peak_kb_raw": 763924,
"system_cpu_ms_per_frame": 249.6,
"compositor_cpu_ms_per_frame": 117.6,
"machine_cpu_pct": 199.1,
"frames_bytes_allocated_p50": 3686400,
"notes": "grim-rs 0.2.0 via Grim::new() (auto-detect); copy target is a wl_shm buffer created inside each capture_* call, not a dma-buf the producer imports. Cursor not overlaid (the frozen argv has no cursor flag; hhstream-live's grim command passes -c). JPEG is Grim::to_jpeg, quality 80, the crate default and the live-stream setting. Raw timer stops when capture_* returns, which is after the crate copies the mapping into a Vec and swizzles XRGB/ARGB to RGBA (and applies any output transform or y-invert). The public API does not expose the mapped buffer before that, so raw includes work the contract says to leave outside the timer. Percentiles are linear interpolation on the sorted samples. cpu_ms_per_frame is getrusage(RUSAGE_SELF) utime+stime over the measured loop only, divided by frames. Frame buffers are freed between frames; that free is inside wall_ms and outside frame_ms. Outputs: Virtual-1@0,0 1280x720 scale 1; hh-chromium-gqkxz3@1280,0 1280x720 scale 1. No GPU on this machine; dma-buf was not used and is not claimed. This process does not move the pointer and does not start or stop a session; the pixels are whatever the compositor is already showing. Last warmup frame nonzero bytes: 3679565."
},
{
"route": "grim-rs",
"description": "grim-rs crate (Rust), long-lived process",
"cell": "output-jpeg",
"run": 1,
"command": "/var/lib/hermes/worktrees/hhs-capture-speed/spikes/capture-speed/001a-grim-rs/target/release/grimrs-capture --output hh-chromium-gqkxz3 --frames 200 --warmup 10 --mode jpeg --json",
"rc": 0,
"elapsed_s": 28.2,
"language": "rust",
"protocol": "ext-image-copy-capture-v1 (auto)",
"output": "hh-chromium-gqkxz3",
"geometry": [
1280,
720
],
"mode": "jpeg",
"warmup": 10,
"frames": 200,
"frame_ms": {
"min": 61.074,
"p50": 124.039,
"mean": 132.54,
"p95": 198.748,
"max": 254.625
},
"fps_sustained": 7.545,
"wall_ms": 26508.219,
"cpu_ms_per_frame": 6.6,
"bytes_p50": 107205.5,
"persistent": true,
"rss_peak_mb": 746.25,
"rss_peak_kb_raw": 764156,
"system_cpu_ms_per_frame": 264.3,
"compositor_cpu_ms_per_frame": 123.1,
"machine_cpu_pct": 199.4,
"frames_bytes_allocated_p50": 3686400,
"notes": "grim-rs 0.2.0 via Grim::new() (auto-detect); copy target is a wl_shm buffer created inside each capture_* call, not a dma-buf the producer imports. Cursor not overlaid (the frozen argv has no cursor flag; hhstream-live's grim command passes -c). JPEG is Grim::to_jpeg, quality 80, the crate default and the live-stream setting. Raw timer stops when capture_* returns, which is after the crate copies the mapping into a Vec and swizzles XRGB/ARGB to RGBA (and applies any output transform or y-invert). The public API does not expose the mapped buffer before that, so raw includes work the contract says to leave outside the timer. Percentiles are linear interpolation on the sorted samples. cpu_ms_per_frame is getrusage(RUSAGE_SELF) utime+stime over the measured loop only, divided by frames. Frame buffers are freed between frames; that free is inside wall_ms and outside frame_ms. Outputs: Virtual-1@0,0 1280x720 scale 1; hh-chromium-gqkxz3@1280,0 1280x720 scale 1. No GPU on this machine; dma-buf was not used and is not claimed. This process does not move the pointer and does not start or stop a session; the pixels are whatever the compositor is already showing. Last warmup frame nonzero bytes: 105382."
},
{
"route": "grim-rs",
"description": "grim-rs crate (Rust), long-lived process",
"cell": "output-jpeg",
"run": 2,
"command": "/var/lib/hermes/worktrees/hhs-capture-speed/spikes/capture-speed/001a-grim-rs/target/release/grimrs-capture --output hh-chromium-gqkxz3 --frames 200 --warmup 10 --mode jpeg --json",
"rc": 0,
"elapsed_s": 28.6,
"language": "rust",
"protocol": "ext-image-copy-capture-v1 (auto)",
"output": "hh-chromium-gqkxz3",
"geometry": [
1280,
720
],
"mode": "jpeg",
"warmup": 10,
"frames": 200,
"frame_ms": {
"min": 59.907,
"p50": 125.214,
"mean": 135.259,
"p95": 207.095,
"max": 324.014
},
"fps_sustained": 7.393,
"wall_ms": 27051.989,
"cpu_ms_per_frame": 6.595,
"bytes_p50": 107203.5,
"persistent": true,
"rss_peak_mb": 746.34,
"rss_peak_kb_raw": 764256,
"system_cpu_ms_per_frame": 269.1,
"compositor_cpu_ms_per_frame": 123.05,
"machine_cpu_pct": 199.0,
"frames_bytes_allocated_p50": 3686400,
"notes": "grim-rs 0.2.0 via Grim::new() (auto-detect); copy target is a wl_shm buffer created inside each capture_* call, not a dma-buf the producer imports. Cursor not overlaid (the frozen argv has no cursor flag; hhstream-live's grim command passes -c). JPEG is Grim::to_jpeg, quality 80, the crate default and the live-stream setting. Raw timer stops when capture_* returns, which is after the crate copies the mapping into a Vec and swizzles XRGB/ARGB to RGBA (and applies any output transform or y-invert). The public API does not expose the mapped buffer before that, so raw includes work the contract says to leave outside the timer. Percentiles are linear interpolation on the sorted samples. cpu_ms_per_frame is getrusage(RUSAGE_SELF) utime+stime over the measured loop only, divided by frames. Frame buffers are freed between frames; that free is inside wall_ms and outside frame_ms. Outputs: Virtual-1@0,0 1280x720 scale 1; hh-chromium-gqkxz3@1280,0 1280x720 scale 1. No GPU on this machine; dma-buf was not used and is not claimed. This process does not move the pointer and does not start or stop a session; the pixels are whatever the compositor is already showing. Last warmup frame nonzero bytes: 105080."
}
]
@@ -0,0 +1,7 @@
{
"window_ms": 10000.0,
"system_cpu_ms": 19850.0,
"compositor_cpu_ms": 9500.0,
"system_cpu_pct": 198.5,
"compositor_cpu_pct": 95.0
}
@@ -0,0 +1,12 @@
grimrs-capture: route grim-rs protocol ext-image-copy-capture-v1 (auto) mode jpeg target hh-chromium-52w2dm warmup 10 frames 200
grimrs-capture: outputs Virtual-1@0,0 1280x720 scale 1; hh-chromium-gqkxz3@1280,0 1280x720 scale 1; hh-chromium-52w2dm@2560,0 1280x720 scale 1
grimrs-capture: warmup 1/10 1280x720 bytes 107829 nonzero 105852
grimrs-capture: warmup 2/10 1280x720 bytes 107829 nonzero 105852
grimrs-capture: warmup 3/10 1280x720 bytes 107829 nonzero 105852
grimrs-capture: warmup 4/10 1280x720 bytes 107829 nonzero 105852
grimrs-capture: warmup 5/10 1280x720 bytes 107829 nonzero 105852
grimrs-capture: warmup 6/10 1280x720 bytes 107829 nonzero 105852
grimrs-capture: warmup 7/10 1280x720 bytes 107829 nonzero 105852
grimrs-capture: warmup 8/10 1280x720 bytes 107829 nonzero 105852
grimrs-capture: warmup 9/10 1280x720 bytes 107829 nonzero 105852
grimrs-capture: warmup 10/10 1280x720 bytes 107829 nonzero 105852
@@ -0,0 +1 @@
{"route":"grim-rs","language":"rust","protocol":"ext-image-copy-capture-v1 (auto)","output":"hh-chromium-52w2dm","geometry":[1280,720],"mode":"jpeg","warmup":10,"frames":200,"frame_ms":{"min":95.334,"p50":186.693,"mean":199.129,"p95":282.41,"max":411.263},"fps_sustained":5.022,"wall_ms":39825.916,"cpu_ms_per_frame":6.7,"bytes_p50":107829.0,"persistent":true,"rss_peak_mb":746.3,"rss_peak_kb_raw":764208,"system_cpu_ms_per_frame":396.0,"compositor_cpu_ms_per_frame":187.75,"machine_cpu_pct":198.9,"frames_bytes_allocated_p50":3686400,"notes":"grim-rs 0.2.0 via Grim::new() (auto-detect); copy target is a wl_shm buffer created inside each capture_* call, not a dma-buf the producer imports. Cursor not overlaid (the frozen argv has no cursor flag; hhstream-live's grim command passes -c). JPEG is Grim::to_jpeg, quality 80, the crate default and the live-stream setting. Raw timer stops when capture_* returns, which is after the crate copies the mapping into a Vec and swizzles XRGB/ARGB to RGBA (and applies any output transform or y-invert). The public API does not expose the mapped buffer before that, so raw includes work the contract says to leave outside the timer. Percentiles are linear interpolation on the sorted samples. cpu_ms_per_frame is getrusage(RUSAGE_SELF) utime+stime over the measured loop only, divided by frames. Frame buffers are freed between frames; that free is inside wall_ms and outside frame_ms. Outputs: Virtual-1@0,0 1280x720 scale 1; hh-chromium-gqkxz3@1280,0 1280x720 scale 1; hh-chromium-52w2dm@2560,0 1280x720 scale 1. No GPU on this machine; dma-buf was not used and is not claimed. This process does not move the pointer and does not start or stop a session; the pixels are whatever the compositor is already showing. Last warmup frame nonzero bytes: 105852."}
@@ -0,0 +1,12 @@
grimrs-capture: route grim-rs protocol ext-image-copy-capture-v1 (auto) mode jpeg target hh-chromium-52w2dm warmup 10 frames 200
grimrs-capture: outputs Virtual-1@0,0 1280x720 scale 1; hh-chromium-gqkxz3@1280,0 1280x720 scale 1; hh-chromium-52w2dm@2560,0 1280x720 scale 1
grimrs-capture: warmup 1/10 1280x720 bytes 107829 nonzero 105852
grimrs-capture: warmup 2/10 1280x720 bytes 107829 nonzero 105852
grimrs-capture: warmup 3/10 1280x720 bytes 107829 nonzero 105852
grimrs-capture: warmup 4/10 1280x720 bytes 107829 nonzero 105852
grimrs-capture: warmup 5/10 1280x720 bytes 107829 nonzero 105852
grimrs-capture: warmup 6/10 1280x720 bytes 107829 nonzero 105852
grimrs-capture: warmup 7/10 1280x720 bytes 107829 nonzero 105852
grimrs-capture: warmup 8/10 1280x720 bytes 107829 nonzero 105852
grimrs-capture: warmup 9/10 1280x720 bytes 107829 nonzero 105852
grimrs-capture: warmup 10/10 1280x720 bytes 107829 nonzero 105852
@@ -0,0 +1 @@
{"route":"grim-rs","language":"rust","protocol":"ext-image-copy-capture-v1 (auto)","output":"hh-chromium-52w2dm","geometry":[1280,720],"mode":"jpeg","warmup":10,"frames":200,"frame_ms":{"min":104.636,"p50":184.629,"mean":198.848,"p95":306.666,"max":368.522},"fps_sustained":5.029,"wall_ms":39771.382,"cpu_ms_per_frame":6.67,"bytes_p50":107829.0,"persistent":true,"rss_peak_mb":746.35,"rss_peak_kb_raw":764264,"system_cpu_ms_per_frame":395.55,"compositor_cpu_ms_per_frame":187.5,"machine_cpu_pct":198.9,"frames_bytes_allocated_p50":3686400,"notes":"grim-rs 0.2.0 via Grim::new() (auto-detect); copy target is a wl_shm buffer created inside each capture_* call, not a dma-buf the producer imports. Cursor not overlaid (the frozen argv has no cursor flag; hhstream-live's grim command passes -c). JPEG is Grim::to_jpeg, quality 80, the crate default and the live-stream setting. Raw timer stops when capture_* returns, which is after the crate copies the mapping into a Vec and swizzles XRGB/ARGB to RGBA (and applies any output transform or y-invert). The public API does not expose the mapped buffer before that, so raw includes work the contract says to leave outside the timer. Percentiles are linear interpolation on the sorted samples. cpu_ms_per_frame is getrusage(RUSAGE_SELF) utime+stime over the measured loop only, divided by frames. Frame buffers are freed between frames; that free is inside wall_ms and outside frame_ms. Outputs: Virtual-1@0,0 1280x720 scale 1; hh-chromium-gqkxz3@1280,0 1280x720 scale 1; hh-chromium-52w2dm@2560,0 1280x720 scale 1. No GPU on this machine; dma-buf was not used and is not claimed. This process does not move the pointer and does not start or stop a session; the pixels are whatever the compositor is already showing. Last warmup frame nonzero bytes: 105852."}
@@ -0,0 +1,12 @@
grimrs-capture: route grim-rs protocol ext-image-copy-capture-v1 (auto) mode raw target hh-chromium-52w2dm warmup 10 frames 200
grimrs-capture: outputs Virtual-1@0,0 1280x720 scale 1; hh-chromium-gqkxz3@1280,0 1280x720 scale 1; hh-chromium-52w2dm@2560,0 1280x720 scale 1
grimrs-capture: warmup 1/10 1280x720 bytes 3686400 nonzero 3679565
grimrs-capture: warmup 2/10 1280x720 bytes 3686400 nonzero 3679565
grimrs-capture: warmup 3/10 1280x720 bytes 3686400 nonzero 3679565
grimrs-capture: warmup 4/10 1280x720 bytes 3686400 nonzero 3679565
grimrs-capture: warmup 5/10 1280x720 bytes 3686400 nonzero 3679565
grimrs-capture: warmup 6/10 1280x720 bytes 3686400 nonzero 3679565
grimrs-capture: warmup 7/10 1280x720 bytes 3686400 nonzero 3679565
grimrs-capture: warmup 8/10 1280x720 bytes 3686400 nonzero 3679565
grimrs-capture: warmup 9/10 1280x720 bytes 3686400 nonzero 3679565
grimrs-capture: warmup 10/10 1280x720 bytes 3686400 nonzero 3679565
@@ -0,0 +1 @@
{"route":"grim-rs","language":"rust","protocol":"ext-image-copy-capture-v1 (auto)","output":"hh-chromium-52w2dm","geometry":[1280,720],"mode":"raw","warmup":10,"frames":200,"frame_ms":{"min":112.37,"p50":188.581,"mean":199.48,"p95":289.422,"max":358.439},"fps_sustained":5.013,"wall_ms":39896.261,"cpu_ms_per_frame":1.677,"bytes_p50":null,"persistent":true,"rss_peak_mb":746.18,"rss_peak_kb_raw":764084,"system_cpu_ms_per_frame":396.45,"compositor_cpu_ms_per_frame":188.85,"machine_cpu_pct":198.7,"frames_bytes_allocated_p50":3686400,"notes":"grim-rs 0.2.0 via Grim::new() (auto-detect); copy target is a wl_shm buffer created inside each capture_* call, not a dma-buf the producer imports. Cursor not overlaid (the frozen argv has no cursor flag; hhstream-live's grim command passes -c). JPEG is Grim::to_jpeg, quality 80, the crate default and the live-stream setting. Raw timer stops when capture_* returns, which is after the crate copies the mapping into a Vec and swizzles XRGB/ARGB to RGBA (and applies any output transform or y-invert). The public API does not expose the mapped buffer before that, so raw includes work the contract says to leave outside the timer. Percentiles are linear interpolation on the sorted samples. cpu_ms_per_frame is getrusage(RUSAGE_SELF) utime+stime over the measured loop only, divided by frames. Frame buffers are freed between frames; that free is inside wall_ms and outside frame_ms. Outputs: Virtual-1@0,0 1280x720 scale 1; hh-chromium-gqkxz3@1280,0 1280x720 scale 1; hh-chromium-52w2dm@2560,0 1280x720 scale 1. No GPU on this machine; dma-buf was not used and is not claimed. This process does not move the pointer and does not start or stop a session; the pixels are whatever the compositor is already showing. Last warmup frame nonzero bytes: 3679565."}
@@ -0,0 +1,12 @@
grimrs-capture: route grim-rs protocol ext-image-copy-capture-v1 (auto) mode raw target hh-chromium-52w2dm warmup 10 frames 200
grimrs-capture: outputs Virtual-1@0,0 1280x720 scale 1; hh-chromium-gqkxz3@1280,0 1280x720 scale 1; hh-chromium-52w2dm@2560,0 1280x720 scale 1
grimrs-capture: warmup 1/10 1280x720 bytes 3686400 nonzero 3679565
grimrs-capture: warmup 2/10 1280x720 bytes 3686400 nonzero 3679565
grimrs-capture: warmup 3/10 1280x720 bytes 3686400 nonzero 3679565
grimrs-capture: warmup 4/10 1280x720 bytes 3686400 nonzero 3679565
grimrs-capture: warmup 5/10 1280x720 bytes 3686400 nonzero 3679565
grimrs-capture: warmup 6/10 1280x720 bytes 3686400 nonzero 3679565
grimrs-capture: warmup 7/10 1280x720 bytes 3686400 nonzero 3679565
grimrs-capture: warmup 8/10 1280x720 bytes 3686400 nonzero 3679565
grimrs-capture: warmup 9/10 1280x720 bytes 3686400 nonzero 3679565
grimrs-capture: warmup 10/10 1280x720 bytes 3686400 nonzero 3679565
@@ -0,0 +1 @@
{"route":"grim-rs","language":"rust","protocol":"ext-image-copy-capture-v1 (auto)","output":"hh-chromium-52w2dm","geometry":[1280,720],"mode":"raw","warmup":10,"frames":200,"frame_ms":{"min":118.647,"p50":187.814,"mean":205.548,"p95":298.366,"max":361.436},"fps_sustained":4.865,"wall_ms":41109.705,"cpu_ms_per_frame":1.652,"bytes_p50":null,"persistent":true,"rss_peak_mb":746.0,"rss_peak_kb_raw":763904,"system_cpu_ms_per_frame":408.6,"compositor_cpu_ms_per_frame":194.3,"machine_cpu_pct":198.8,"frames_bytes_allocated_p50":3686400,"notes":"grim-rs 0.2.0 via Grim::new() (auto-detect); copy target is a wl_shm buffer created inside each capture_* call, not a dma-buf the producer imports. Cursor not overlaid (the frozen argv has no cursor flag; hhstream-live's grim command passes -c). JPEG is Grim::to_jpeg, quality 80, the crate default and the live-stream setting. Raw timer stops when capture_* returns, which is after the crate copies the mapping into a Vec and swizzles XRGB/ARGB to RGBA (and applies any output transform or y-invert). The public API does not expose the mapped buffer before that, so raw includes work the contract says to leave outside the timer. Percentiles are linear interpolation on the sorted samples. cpu_ms_per_frame is getrusage(RUSAGE_SELF) utime+stime over the measured loop only, divided by frames. Frame buffers are freed between frames; that free is inside wall_ms and outside frame_ms. Outputs: Virtual-1@0,0 1280x720 scale 1; hh-chromium-gqkxz3@1280,0 1280x720 scale 1; hh-chromium-52w2dm@2560,0 1280x720 scale 1. No GPU on this machine; dma-buf was not used and is not claimed. This process does not move the pointer and does not start or stop a session; the pixels are whatever the compositor is already showing. Last warmup frame nonzero bytes: 3679565."}
Whitespace-only changes.
@@ -0,0 +1 @@
{"route": "grim-subprocess", "language": "python", "protocol": "zwlr-screencopy-v1 (grim's client)", "output": "hh-chromium-52w2dm", "geometry": [1280, 720], "mode": "jpeg", "warmup": 10, "frames": 200, "frame_ms": {"min": 108.417, "p50": 183.652, "mean": 193.275, "p95": 266.099, "max": 370.433}, "fps_sustained": 5.17, "wall_ms": 38983.67, "bytes_p50": 129969, "persistent": false, "notes": "one grim process per frame, fork and encode inside the timed span; territory: --output; jpeg q80", "cpu_ms_per_frame": 5.1, "rss_peak_mb": 18.5, "rss_peak_kb_raw": 18940, "system_cpu_ms_per_frame": 387.9, "compositor_cpu_ms_per_frame": 182.8, "machine_cpu_pct": 199.0, "machine_cpu_pct_of_all_cores": 99.5, "frames_bytes_allocated_p50": null, "resource_notes": ""}
Whitespace-only changes.
@@ -0,0 +1 @@
{"route": "grim-subprocess", "language": "python", "protocol": "zwlr-screencopy-v1 (grim's client)", "output": "hh-chromium-52w2dm", "geometry": [1280, 720], "mode": "jpeg", "warmup": 10, "frames": 200, "frame_ms": {"min": 112.453, "p50": 184.392, "mean": 191.879, "p95": 255.498, "max": 387.236}, "fps_sustained": 5.21, "wall_ms": 38705.59, "bytes_p50": 129969, "persistent": false, "notes": "one grim process per frame, fork and encode inside the timed span; territory: --output; jpeg q80", "cpu_ms_per_frame": 5.057, "rss_peak_mb": 18.61, "rss_peak_kb_raw": 19056, "system_cpu_ms_per_frame": 384.75, "compositor_cpu_ms_per_frame": 181.6, "machine_cpu_pct": 198.8, "machine_cpu_pct_of_all_cores": 99.4, "frames_bytes_allocated_p50": null, "resource_notes": ""}
Whitespace-only changes.
@@ -0,0 +1 @@
{"route": "grim-subprocess", "language": "python", "protocol": "zwlr-screencopy-v1 (grim's client)", "output": "hh-chromium-52w2dm", "geometry": [1280, 720], "mode": "raw", "warmup": 10, "frames": 200, "frame_ms": {"min": 102.795, "p50": 185.109, "mean": 191.951, "p95": 290.786, "max": 349.333}, "fps_sustained": 5.21, "wall_ms": 38717.94, "bytes_p50": 2764816, "persistent": false, "notes": "one grim process per frame, fork and encode inside the timed span; territory: --output; ppm, unencoded", "cpu_ms_per_frame": 3.959, "rss_peak_mb": 20.57, "rss_peak_kb_raw": 21064, "system_cpu_ms_per_frame": 385.3, "compositor_cpu_ms_per_frame": 182.9, "machine_cpu_pct": 199.0, "machine_cpu_pct_of_all_cores": 99.5, "frames_bytes_allocated_p50": 3686400, "resource_notes": ""}
Whitespace-only changes.
@@ -0,0 +1 @@
{"route": "grim-subprocess", "language": "python", "protocol": "zwlr-screencopy-v1 (grim's client)", "output": "hh-chromium-52w2dm", "geometry": [1280, 720], "mode": "raw", "warmup": 10, "frames": 200, "frame_ms": {"min": 118.504, "p50": 185.863, "mean": 194.485, "p95": 292.598, "max": 338.927}, "fps_sustained": 5.14, "wall_ms": 39290.23, "bytes_p50": 2764816, "persistent": false, "notes": "one grim process per frame, fork and encode inside the timed span; territory: --output; ppm, unencoded", "cpu_ms_per_frame": 3.963, "rss_peak_mb": 20.51, "rss_peak_kb_raw": 21000, "system_cpu_ms_per_frame": 390.65, "compositor_cpu_ms_per_frame": 184.4, "machine_cpu_pct": 198.9, "machine_cpu_pct_of_all_cores": 99.4, "frames_bytes_allocated_p50": 3686400, "resource_notes": ""}
Whitespace-only changes.
@@ -0,0 +1 @@
{"route": "pywayland-in-process", "language": "python", "protocol": "ext-image-copy-capture-v1", "output": "hh-chromium-52w2dm", "geometry": [1280, 720], "mode": "jpeg", "warmup": 10, "frames": 200, "frame_ms": {"min": 30.949, "p50": 68.946, "mean": 90.495, "p95": 171.416, "max": 186.502}, "stage_ms": {"capture": 88.21, "encode": 2.201, "stitch": 0.0}, "fps_sustained": 11.05, "wall_ms": 18099.38, "bytes_p50": 106564, "persistent": true, "notes": "one connection, buffers reused, turbojpeg q80 in-process; 1 output(s); forcing a repaint per frame", "cpu_ms_per_frame": 1.772, "rss_peak_mb": 29.09, "rss_peak_kb_raw": 29792, "system_cpu_ms_per_frame": 179.8, "compositor_cpu_ms_per_frame": 86.9, "machine_cpu_pct": 198.7, "machine_cpu_pct_of_all_cores": 99.3, "frames_bytes_allocated_p50": 3686400, "resource_notes": ""}
Whitespace-only changes.
@@ -0,0 +1 @@
{"route": "pywayland-in-process", "language": "python", "protocol": "ext-image-copy-capture-v1", "output": "hh-chromium-52w2dm", "geometry": [1280, 720], "mode": "jpeg", "warmup": 10, "frames": 200, "frame_ms": {"min": 42.71, "p50": 71.213, "mean": 91.291, "p95": 175.214, "max": 183.55}, "stage_ms": {"capture": 88.969, "encode": 2.274, "stitch": 0.0}, "fps_sustained": 10.95, "wall_ms": 18258.58, "bytes_p50": 106564, "persistent": true, "notes": "one connection, buffers reused, turbojpeg q80 in-process; 1 output(s); forcing a repaint per frame", "cpu_ms_per_frame": 1.768, "rss_peak_mb": 29.04, "rss_peak_kb_raw": 29732, "system_cpu_ms_per_frame": 181.3, "compositor_cpu_ms_per_frame": 87.45, "machine_cpu_pct": 198.6, "machine_cpu_pct_of_all_cores": 99.3, "frames_bytes_allocated_p50": 3686400, "resource_notes": ""}
Whitespace-only changes.
@@ -0,0 +1 @@
{"route": "pywayland-in-process", "language": "python", "protocol": "ext-image-copy-capture-v1", "output": "hh-chromium-52w2dm", "geometry": [1280, 720], "mode": "raw", "warmup": 10, "frames": 200, "frame_ms": {"min": 31.081, "p50": 73.873, "mean": 89.249, "p95": 168.05, "max": 191.367}, "stage_ms": {"capture": 89.238, "encode": 0.0, "stitch": 0.0}, "fps_sustained": 11.2, "wall_ms": 17849.92, "bytes_p50": null, "persistent": true, "notes": "one connection, buffers reused, turbojpeg q80 in-process; 1 output(s); forcing a repaint per frame", "cpu_ms_per_frame": 0.331, "rss_peak_mb": 24.58, "rss_peak_kb_raw": 25172, "system_cpu_ms_per_frame": 177.4, "compositor_cpu_ms_per_frame": 86.45, "machine_cpu_pct": 198.8, "machine_cpu_pct_of_all_cores": 99.4, "frames_bytes_allocated_p50": 3686400, "resource_notes": ""}
Whitespace-only changes.
@@ -0,0 +1 @@
{"route": "pywayland-in-process", "language": "python", "protocol": "ext-image-copy-capture-v1", "output": "hh-chromium-52w2dm", "geometry": [1280, 720], "mode": "raw", "warmup": 10, "frames": 200, "frame_ms": {"min": 14.303, "p50": 68.229, "mean": 89.063, "p95": 175.379, "max": 197.257}, "stage_ms": {"capture": 89.052, "encode": 0.0, "stitch": 0.0}, "fps_sustained": 11.23, "wall_ms": 17812.7, "bytes_p50": null, "persistent": true, "notes": "one connection, buffers reused, turbojpeg q80 in-process; 1 output(s); forcing a repaint per frame", "cpu_ms_per_frame": 0.321, "rss_peak_mb": 24.57, "rss_peak_kb_raw": 25156, "system_cpu_ms_per_frame": 176.8, "compositor_cpu_ms_per_frame": 85.85, "machine_cpu_pct": 198.5, "machine_cpu_pct_of_all_cores": 99.3, "frames_bytes_allocated_p50": 3686400, "resource_notes": ""}
@@ -0,0 +1 @@
{"route": "pywayland-in-process", "language": "python", "protocol": "zwlr-screencopy-v1", "output": "hh-chromium-52w2dm", "geometry": [1280, 720], "mode": "jpeg", "warmup": 10, "frames": 200, "frame_ms": {"min": 44.66, "p50": 71.331, "mean": 92.49, "p95": 175.37, "max": 193.605}, "stage_ms": {"capture": 90.198, "encode": 2.244, "stitch": 0.0}, "fps_sustained": 10.81, "wall_ms": 18503.41, "bytes_p50": 106564, "persistent": true, "notes": "one connection, buffers reused, turbojpeg q80 in-process; 1 output(s); forcing a repaint per frame", "cpu_ms_per_frame": 1.781, "rss_peak_mb": 28.95, "rss_peak_kb_raw": 29640, "system_cpu_ms_per_frame": 183.6, "compositor_cpu_ms_per_frame": 88.8, "machine_cpu_pct": 198.4, "machine_cpu_pct_of_all_cores": 99.2, "frames_bytes_allocated_p50": 3686400, "resource_notes": ""}
@@ -0,0 +1 @@
{"route": "pywayland-in-process", "language": "python", "protocol": "zwlr-screencopy-v1", "output": "hh-chromium-52w2dm", "geometry": [1280, 720], "mode": "jpeg", "warmup": 10, "frames": 200, "frame_ms": {"min": 32.744, "p50": 73.495, "mean": 91.299, "p95": 169.886, "max": 185.797}, "stage_ms": {"capture": 89.038, "encode": 2.228, "stitch": 0.0}, "fps_sustained": 10.95, "wall_ms": 18260.25, "bytes_p50": 106564, "persistent": true, "notes": "one connection, buffers reused, turbojpeg q80 in-process; 1 output(s); forcing a repaint per frame", "cpu_ms_per_frame": 1.758, "rss_peak_mb": 28.98, "rss_peak_kb_raw": 29676, "system_cpu_ms_per_frame": 181.45, "compositor_cpu_ms_per_frame": 87.75, "machine_cpu_pct": 198.7, "machine_cpu_pct_of_all_cores": 99.4, "frames_bytes_allocated_p50": 3686400, "resource_notes": ""}
@@ -0,0 +1 @@
{"route": "pywayland-in-process", "language": "python", "protocol": "zwlr-screencopy-v1", "output": "hh-chromium-52w2dm", "geometry": [1280, 720], "mode": "raw", "warmup": 10, "frames": 200, "frame_ms": {"min": 31.169, "p50": 67.538, "mean": 90.985, "p95": 180.087, "max": 211.502}, "stage_ms": {"capture": 90.974, "encode": 0.0, "stitch": 0.0}, "fps_sustained": 10.99, "wall_ms": 18197.14, "bytes_p50": null, "persistent": true, "notes": "one connection, buffers reused, turbojpeg q80 in-process; 1 output(s); forcing a repaint per frame", "cpu_ms_per_frame": 0.343, "rss_peak_mb": 24.29, "rss_peak_kb_raw": 24872, "system_cpu_ms_per_frame": 180.95, "compositor_cpu_ms_per_frame": 87.5, "machine_cpu_pct": 198.9, "machine_cpu_pct_of_all_cores": 99.4, "frames_bytes_allocated_p50": 3686400, "resource_notes": ""}
@@ -0,0 +1 @@
{"route": "pywayland-in-process", "language": "python", "protocol": "zwlr-screencopy-v1", "output": "hh-chromium-52w2dm", "geometry": [1280, 720], "mode": "raw", "warmup": 10, "frames": 200, "frame_ms": {"min": 30.01, "p50": 72.161, "mean": 91.595, "p95": 175.282, "max": 227.322}, "stage_ms": {"capture": 91.582, "encode": 0.0, "stitch": 0.0}, "fps_sustained": 10.92, "wall_ms": 18319.19, "bytes_p50": null, "persistent": true, "notes": "one connection, buffers reused, turbojpeg q80 in-process; 1 output(s); forcing a repaint per frame", "cpu_ms_per_frame": 0.339, "rss_peak_mb": 24.43, "rss_peak_kb_raw": 25012, "system_cpu_ms_per_frame": 181.85, "compositor_cpu_ms_per_frame": 88.15, "machine_cpu_pct": 198.5, "machine_cpu_pct_of_all_cores": 99.3, "frames_bytes_allocated_p50": 3686400, "resource_notes": ""}
@@ -0,0 +1,674 @@
[
{
"route": "grim-subprocess",
"description": "grim subprocess per frame (shipped)",
"cell": "output-raw",
"run": 1,
"command": "/nix/store/i7d46siafmlrgcmj9nzvwj7y64p34yaz-python3-3.14.7-env/bin/python3 /var/lib/hermes/worktrees/hhs-capture-speed/spikes/capture-speed/producers/grim_producer.py --output hh-chromium-52w2dm --frames 200 --warmup 10 --mode raw --json",
"rc": 0,
"elapsed_s": 40.9,
"language": "python",
"protocol": "zwlr-screencopy-v1 (grim's client)",
"output": "hh-chromium-52w2dm",
"geometry": [
1280,
720
],
"mode": "raw",
"warmup": 10,
"frames": 200,
"frame_ms": {
"min": 102.795,
"p50": 185.109,
"mean": 191.951,
"p95": 290.786,
"max": 349.333
},
"fps_sustained": 5.21,
"wall_ms": 38717.94,
"bytes_p50": 2764816,
"persistent": false,
"notes": "one grim process per frame, fork and encode inside the timed span; territory: --output; ppm, unencoded",
"cpu_ms_per_frame": 3.959,
"rss_peak_mb": 20.57,
"rss_peak_kb_raw": 21064,
"system_cpu_ms_per_frame": 385.3,
"compositor_cpu_ms_per_frame": 182.9,
"machine_cpu_pct": 199.0,
"machine_cpu_pct_of_all_cores": 99.5,
"frames_bytes_allocated_p50": 3686400,
"resource_notes": ""
},
{
"route": "grim-subprocess",
"description": "grim subprocess per frame (shipped)",
"cell": "output-raw",
"run": 2,
"command": "/nix/store/i7d46siafmlrgcmj9nzvwj7y64p34yaz-python3-3.14.7-env/bin/python3 /var/lib/hermes/worktrees/hhs-capture-speed/spikes/capture-speed/producers/grim_producer.py --output hh-chromium-52w2dm --frames 200 --warmup 10 --mode raw --json",
"rc": 0,
"elapsed_s": 41.3,
"language": "python",
"protocol": "zwlr-screencopy-v1 (grim's client)",
"output": "hh-chromium-52w2dm",
"geometry": [
1280,
720
],
"mode": "raw",
"warmup": 10,
"frames": 200,
"frame_ms": {
"min": 118.504,
"p50": 185.863,
"mean": 194.485,
"p95": 292.598,
"max": 338.927
},
"fps_sustained": 5.14,
"wall_ms": 39290.23,
"bytes_p50": 2764816,
"persistent": false,
"notes": "one grim process per frame, fork and encode inside the timed span; territory: --output; ppm, unencoded",
"cpu_ms_per_frame": 3.963,
"rss_peak_mb": 20.51,
"rss_peak_kb_raw": 21000,
"system_cpu_ms_per_frame": 390.65,
"compositor_cpu_ms_per_frame": 184.4,
"machine_cpu_pct": 198.9,
"machine_cpu_pct_of_all_cores": 99.4,
"frames_bytes_allocated_p50": 3686400,
"resource_notes": ""
},
{
"route": "grim-subprocess",
"description": "grim subprocess per frame (shipped)",
"cell": "output-jpeg",
"run": 1,
"command": "/nix/store/i7d46siafmlrgcmj9nzvwj7y64p34yaz-python3-3.14.7-env/bin/python3 /var/lib/hermes/worktrees/hhs-capture-speed/spikes/capture-speed/producers/grim_producer.py --output hh-chromium-52w2dm --frames 200 --warmup 10 --mode jpeg --json",
"rc": 0,
"elapsed_s": 41.0,
"language": "python",
"protocol": "zwlr-screencopy-v1 (grim's client)",
"output": "hh-chromium-52w2dm",
"geometry": [
1280,
720
],
"mode": "jpeg",
"warmup": 10,
"frames": 200,
"frame_ms": {
"min": 108.417,
"p50": 183.652,
"mean": 193.275,
"p95": 266.099,
"max": 370.433
},
"fps_sustained": 5.17,
"wall_ms": 38983.67,
"bytes_p50": 129969,
"persistent": false,
"notes": "one grim process per frame, fork and encode inside the timed span; territory: --output; jpeg q80",
"cpu_ms_per_frame": 5.1,
"rss_peak_mb": 18.5,
"rss_peak_kb_raw": 18940,
"system_cpu_ms_per_frame": 387.9,
"compositor_cpu_ms_per_frame": 182.8,
"machine_cpu_pct": 199.0,
"machine_cpu_pct_of_all_cores": 99.5,
"frames_bytes_allocated_p50": null,
"resource_notes": ""
},
{
"route": "grim-subprocess",
"description": "grim subprocess per frame (shipped)",
"cell": "output-jpeg",
"run": 2,
"command": "/nix/store/i7d46siafmlrgcmj9nzvwj7y64p34yaz-python3-3.14.7-env/bin/python3 /var/lib/hermes/worktrees/hhs-capture-speed/spikes/capture-speed/producers/grim_producer.py --output hh-chromium-52w2dm --frames 200 --warmup 10 --mode jpeg --json",
"rc": 0,
"elapsed_s": 40.8,
"language": "python",
"protocol": "zwlr-screencopy-v1 (grim's client)",
"output": "hh-chromium-52w2dm",
"geometry": [
1280,
720
],
"mode": "jpeg",
"warmup": 10,
"frames": 200,
"frame_ms": {
"min": 112.453,
"p50": 184.392,
"mean": 191.879,
"p95": 255.498,
"max": 387.236
},
"fps_sustained": 5.21,
"wall_ms": 38705.59,
"bytes_p50": 129969,
"persistent": false,
"notes": "one grim process per frame, fork and encode inside the timed span; territory: --output; jpeg q80",
"cpu_ms_per_frame": 5.057,
"rss_peak_mb": 18.61,
"rss_peak_kb_raw": 19056,
"system_cpu_ms_per_frame": 384.75,
"compositor_cpu_ms_per_frame": 181.6,
"machine_cpu_pct": 198.8,
"machine_cpu_pct_of_all_cores": 99.4,
"frames_bytes_allocated_p50": null,
"resource_notes": ""
},
{
"route": "pywayland-in-process",
"description": "pywayland in-process, zwlr-screencopy-v1",
"cell": "output-raw",
"run": 1,
"command": "/nix/store/i7d46siafmlrgcmj9nzvwj7y64p34yaz-python3-3.14.7-env/bin/python3 /var/lib/hermes/worktrees/hhs-capture-speed/spikes/capture-speed/producers/wayland_producer.py --output hh-chromium-52w2dm --protocol screencopy --frames 200 --warmup 10 --mode raw --json",
"rc": 0,
"elapsed_s": 19.6,
"language": "python",
"protocol": "zwlr-screencopy-v1",
"output": "hh-chromium-52w2dm",
"geometry": [
1280,
720
],
"mode": "raw",
"warmup": 10,
"frames": 200,
"frame_ms": {
"min": 31.169,
"p50": 67.538,
"mean": 90.985,
"p95": 180.087,
"max": 211.502
},
"stage_ms": {
"capture": 90.974,
"encode": 0.0,
"stitch": 0.0
},
"fps_sustained": 10.99,
"wall_ms": 18197.14,
"bytes_p50": null,
"persistent": true,
"notes": "one connection, buffers reused, turbojpeg q80 in-process; 1 output(s); forcing a repaint per frame",
"cpu_ms_per_frame": 0.343,
"rss_peak_mb": 24.29,
"rss_peak_kb_raw": 24872,
"system_cpu_ms_per_frame": 180.95,
"compositor_cpu_ms_per_frame": 87.5,
"machine_cpu_pct": 198.9,
"machine_cpu_pct_of_all_cores": 99.4,
"frames_bytes_allocated_p50": 3686400,
"resource_notes": ""
},
{
"route": "pywayland-in-process",
"description": "pywayland in-process, zwlr-screencopy-v1",
"cell": "output-raw",
"run": 2,
"command": "/nix/store/i7d46siafmlrgcmj9nzvwj7y64p34yaz-python3-3.14.7-env/bin/python3 /var/lib/hermes/worktrees/hhs-capture-speed/spikes/capture-speed/producers/wayland_producer.py --output hh-chromium-52w2dm --protocol screencopy --frames 200 --warmup 10 --mode raw --json",
"rc": 0,
"elapsed_s": 19.4,
"language": "python",
"protocol": "zwlr-screencopy-v1",
"output": "hh-chromium-52w2dm",
"geometry": [
1280,
720
],
"mode": "raw",
"warmup": 10,
"frames": 200,
"frame_ms": {
"min": 30.01,
"p50": 72.161,
"mean": 91.595,
"p95": 175.282,
"max": 227.322
},
"stage_ms": {
"capture": 91.582,
"encode": 0.0,
"stitch": 0.0
},
"fps_sustained": 10.92,
"wall_ms": 18319.19,
"bytes_p50": null,
"persistent": true,
"notes": "one connection, buffers reused, turbojpeg q80 in-process; 1 output(s); forcing a repaint per frame",
"cpu_ms_per_frame": 0.339,
"rss_peak_mb": 24.43,
"rss_peak_kb_raw": 25012,
"system_cpu_ms_per_frame": 181.85,
"compositor_cpu_ms_per_frame": 88.15,
"machine_cpu_pct": 198.5,
"machine_cpu_pct_of_all_cores": 99.3,
"frames_bytes_allocated_p50": 3686400,
"resource_notes": ""
},
{
"route": "pywayland-in-process",
"description": "pywayland in-process, zwlr-screencopy-v1",
"cell": "output-jpeg",
"run": 1,
"command": "/nix/store/i7d46siafmlrgcmj9nzvwj7y64p34yaz-python3-3.14.7-env/bin/python3 /var/lib/hermes/worktrees/hhs-capture-speed/spikes/capture-speed/producers/wayland_producer.py --output hh-chromium-52w2dm --protocol screencopy --frames 200 --warmup 10 --mode jpeg --json",
"rc": 0,
"elapsed_s": 19.8,
"language": "python",
"protocol": "zwlr-screencopy-v1",
"output": "hh-chromium-52w2dm",
"geometry": [
1280,
720
],
"mode": "jpeg",
"warmup": 10,
"frames": 200,
"frame_ms": {
"min": 44.66,
"p50": 71.331,
"mean": 92.49,
"p95": 175.37,
"max": 193.605
},
"stage_ms": {
"capture": 90.198,
"encode": 2.244,
"stitch": 0.0
},
"fps_sustained": 10.81,
"wall_ms": 18503.41,
"bytes_p50": 106564,
"persistent": true,
"notes": "one connection, buffers reused, turbojpeg q80 in-process; 1 output(s); forcing a repaint per frame",
"cpu_ms_per_frame": 1.781,
"rss_peak_mb": 28.95,
"rss_peak_kb_raw": 29640,
"system_cpu_ms_per_frame": 183.6,
"compositor_cpu_ms_per_frame": 88.8,
"machine_cpu_pct": 198.4,
"machine_cpu_pct_of_all_cores": 99.2,
"frames_bytes_allocated_p50": 3686400,
"resource_notes": ""
},
{
"route": "pywayland-in-process",
"description": "pywayland in-process, zwlr-screencopy-v1",
"cell": "output-jpeg",
"run": 2,
"command": "/nix/store/i7d46siafmlrgcmj9nzvwj7y64p34yaz-python3-3.14.7-env/bin/python3 /var/lib/hermes/worktrees/hhs-capture-speed/spikes/capture-speed/producers/wayland_producer.py --output hh-chromium-52w2dm --protocol screencopy --frames 200 --warmup 10 --mode jpeg --json",
"rc": 0,
"elapsed_s": 19.6,
"language": "python",
"protocol": "zwlr-screencopy-v1",
"output": "hh-chromium-52w2dm",
"geometry": [
1280,
720
],
"mode": "jpeg",
"warmup": 10,
"frames": 200,
"frame_ms": {
"min": 32.744,
"p50": 73.495,
"mean": 91.299,
"p95": 169.886,
"max": 185.797
},
"stage_ms": {
"capture": 89.038,
"encode": 2.228,
"stitch": 0.0
},
"fps_sustained": 10.95,
"wall_ms": 18260.25,
"bytes_p50": 106564,
"persistent": true,
"notes": "one connection, buffers reused, turbojpeg q80 in-process; 1 output(s); forcing a repaint per frame",
"cpu_ms_per_frame": 1.758,
"rss_peak_mb": 28.98,
"rss_peak_kb_raw": 29676,
"system_cpu_ms_per_frame": 181.45,
"compositor_cpu_ms_per_frame": 87.75,
"machine_cpu_pct": 198.7,
"machine_cpu_pct_of_all_cores": 99.4,
"frames_bytes_allocated_p50": 3686400,
"resource_notes": ""
},
{
"route": "pywayland-in-process",
"description": "pywayland in-process, ext-image-copy-capture-v1",
"cell": "output-raw",
"run": 1,
"command": "/nix/store/i7d46siafmlrgcmj9nzvwj7y64p34yaz-python3-3.14.7-env/bin/python3 /var/lib/hermes/worktrees/hhs-capture-speed/spikes/capture-speed/producers/wayland_producer.py --output hh-chromium-52w2dm --protocol ext --frames 200 --warmup 10 --mode raw --json",
"rc": 0,
"elapsed_s": 19.0,
"language": "python",
"protocol": "ext-image-copy-capture-v1",
"output": "hh-chromium-52w2dm",
"geometry": [
1280,
720
],
"mode": "raw",
"warmup": 10,
"frames": 200,
"frame_ms": {
"min": 31.081,
"p50": 73.873,
"mean": 89.249,
"p95": 168.05,
"max": 191.367
},
"stage_ms": {
"capture": 89.238,
"encode": 0.0,
"stitch": 0.0
},
"fps_sustained": 11.2,
"wall_ms": 17849.92,
"bytes_p50": null,
"persistent": true,
"notes": "one connection, buffers reused, turbojpeg q80 in-process; 1 output(s); forcing a repaint per frame",
"cpu_ms_per_frame": 0.331,
"rss_peak_mb": 24.58,
"rss_peak_kb_raw": 25172,
"system_cpu_ms_per_frame": 177.4,
"compositor_cpu_ms_per_frame": 86.45,
"machine_cpu_pct": 198.8,
"machine_cpu_pct_of_all_cores": 99.4,
"frames_bytes_allocated_p50": 3686400,
"resource_notes": ""
},
{
"route": "pywayland-in-process",
"description": "pywayland in-process, ext-image-copy-capture-v1",
"cell": "output-raw",
"run": 2,
"command": "/nix/store/i7d46siafmlrgcmj9nzvwj7y64p34yaz-python3-3.14.7-env/bin/python3 /var/lib/hermes/worktrees/hhs-capture-speed/spikes/capture-speed/producers/wayland_producer.py --output hh-chromium-52w2dm --protocol ext --frames 200 --warmup 10 --mode raw --json",
"rc": 0,
"elapsed_s": 18.9,
"language": "python",
"protocol": "ext-image-copy-capture-v1",
"output": "hh-chromium-52w2dm",
"geometry": [
1280,
720
],
"mode": "raw",
"warmup": 10,
"frames": 200,
"frame_ms": {
"min": 14.303,
"p50": 68.229,
"mean": 89.063,
"p95": 175.379,
"max": 197.257
},
"stage_ms": {
"capture": 89.052,
"encode": 0.0,
"stitch": 0.0
},
"fps_sustained": 11.23,
"wall_ms": 17812.7,
"bytes_p50": null,
"persistent": true,
"notes": "one connection, buffers reused, turbojpeg q80 in-process; 1 output(s); forcing a repaint per frame",
"cpu_ms_per_frame": 0.321,
"rss_peak_mb": 24.57,
"rss_peak_kb_raw": 25156,
"system_cpu_ms_per_frame": 176.8,
"compositor_cpu_ms_per_frame": 85.85,
"machine_cpu_pct": 198.5,
"machine_cpu_pct_of_all_cores": 99.3,
"frames_bytes_allocated_p50": 3686400,
"resource_notes": ""
},
{
"route": "pywayland-in-process",
"description": "pywayland in-process, ext-image-copy-capture-v1",
"cell": "output-jpeg",
"run": 1,
"command": "/nix/store/i7d46siafmlrgcmj9nzvwj7y64p34yaz-python3-3.14.7-env/bin/python3 /var/lib/hermes/worktrees/hhs-capture-speed/spikes/capture-speed/producers/wayland_producer.py --output hh-chromium-52w2dm --protocol ext --frames 200 --warmup 10 --mode jpeg --json",
"rc": 0,
"elapsed_s": 19.5,
"language": "python",
"protocol": "ext-image-copy-capture-v1",
"output": "hh-chromium-52w2dm",
"geometry": [
1280,
720
],
"mode": "jpeg",
"warmup": 10,
"frames": 200,
"frame_ms": {
"min": 30.949,
"p50": 68.946,
"mean": 90.495,
"p95": 171.416,
"max": 186.502
},
"stage_ms": {
"capture": 88.21,
"encode": 2.201,
"stitch": 0.0
},
"fps_sustained": 11.05,
"wall_ms": 18099.38,
"bytes_p50": 106564,
"persistent": true,
"notes": "one connection, buffers reused, turbojpeg q80 in-process; 1 output(s); forcing a repaint per frame",
"cpu_ms_per_frame": 1.772,
"rss_peak_mb": 29.09,
"rss_peak_kb_raw": 29792,
"system_cpu_ms_per_frame": 179.8,
"compositor_cpu_ms_per_frame": 86.9,
"machine_cpu_pct": 198.7,
"machine_cpu_pct_of_all_cores": 99.3,
"frames_bytes_allocated_p50": 3686400,
"resource_notes": ""
},
{
"route": "pywayland-in-process",
"description": "pywayland in-process, ext-image-copy-capture-v1",
"cell": "output-jpeg",
"run": 2,
"command": "/nix/store/i7d46siafmlrgcmj9nzvwj7y64p34yaz-python3-3.14.7-env/bin/python3 /var/lib/hermes/worktrees/hhs-capture-speed/spikes/capture-speed/producers/wayland_producer.py --output hh-chromium-52w2dm --protocol ext --frames 200 --warmup 10 --mode jpeg --json",
"rc": 0,
"elapsed_s": 19.5,
"language": "python",
"protocol": "ext-image-copy-capture-v1",
"output": "hh-chromium-52w2dm",
"geometry": [
1280,
720
],
"mode": "jpeg",
"warmup": 10,
"frames": 200,
"frame_ms": {
"min": 42.71,
"p50": 71.213,
"mean": 91.291,
"p95": 175.214,
"max": 183.55
},
"stage_ms": {
"capture": 88.969,
"encode": 2.274,
"stitch": 0.0
},
"fps_sustained": 10.95,
"wall_ms": 18258.58,
"bytes_p50": 106564,
"persistent": true,
"notes": "one connection, buffers reused, turbojpeg q80 in-process; 1 output(s); forcing a repaint per frame",
"cpu_ms_per_frame": 1.768,
"rss_peak_mb": 29.04,
"rss_peak_kb_raw": 29732,
"system_cpu_ms_per_frame": 181.3,
"compositor_cpu_ms_per_frame": 87.45,
"machine_cpu_pct": 198.6,
"machine_cpu_pct_of_all_cores": 99.3,
"frames_bytes_allocated_p50": 3686400,
"resource_notes": ""
},
{
"route": "grim-rs",
"description": "grim-rs crate (Rust), long-lived process",
"cell": "output-raw",
"run": 1,
"command": "/var/lib/hermes/worktrees/hhs-capture-speed/spikes/capture-speed/001a-grim-rs/target/release/grimrs-capture --output hh-chromium-52w2dm --frames 200 --warmup 10 --mode raw --json",
"rc": 0,
"elapsed_s": 42.1,
"language": "rust",
"protocol": "ext-image-copy-capture-v1 (auto)",
"output": "hh-chromium-52w2dm",
"geometry": [
1280,
720
],
"mode": "raw",
"warmup": 10,
"frames": 200,
"frame_ms": {
"min": 112.37,
"p50": 188.581,
"mean": 199.48,
"p95": 289.422,
"max": 358.439
},
"fps_sustained": 5.013,
"wall_ms": 39896.261,
"cpu_ms_per_frame": 1.677,
"bytes_p50": null,
"persistent": true,
"rss_peak_mb": 746.18,
"rss_peak_kb_raw": 764084,
"system_cpu_ms_per_frame": 396.45,
"compositor_cpu_ms_per_frame": 188.85,
"machine_cpu_pct": 198.7,
"frames_bytes_allocated_p50": 3686400,
"notes": "grim-rs 0.2.0 via Grim::new() (auto-detect); copy target is a wl_shm buffer created inside each capture_* call, not a dma-buf the producer imports. Cursor not overlaid (the frozen argv has no cursor flag; hhstream-live's grim command passes -c). JPEG is Grim::to_jpeg, quality 80, the crate default and the live-stream setting. Raw timer stops when capture_* returns, which is after the crate copies the mapping into a Vec and swizzles XRGB/ARGB to RGBA (and applies any output transform or y-invert). The public API does not expose the mapped buffer before that, so raw includes work the contract says to leave outside the timer. Percentiles are linear interpolation on the sorted samples. cpu_ms_per_frame is getrusage(RUSAGE_SELF) utime+stime over the measured loop only, divided by frames. Frame buffers are freed between frames; that free is inside wall_ms and outside frame_ms. Outputs: Virtual-1@0,0 1280x720 scale 1; hh-chromium-gqkxz3@1280,0 1280x720 scale 1; hh-chromium-52w2dm@2560,0 1280x720 scale 1. No GPU on this machine; dma-buf was not used and is not claimed. This process does not move the pointer and does not start or stop a session; the pixels are whatever the compositor is already showing. Last warmup frame nonzero bytes: 3679565."
},
{
"route": "grim-rs",
"description": "grim-rs crate (Rust), long-lived process",
"cell": "output-raw",
"run": 2,
"command": "/var/lib/hermes/worktrees/hhs-capture-speed/spikes/capture-speed/001a-grim-rs/target/release/grimrs-capture --output hh-chromium-52w2dm --frames 200 --warmup 10 --mode raw --json",
"rc": 0,
"elapsed_s": 43.4,
"language": "rust",
"protocol": "ext-image-copy-capture-v1 (auto)",
"output": "hh-chromium-52w2dm",
"geometry": [
1280,
720
],
"mode": "raw",
"warmup": 10,
"frames": 200,
"frame_ms": {
"min": 118.647,
"p50": 187.814,
"mean": 205.548,
"p95": 298.366,
"max": 361.436
},
"fps_sustained": 4.865,
"wall_ms": 41109.705,
"cpu_ms_per_frame": 1.652,
"bytes_p50": null,
"persistent": true,
"rss_peak_mb": 746.0,
"rss_peak_kb_raw": 763904,
"system_cpu_ms_per_frame": 408.6,
"compositor_cpu_ms_per_frame": 194.3,
"machine_cpu_pct": 198.8,
"frames_bytes_allocated_p50": 3686400,
"notes": "grim-rs 0.2.0 via Grim::new() (auto-detect); copy target is a wl_shm buffer created inside each capture_* call, not a dma-buf the producer imports. Cursor not overlaid (the frozen argv has no cursor flag; hhstream-live's grim command passes -c). JPEG is Grim::to_jpeg, quality 80, the crate default and the live-stream setting. Raw timer stops when capture_* returns, which is after the crate copies the mapping into a Vec and swizzles XRGB/ARGB to RGBA (and applies any output transform or y-invert). The public API does not expose the mapped buffer before that, so raw includes work the contract says to leave outside the timer. Percentiles are linear interpolation on the sorted samples. cpu_ms_per_frame is getrusage(RUSAGE_SELF) utime+stime over the measured loop only, divided by frames. Frame buffers are freed between frames; that free is inside wall_ms and outside frame_ms. Outputs: Virtual-1@0,0 1280x720 scale 1; hh-chromium-gqkxz3@1280,0 1280x720 scale 1; hh-chromium-52w2dm@2560,0 1280x720 scale 1. No GPU on this machine; dma-buf was not used and is not claimed. This process does not move the pointer and does not start or stop a session; the pixels are whatever the compositor is already showing. Last warmup frame nonzero bytes: 3679565."
},
{
"route": "grim-rs",
"description": "grim-rs crate (Rust), long-lived process",
"cell": "output-jpeg",
"run": 1,
"command": "/var/lib/hermes/worktrees/hhs-capture-speed/spikes/capture-speed/001a-grim-rs/target/release/grimrs-capture --output hh-chromium-52w2dm --frames 200 --warmup 10 --mode jpeg --json",
"rc": 0,
"elapsed_s": 42.2,
"language": "rust",
"protocol": "ext-image-copy-capture-v1 (auto)",
"output": "hh-chromium-52w2dm",
"geometry": [
1280,
720
],
"mode": "jpeg",
"warmup": 10,
"frames": 200,
"frame_ms": {
"min": 95.334,
"p50": 186.693,
"mean": 199.129,
"p95": 282.41,
"max": 411.263
},
"fps_sustained": 5.022,
"wall_ms": 39825.916,
"cpu_ms_per_frame": 6.7,
"bytes_p50": 107829.0,
"persistent": true,
"rss_peak_mb": 746.3,
"rss_peak_kb_raw": 764208,
"system_cpu_ms_per_frame": 396.0,
"compositor_cpu_ms_per_frame": 187.75,
"machine_cpu_pct": 198.9,
"frames_bytes_allocated_p50": 3686400,
"notes": "grim-rs 0.2.0 via Grim::new() (auto-detect); copy target is a wl_shm buffer created inside each capture_* call, not a dma-buf the producer imports. Cursor not overlaid (the frozen argv has no cursor flag; hhstream-live's grim command passes -c). JPEG is Grim::to_jpeg, quality 80, the crate default and the live-stream setting. Raw timer stops when capture_* returns, which is after the crate copies the mapping into a Vec and swizzles XRGB/ARGB to RGBA (and applies any output transform or y-invert). The public API does not expose the mapped buffer before that, so raw includes work the contract says to leave outside the timer. Percentiles are linear interpolation on the sorted samples. cpu_ms_per_frame is getrusage(RUSAGE_SELF) utime+stime over the measured loop only, divided by frames. Frame buffers are freed between frames; that free is inside wall_ms and outside frame_ms. Outputs: Virtual-1@0,0 1280x720 scale 1; hh-chromium-gqkxz3@1280,0 1280x720 scale 1; hh-chromium-52w2dm@2560,0 1280x720 scale 1. No GPU on this machine; dma-buf was not used and is not claimed. This process does not move the pointer and does not start or stop a session; the pixels are whatever the compositor is already showing. Last warmup frame nonzero bytes: 105852."
},
{
"route": "grim-rs",
"description": "grim-rs crate (Rust), long-lived process",
"cell": "output-jpeg",
"run": 2,
"command": "/var/lib/hermes/worktrees/hhs-capture-speed/spikes/capture-speed/001a-grim-rs/target/release/grimrs-capture --output hh-chromium-52w2dm --frames 200 --warmup 10 --mode jpeg --json",
"rc": 0,
"elapsed_s": 42.2,
"language": "rust",
"protocol": "ext-image-copy-capture-v1 (auto)",
"output": "hh-chromium-52w2dm",
"geometry": [
1280,
720
],
"mode": "jpeg",
"warmup": 10,
"frames": 200,
"frame_ms": {
"min": 104.636,
"p50": 184.629,
"mean": 198.848,
"p95": 306.666,
"max": 368.522
},
"fps_sustained": 5.029,
"wall_ms": 39771.382,
"cpu_ms_per_frame": 6.67,
"bytes_p50": 107829.0,
"persistent": true,
"rss_peak_mb": 746.35,
"rss_peak_kb_raw": 764264,
"system_cpu_ms_per_frame": 395.55,
"compositor_cpu_ms_per_frame": 187.5,
"machine_cpu_pct": 198.9,
"frames_bytes_allocated_p50": 3686400,
"notes": "grim-rs 0.2.0 via Grim::new() (auto-detect); copy target is a wl_shm buffer created inside each capture_* call, not a dma-buf the producer imports. Cursor not overlaid (the frozen argv has no cursor flag; hhstream-live's grim command passes -c). JPEG is Grim::to_jpeg, quality 80, the crate default and the live-stream setting. Raw timer stops when capture_* returns, which is after the crate copies the mapping into a Vec and swizzles XRGB/ARGB to RGBA (and applies any output transform or y-invert). The public API does not expose the mapped buffer before that, so raw includes work the contract says to leave outside the timer. Percentiles are linear interpolation on the sorted samples. cpu_ms_per_frame is getrusage(RUSAGE_SELF) utime+stime over the measured loop only, divided by frames. Frame buffers are freed between frames; that free is inside wall_ms and outside frame_ms. Outputs: Virtual-1@0,0 1280x720 scale 1; hh-chromium-gqkxz3@1280,0 1280x720 scale 1; hh-chromium-52w2dm@2560,0 1280x720 scale 1. No GPU on this machine; dma-buf was not used and is not claimed. This process does not move the pointer and does not start or stop a session; the pixels are whatever the compositor is already showing. Last warmup frame nonzero bytes: 105852."
}
]
@@ -1,7 +1,7 @@
{
"window_ms": 10000.0,
"system_cpu_ms": 17920.0,
"compositor_cpu_ms": 6900.0,
"system_cpu_pct": 179.2,
"compositor_cpu_pct": 69.0
"system_cpu_ms": 19980.0,
"compositor_cpu_ms": 2630.0,
"system_cpu_pct": 199.8,
"compositor_cpu_pct": 26.3
}
@@ -0,0 +1,12 @@
grimrs-capture: route grim-rs protocol ext-image-copy-capture-v1 (auto) mode jpeg target all warmup 10 frames 200
grimrs-capture: outputs Virtual-1@0,0 1280x720 scale 1
grimrs-capture: warmup 1/10 1280x720 bytes 218563 nonzero 217104
grimrs-capture: warmup 2/10 1280x720 bytes 218563 nonzero 217104
grimrs-capture: warmup 3/10 1280x720 bytes 218563 nonzero 217104
grimrs-capture: warmup 4/10 1280x720 bytes 218563 nonzero 217104
grimrs-capture: warmup 5/10 1280x720 bytes 218563 nonzero 217104
grimrs-capture: warmup 6/10 1280x720 bytes 218563 nonzero 217104
grimrs-capture: warmup 7/10 1280x720 bytes 218563 nonzero 217104
grimrs-capture: warmup 8/10 1280x720 bytes 218563 nonzero 217104
grimrs-capture: warmup 9/10 1280x720 bytes 218563 nonzero 217104
grimrs-capture: warmup 10/10 1280x720 bytes 218563 nonzero 217104
@@ -0,0 +1 @@
{"route":"grim-rs","language":"rust","protocol":"ext-image-copy-capture-v1 (auto)","output":"all","geometry":[1280,720],"mode":"jpeg","warmup":10,"frames":200,"frame_ms":{"min":43.383,"p50":78.211,"mean":76.591,"p95":91.117,"max":120.327},"fps_sustained":13.056,"wall_ms":15318.39,"cpu_ms_per_frame":7.657,"bytes_p50":218563.0,"persistent":true,"rss_peak_mb":745.62,"rss_peak_kb_raw":763516,"system_cpu_ms_per_frame":153.05,"compositor_cpu_ms_per_frame":48.15,"machine_cpu_pct":199.8,"frames_bytes_allocated_p50":3686400,"notes":"grim-rs 0.2.0 via Grim::new() (auto-detect); copy target is a wl_shm buffer created inside each capture_* call, not a dma-buf the producer imports. Cursor not overlaid (the frozen argv has no cursor flag; hhstream-live's grim command passes -c). JPEG is Grim::to_jpeg, quality 80, the crate default and the live-stream setting. Raw timer stops when capture_* returns, which is after the crate copies the mapping into a Vec and swizzles XRGB/ARGB to RGBA (and applies any output transform or y-invert). The public API does not expose the mapped buffer before that, so raw includes work the contract says to leave outside the timer. Percentiles are linear interpolation on the sorted samples. cpu_ms_per_frame is getrusage(RUSAGE_SELF) utime+stime over the measured loop only, divided by frames. Frame buffers are freed between frames; that free is inside wall_ms and outside frame_ms. Outputs: Virtual-1@0,0 1280x720 scale 1. No GPU on this machine; dma-buf was not used and is not claimed. This process does not move the pointer and does not start or stop a session; the pixels are whatever the compositor is already showing. Last warmup frame nonzero bytes: 217104."}
@@ -0,0 +1,12 @@
grimrs-capture: route grim-rs protocol ext-image-copy-capture-v1 (auto) mode jpeg target all warmup 10 frames 200
grimrs-capture: outputs Virtual-1@0,0 1280x720 scale 1
grimrs-capture: warmup 1/10 1280x720 bytes 218563 nonzero 217104
grimrs-capture: warmup 2/10 1280x720 bytes 218563 nonzero 217104
grimrs-capture: warmup 3/10 1280x720 bytes 218563 nonzero 217104
grimrs-capture: warmup 4/10 1280x720 bytes 218563 nonzero 217104
grimrs-capture: warmup 5/10 1280x720 bytes 218563 nonzero 217104
grimrs-capture: warmup 6/10 1280x720 bytes 218563 nonzero 217104
grimrs-capture: warmup 7/10 1280x720 bytes 218563 nonzero 217104
grimrs-capture: warmup 8/10 1280x720 bytes 218563 nonzero 217104
grimrs-capture: warmup 9/10 1280x720 bytes 218563 nonzero 217104
grimrs-capture: warmup 10/10 1280x720 bytes 218563 nonzero 217104
@@ -0,0 +1 @@
{"route":"grim-rs","language":"rust","protocol":"ext-image-copy-capture-v1 (auto)","output":"all","geometry":[1280,720],"mode":"jpeg","warmup":10,"frames":200,"frame_ms":{"min":41.516,"p50":75.597,"mean":74.024,"p95":92.187,"max":126.722},"fps_sustained":13.509,"wall_ms":14805.03,"cpu_ms_per_frame":7.682,"bytes_p50":218563.0,"persistent":true,"rss_peak_mb":745.52,"rss_peak_kb_raw":763416,"system_cpu_ms_per_frame":147.85,"compositor_cpu_ms_per_frame":47.5,"machine_cpu_pct":199.7,"frames_bytes_allocated_p50":3686400,"notes":"grim-rs 0.2.0 via Grim::new() (auto-detect); copy target is a wl_shm buffer created inside each capture_* call, not a dma-buf the producer imports. Cursor not overlaid (the frozen argv has no cursor flag; hhstream-live's grim command passes -c). JPEG is Grim::to_jpeg, quality 80, the crate default and the live-stream setting. Raw timer stops when capture_* returns, which is after the crate copies the mapping into a Vec and swizzles XRGB/ARGB to RGBA (and applies any output transform or y-invert). The public API does not expose the mapped buffer before that, so raw includes work the contract says to leave outside the timer. Percentiles are linear interpolation on the sorted samples. cpu_ms_per_frame is getrusage(RUSAGE_SELF) utime+stime over the measured loop only, divided by frames. Frame buffers are freed between frames; that free is inside wall_ms and outside frame_ms. Outputs: Virtual-1@0,0 1280x720 scale 1. No GPU on this machine; dma-buf was not used and is not claimed. This process does not move the pointer and does not start or stop a session; the pixels are whatever the compositor is already showing. Last warmup frame nonzero bytes: 217104."}
@@ -0,0 +1,12 @@
grimrs-capture: route grim-rs protocol ext-image-copy-capture-v1 (auto) mode raw target all warmup 10 frames 200
grimrs-capture: outputs Virtual-1@0,0 1280x720 scale 1
grimrs-capture: warmup 1/10 1280x720 bytes 3686400 nonzero 3673876
grimrs-capture: warmup 2/10 1280x720 bytes 3686400 nonzero 3673876
grimrs-capture: warmup 3/10 1280x720 bytes 3686400 nonzero 3673876
grimrs-capture: warmup 4/10 1280x720 bytes 3686400 nonzero 3673876
grimrs-capture: warmup 5/10 1280x720 bytes 3686400 nonzero 3673876
grimrs-capture: warmup 6/10 1280x720 bytes 3686400 nonzero 3673876
grimrs-capture: warmup 7/10 1280x720 bytes 3686400 nonzero 3673876
grimrs-capture: warmup 8/10 1280x720 bytes 3686400 nonzero 3673876
grimrs-capture: warmup 9/10 1280x720 bytes 3686400 nonzero 3673876
grimrs-capture: warmup 10/10 1280x720 bytes 3686400 nonzero 3673876
@@ -0,0 +1 @@
{"route":"grim-rs","language":"rust","protocol":"ext-image-copy-capture-v1 (auto)","output":"all","geometry":[1280,720],"mode":"raw","warmup":10,"frames":200,"frame_ms":{"min":30.553,"p50":69.147,"mean":66.826,"p95":85.813,"max":110.168},"fps_sustained":14.964,"wall_ms":13365.37,"cpu_ms_per_frame":1.754,"bytes_p50":null,"persistent":true,"rss_peak_mb":745.64,"rss_peak_kb_raw":763536,"system_cpu_ms_per_frame":133.5,"compositor_cpu_ms_per_frame":46.05,"machine_cpu_pct":199.8,"frames_bytes_allocated_p50":3686400,"notes":"grim-rs 0.2.0 via Grim::new() (auto-detect); copy target is a wl_shm buffer created inside each capture_* call, not a dma-buf the producer imports. Cursor not overlaid (the frozen argv has no cursor flag; hhstream-live's grim command passes -c). JPEG is Grim::to_jpeg, quality 80, the crate default and the live-stream setting. Raw timer stops when capture_* returns, which is after the crate copies the mapping into a Vec and swizzles XRGB/ARGB to RGBA (and applies any output transform or y-invert). The public API does not expose the mapped buffer before that, so raw includes work the contract says to leave outside the timer. Percentiles are linear interpolation on the sorted samples. cpu_ms_per_frame is getrusage(RUSAGE_SELF) utime+stime over the measured loop only, divided by frames. Frame buffers are freed between frames; that free is inside wall_ms and outside frame_ms. Outputs: Virtual-1@0,0 1280x720 scale 1. No GPU on this machine; dma-buf was not used and is not claimed. This process does not move the pointer and does not start or stop a session; the pixels are whatever the compositor is already showing. Last warmup frame nonzero bytes: 3673876."}
@@ -0,0 +1,12 @@
grimrs-capture: route grim-rs protocol ext-image-copy-capture-v1 (auto) mode raw target all warmup 10 frames 200
grimrs-capture: outputs Virtual-1@0,0 1280x720 scale 1
grimrs-capture: warmup 1/10 1280x720 bytes 3686400 nonzero 3673876
grimrs-capture: warmup 2/10 1280x720 bytes 3686400 nonzero 3673876
grimrs-capture: warmup 3/10 1280x720 bytes 3686400 nonzero 3673876
grimrs-capture: warmup 4/10 1280x720 bytes 3686400 nonzero 3673876
grimrs-capture: warmup 5/10 1280x720 bytes 3686400 nonzero 3673876
grimrs-capture: warmup 6/10 1280x720 bytes 3686400 nonzero 3673876
grimrs-capture: warmup 7/10 1280x720 bytes 3686400 nonzero 3673876
grimrs-capture: warmup 8/10 1280x720 bytes 3686400 nonzero 3673876
grimrs-capture: warmup 9/10 1280x720 bytes 3686400 nonzero 3673876
grimrs-capture: warmup 10/10 1280x720 bytes 3686400 nonzero 3673876
@@ -0,0 +1 @@
{"route":"grim-rs","language":"rust","protocol":"ext-image-copy-capture-v1 (auto)","output":"all","geometry":[1280,720],"mode":"raw","warmup":10,"frames":200,"frame_ms":{"min":42.327,"p50":74.965,"mean":71.806,"p95":82.012,"max":98.162},"fps_sustained":13.927,"wall_ms":14361.366,"cpu_ms_per_frame":1.721,"bytes_p50":null,"persistent":true,"rss_peak_mb":745.63,"rss_peak_kb_raw":763520,"system_cpu_ms_per_frame":143.5,"compositor_cpu_ms_per_frame":46.9,"machine_cpu_pct":199.8,"frames_bytes_allocated_p50":3686400,"notes":"grim-rs 0.2.0 via Grim::new() (auto-detect); copy target is a wl_shm buffer created inside each capture_* call, not a dma-buf the producer imports. Cursor not overlaid (the frozen argv has no cursor flag; hhstream-live's grim command passes -c). JPEG is Grim::to_jpeg, quality 80, the crate default and the live-stream setting. Raw timer stops when capture_* returns, which is after the crate copies the mapping into a Vec and swizzles XRGB/ARGB to RGBA (and applies any output transform or y-invert). The public API does not expose the mapped buffer before that, so raw includes work the contract says to leave outside the timer. Percentiles are linear interpolation on the sorted samples. cpu_ms_per_frame is getrusage(RUSAGE_SELF) utime+stime over the measured loop only, divided by frames. Frame buffers are freed between frames; that free is inside wall_ms and outside frame_ms. Outputs: Virtual-1@0,0 1280x720 scale 1. No GPU on this machine; dma-buf was not used and is not claimed. This process does not move the pointer and does not start or stop a session; the pixels are whatever the compositor is already showing. Last warmup frame nonzero bytes: 3673876."}
@@ -0,0 +1 @@
grimrs-capture: output "hh-chromium-nxopdz" not found. Outputs: Virtual-1@0,0 1280x720 scale 1
Whitespace-only changes.
@@ -0,0 +1 @@
grimrs-capture: output "hh-chromium-nxopdz" not found. Outputs: Virtual-1@0,0 1280x720 scale 1
Whitespace-only changes.
@@ -0,0 +1 @@
grimrs-capture: output "hh-chromium-nxopdz" not found. Outputs: Virtual-1@0,0 1280x720 scale 1
Whitespace-only changes.
@@ -0,0 +1 @@
grimrs-capture: output "hh-chromium-nxopdz" not found. Outputs: Virtual-1@0,0 1280x720 scale 1
Whitespace-only changes.
Whitespace-only changes.
@@ -0,0 +1 @@
{"route": "pywayland-in-process", "language": "python", "protocol": "ext-image-copy-capture-v1", "output": "all", "geometry": [1280, 720], "mode": "jpeg", "warmup": 10, "frames": 200, "frame_ms": {"min": 27.78, "p50": 42.962, "mean": 45.199, "p95": 68.209, "max": 107.887}, "stage_ms": {"capture": 41.147, "encode": 4.036, "stitch": 0.0}, "fps_sustained": 22.12, "wall_ms": 9050.3, "bytes_p50": 218019, "persistent": true, "notes": "one connection, buffers reused, turbojpeg q80 in-process; 1 output(s); forcing a repaint per frame", "cpu_ms_per_frame": 2.042, "rss_peak_mb": 29.34, "rss_peak_kb_raw": 30048, "system_cpu_ms_per_frame": 90.2, "compositor_cpu_ms_per_frame": 35.7, "machine_cpu_pct": 199.3, "machine_cpu_pct_of_all_cores": 99.7, "frames_bytes_allocated_p50": 3686400, "resource_notes": ""}
@@ -0,0 +1,700 @@
[
{
"route": "grim-subprocess",
"description": "grim subprocess per frame (shipped)",
"cell": "all-raw",
"run": 1,
"rc": 0,
"language": "python",
"protocol": "zwlr-screencopy-v1 (grim's client)",
"output": "all",
"geometry": [
1280,
720
],
"mode": "raw",
"warmup": 10,
"frames": 200,
"frame_ms": {
"min": 36.553,
"p50": 56.179,
"mean": 60.257,
"p95": 86.708,
"max": 129.015
},
"fps_sustained": 16.6,
"wall_ms": 12454.97,
"bytes_p50": 2764816,
"persistent": false,
"notes": "one grim process per frame, fork and encode inside the timed span; territory: --all (every output stitched); ppm, unencoded",
"cpu_ms_per_frame": 4.212,
"rss_peak_mb": 20.54,
"rss_peak_kb_raw": 21036,
"system_cpu_ms_per_frame": 124.45,
"compositor_cpu_ms_per_frame": 52.75,
"machine_cpu_pct": 199.8,
"machine_cpu_pct_of_all_cores": 99.9,
"frames_bytes_allocated_p50": 3686400,
"resource_notes": ""
},
{
"route": "grim-subprocess",
"description": "grim subprocess per frame (shipped)",
"cell": "all-raw",
"run": 2,
"rc": 0,
"language": "python",
"protocol": "zwlr-screencopy-v1 (grim's client)",
"output": "all",
"geometry": [
1280,
720
],
"mode": "raw",
"warmup": 10,
"frames": 200,
"frame_ms": {
"min": 31.631,
"p50": 52.197,
"mean": 53.412,
"p95": 66.973,
"max": 106.298
},
"fps_sustained": 18.72,
"wall_ms": 11219.4,
"bytes_p50": 2764816,
"persistent": false,
"notes": "one grim process per frame, fork and encode inside the timed span; territory: --all (every output stitched); ppm, unencoded",
"cpu_ms_per_frame": 4.265,
"rss_peak_mb": 20.43,
"rss_peak_kb_raw": 20916,
"system_cpu_ms_per_frame": 112.1,
"compositor_cpu_ms_per_frame": 45.4,
"machine_cpu_pct": 199.8,
"machine_cpu_pct_of_all_cores": 99.9,
"frames_bytes_allocated_p50": 3686400,
"resource_notes": ""
},
{
"route": "grim-subprocess",
"description": "grim subprocess per frame (shipped)",
"cell": "all-jpeg",
"run": 1,
"rc": 0,
"language": "python",
"protocol": "zwlr-screencopy-v1 (grim's client)",
"output": "all",
"geometry": [
1280,
720
],
"mode": "jpeg",
"warmup": 10,
"frames": 200,
"frame_ms": {
"min": 37.184,
"p50": 63.692,
"mean": 64.745,
"p95": 84.754,
"max": 97.661
},
"fps_sustained": 15.45,
"wall_ms": 13722.74,
"bytes_p50": 258815,
"persistent": false,
"notes": "one grim process per frame, fork and encode inside the timed span; territory: --all (every output stitched); jpeg q80",
"cpu_ms_per_frame": 6.062,
"rss_peak_mb": 19.72,
"rss_peak_kb_raw": 20192,
"system_cpu_ms_per_frame": 137.15,
"compositor_cpu_ms_per_frame": 47.35,
"machine_cpu_pct": 199.9,
"machine_cpu_pct_of_all_cores": 99.9,
"frames_bytes_allocated_p50": null,
"resource_notes": ""
},
{
"route": "grim-subprocess",
"description": "grim subprocess per frame (shipped)",
"cell": "all-jpeg",
"run": 2,
"rc": 0,
"language": "python",
"protocol": "zwlr-screencopy-v1 (grim's client)",
"output": "all",
"geometry": [
1280,
720
],
"mode": "jpeg",
"warmup": 10,
"frames": 200,
"frame_ms": {
"min": 35.371,
"p50": 56.993,
"mean": 61.547,
"p95": 85.241,
"max": 109.849
},
"fps_sustained": 16.25,
"wall_ms": 13006.45,
"bytes_p50": 258815,
"persistent": false,
"notes": "one grim process per frame, fork and encode inside the timed span; territory: --all (every output stitched); jpeg q80",
"cpu_ms_per_frame": 6.015,
"rss_peak_mb": 19.74,
"rss_peak_kb_raw": 20216,
"system_cpu_ms_per_frame": 129.95,
"compositor_cpu_ms_per_frame": 46.35,
"machine_cpu_pct": 199.8,
"machine_cpu_pct_of_all_cores": 99.9,
"frames_bytes_allocated_p50": null,
"resource_notes": ""
},
{
"route": "pywayland-in-process",
"description": "pywayland in-process, zwlr-screencopy-v1",
"cell": "all-raw",
"run": 1,
"rc": 0,
"language": "python",
"protocol": "zwlr-screencopy-v1",
"output": "all",
"geometry": [
1280,
720
],
"mode": "raw",
"warmup": 10,
"frames": 200,
"frame_ms": {
"min": 27.641,
"p50": 46.404,
"mean": 44.688,
"p95": 58.353,
"max": 72.148
},
"stage_ms": {
"capture": 44.681,
"encode": 0.0,
"stitch": 0.0
},
"fps_sustained": 22.38,
"wall_ms": 8937.76,
"bytes_p50": null,
"persistent": true,
"notes": "one connection, buffers reused, turbojpeg q80 in-process; 1 output(s); forcing a repaint per frame",
"cpu_ms_per_frame": 0.249,
"rss_peak_mb": 24.58,
"rss_peak_kb_raw": 25172,
"system_cpu_ms_per_frame": 89.35,
"compositor_cpu_ms_per_frame": 30.9,
"machine_cpu_pct": 199.9,
"machine_cpu_pct_of_all_cores": 100.0,
"frames_bytes_allocated_p50": 3686400,
"resource_notes": ""
},
{
"route": "pywayland-in-process",
"description": "pywayland in-process, zwlr-screencopy-v1",
"cell": "all-raw",
"run": 2,
"rc": 0,
"language": "python",
"protocol": "zwlr-screencopy-v1",
"output": "all",
"geometry": [
1280,
720
],
"mode": "raw",
"warmup": 10,
"frames": 200,
"frame_ms": {
"min": 27.798,
"p50": 48.248,
"mean": 48.139,
"p95": 66.304,
"max": 84.225
},
"stage_ms": {
"capture": 48.132,
"encode": 0.0,
"stitch": 0.0
},
"fps_sustained": 20.77,
"wall_ms": 9627.89,
"bytes_p50": null,
"persistent": true,
"notes": "one connection, buffers reused, turbojpeg q80 in-process; 1 output(s); forcing a repaint per frame",
"cpu_ms_per_frame": 0.278,
"rss_peak_mb": 24.4,
"rss_peak_kb_raw": 24988,
"system_cpu_ms_per_frame": 96.1,
"compositor_cpu_ms_per_frame": 34.45,
"machine_cpu_pct": 199.6,
"machine_cpu_pct_of_all_cores": 99.8,
"frames_bytes_allocated_p50": 3686400,
"resource_notes": ""
},
{
"route": "pywayland-in-process",
"description": "pywayland in-process, zwlr-screencopy-v1",
"cell": "all-jpeg",
"run": 1,
"rc": 0,
"language": "python",
"protocol": "zwlr-screencopy-v1",
"output": "all",
"geometry": [
1280,
720
],
"mode": "jpeg",
"warmup": 10,
"frames": 200,
"frame_ms": {
"min": 28.34,
"p50": 51.442,
"mean": 53.317,
"p95": 68.451,
"max": 71.774
},
"stage_ms": {
"capture": 49.628,
"encode": 3.674,
"stitch": 0.0
},
"fps_sustained": 18.76,
"wall_ms": 10665.15,
"bytes_p50": 218019,
"persistent": true,
"notes": "one connection, buffers reused, turbojpeg q80 in-process; 1 output(s); forcing a repaint per frame",
"cpu_ms_per_frame": 2.033,
"rss_peak_mb": 29.53,
"rss_peak_kb_raw": 30236,
"system_cpu_ms_per_frame": 106.65,
"compositor_cpu_ms_per_frame": 35.65,
"machine_cpu_pct": 200.0,
"machine_cpu_pct_of_all_cores": 100.0,
"frames_bytes_allocated_p50": 3686400,
"resource_notes": ""
},
{
"route": "pywayland-in-process",
"description": "pywayland in-process, zwlr-screencopy-v1",
"cell": "all-jpeg",
"run": 2,
"rc": 0,
"language": "python",
"protocol": "zwlr-screencopy-v1",
"output": "all",
"geometry": [
1280,
720
],
"mode": "jpeg",
"warmup": 10,
"frames": 200,
"frame_ms": {
"min": 25.792,
"p50": 50.701,
"mean": 52.092,
"p95": 66.879,
"max": 80.066
},
"stage_ms": {
"capture": 47.785,
"encode": 4.258,
"stitch": 0.0
},
"fps_sustained": 19.2,
"wall_ms": 10420.83,
"bytes_p50": 218019,
"persistent": true,
"notes": "one connection, buffers reused, turbojpeg q80 in-process; 1 output(s); forcing a repaint per frame",
"cpu_ms_per_frame": 2.038,
"rss_peak_mb": 29.45,
"rss_peak_kb_raw": 30152,
"system_cpu_ms_per_frame": 104.25,
"compositor_cpu_ms_per_frame": 34.1,
"machine_cpu_pct": 200.1,
"machine_cpu_pct_of_all_cores": 100.0,
"frames_bytes_allocated_p50": 3686400,
"resource_notes": ""
},
{
"route": "pywayland-in-process",
"description": "pywayland in-process, ext-image-copy-capture-v1",
"cell": "all-raw",
"run": 1,
"rc": 0,
"language": "python",
"protocol": "ext-image-copy-capture-v1",
"output": "all",
"geometry": [
1280,
720
],
"mode": "raw",
"warmup": 10,
"frames": 200,
"frame_ms": {
"min": 28.674,
"p50": 50.052,
"mean": 50.651,
"p95": 66.015,
"max": 80.998
},
"stage_ms": {
"capture": 50.644,
"encode": 0.0,
"stitch": 0.0
},
"fps_sustained": 19.74,
"wall_ms": 10130.39,
"bytes_p50": null,
"persistent": true,
"notes": "one connection, buffers reused, turbojpeg q80 in-process; 1 output(s); forcing a repaint per frame",
"cpu_ms_per_frame": 0.278,
"rss_peak_mb": 24.53,
"rss_peak_kb_raw": 25116,
"system_cpu_ms_per_frame": 101.3,
"compositor_cpu_ms_per_frame": 35.35,
"machine_cpu_pct": 200.0,
"machine_cpu_pct_of_all_cores": 100.0,
"frames_bytes_allocated_p50": 3686400,
"resource_notes": ""
},
{
"route": "pywayland-in-process",
"description": "pywayland in-process, ext-image-copy-capture-v1",
"cell": "all-raw",
"run": 2,
"rc": 0,
"language": "python",
"protocol": "ext-image-copy-capture-v1",
"output": "all",
"geometry": [
1280,
720
],
"mode": "raw",
"warmup": 10,
"frames": 200,
"frame_ms": {
"min": 28.289,
"p50": 40.034,
"mean": 41.43,
"p95": 58.108,
"max": 76.787
},
"stage_ms": {
"capture": 41.423,
"encode": 0.0,
"stitch": 0.0
},
"fps_sustained": 24.14,
"wall_ms": 8286.13,
"bytes_p50": null,
"persistent": true,
"notes": "one connection, buffers reused, turbojpeg q80 in-process; 1 output(s); forcing a repaint per frame",
"cpu_ms_per_frame": 0.27,
"rss_peak_mb": 24.47,
"rss_peak_kb_raw": 25060,
"system_cpu_ms_per_frame": 82.6,
"compositor_cpu_ms_per_frame": 33.45,
"machine_cpu_pct": 199.4,
"machine_cpu_pct_of_all_cores": 99.7,
"frames_bytes_allocated_p50": 3686400,
"resource_notes": ""
},
{
"route": "pywayland-in-process",
"description": "pywayland in-process, ext-image-copy-capture-v1",
"cell": "all-jpeg",
"run": 1,
"rc": 0,
"language": "python",
"protocol": "ext-image-copy-capture-v1",
"output": "all",
"geometry": [
1280,
720
],
"mode": "jpeg",
"warmup": 10,
"frames": 200,
"frame_ms": {
"min": 26.134,
"p50": 53.679,
"mean": 53.121,
"p95": 71.219,
"max": 87.846
},
"stage_ms": {
"capture": 49.302,
"encode": 3.803,
"stitch": 0.0
},
"fps_sustained": 18.82,
"wall_ms": 10624.64,
"bytes_p50": 218019,
"persistent": true,
"notes": "one connection, buffers reused, turbojpeg q80 in-process; 1 output(s); forcing a repaint per frame",
"cpu_ms_per_frame": 2.041,
"rss_peak_mb": 29.34,
"rss_peak_kb_raw": 30048,
"system_cpu_ms_per_frame": 106.15,
"compositor_cpu_ms_per_frame": 38.7,
"machine_cpu_pct": 199.8,
"machine_cpu_pct_of_all_cores": 99.9,
"frames_bytes_allocated_p50": 3686400,
"resource_notes": ""
},
{
"route": "pywayland-in-process",
"description": "pywayland in-process, ext-image-copy-capture-v1",
"cell": "all-jpeg",
"run": 2,
"command": "/nix/store/i7d46siafmlrgcmj9nzvwj7y64p34yaz-python3-3.14.7-env/bin/python3 /var/lib/hermes/worktrees/hhs-capture-speed/spikes/capture-speed/producers/wayland_producer.py --all --protocol ext --frames 200 --warmup 10 --mode jpeg --json",
"rc": 0,
"elapsed_s": 9.9,
"language": "python",
"protocol": "ext-image-copy-capture-v1",
"output": "all",
"geometry": [
1280,
720
],
"mode": "jpeg",
"warmup": 10,
"frames": 200,
"frame_ms": {
"min": 27.78,
"p50": 42.962,
"mean": 45.199,
"p95": 68.209,
"max": 107.887
},
"stage_ms": {
"capture": 41.147,
"encode": 4.036,
"stitch": 0.0
},
"fps_sustained": 22.12,
"wall_ms": 9050.3,
"bytes_p50": 218019,
"persistent": true,
"notes": "one connection, buffers reused, turbojpeg q80 in-process; 1 output(s); forcing a repaint per frame",
"cpu_ms_per_frame": 2.042,
"rss_peak_mb": 29.34,
"rss_peak_kb_raw": 30048,
"system_cpu_ms_per_frame": 90.2,
"compositor_cpu_ms_per_frame": 35.7,
"machine_cpu_pct": 199.3,
"machine_cpu_pct_of_all_cores": 99.7,
"frames_bytes_allocated_p50": 3686400,
"resource_notes": ""
},
{
"route": "grim-rs",
"description": "grim-rs crate (Rust), long-lived process",
"cell": "output-raw",
"run": 1,
"command": "/var/lib/hermes/worktrees/hhs-capture-speed/spikes/capture-speed/001a-grim-rs/target/release/grimrs-capture --output hh-chromium-nxopdz --frames 200 --warmup 10 --mode raw --json",
"rc": 1,
"elapsed_s": 0.0,
"error": [
"grimrs-capture: output \"hh-chromium-nxopdz\" not found. Outputs: Virtual-1@0,0 1280x720 scale 1"
]
},
{
"route": "grim-rs",
"description": "grim-rs crate (Rust), long-lived process",
"cell": "output-raw",
"run": 2,
"command": "/var/lib/hermes/worktrees/hhs-capture-speed/spikes/capture-speed/001a-grim-rs/target/release/grimrs-capture --output hh-chromium-nxopdz --frames 200 --warmup 10 --mode raw --json",
"rc": 1,
"elapsed_s": 0.0,
"error": [
"grimrs-capture: output \"hh-chromium-nxopdz\" not found. Outputs: Virtual-1@0,0 1280x720 scale 1"
]
},
{
"route": "grim-rs",
"description": "grim-rs crate (Rust), long-lived process",
"cell": "output-jpeg",
"run": 1,
"command": "/var/lib/hermes/worktrees/hhs-capture-speed/spikes/capture-speed/001a-grim-rs/target/release/grimrs-capture --output hh-chromium-nxopdz --frames 200 --warmup 10 --mode jpeg --json",
"rc": 1,
"elapsed_s": 0.0,
"error": [
"grimrs-capture: output \"hh-chromium-nxopdz\" not found. Outputs: Virtual-1@0,0 1280x720 scale 1"
]
},
{
"route": "grim-rs",
"description": "grim-rs crate (Rust), long-lived process",
"cell": "output-jpeg",
"run": 2,
"command": "/var/lib/hermes/worktrees/hhs-capture-speed/spikes/capture-speed/001a-grim-rs/target/release/grimrs-capture --output hh-chromium-nxopdz --frames 200 --warmup 10 --mode jpeg --json",
"rc": 1,
"elapsed_s": 0.0,
"error": [
"grimrs-capture: output \"hh-chromium-nxopdz\" not found. Outputs: Virtual-1@0,0 1280x720 scale 1"
]
},
{
"route": "grim-rs",
"description": "grim-rs crate (Rust), long-lived process",
"cell": "all-raw",
"run": 1,
"command": "/var/lib/hermes/worktrees/hhs-capture-speed/spikes/capture-speed/001a-grim-rs/target/release/grimrs-capture --all --frames 200 --warmup 10 --mode raw --json",
"rc": 0,
"elapsed_s": 14.0,
"language": "rust",
"protocol": "ext-image-copy-capture-v1 (auto)",
"output": "all",
"geometry": [
1280,
720
],
"mode": "raw",
"warmup": 10,
"frames": 200,
"frame_ms": {
"min": 30.553,
"p50": 69.147,
"mean": 66.826,
"p95": 85.813,
"max": 110.168
},
"fps_sustained": 14.964,
"wall_ms": 13365.37,
"cpu_ms_per_frame": 1.754,
"bytes_p50": null,
"persistent": true,
"rss_peak_mb": 745.64,
"rss_peak_kb_raw": 763536,
"system_cpu_ms_per_frame": 133.5,
"compositor_cpu_ms_per_frame": 46.05,
"machine_cpu_pct": 199.8,
"frames_bytes_allocated_p50": 3686400,
"notes": "grim-rs 0.2.0 via Grim::new() (auto-detect); copy target is a wl_shm buffer created inside each capture_* call, not a dma-buf the producer imports. Cursor not overlaid (the frozen argv has no cursor flag; hhstream-live's grim command passes -c). JPEG is Grim::to_jpeg, quality 80, the crate default and the live-stream setting. Raw timer stops when capture_* returns, which is after the crate copies the mapping into a Vec and swizzles XRGB/ARGB to RGBA (and applies any output transform or y-invert). The public API does not expose the mapped buffer before that, so raw includes work the contract says to leave outside the timer. Percentiles are linear interpolation on the sorted samples. cpu_ms_per_frame is getrusage(RUSAGE_SELF) utime+stime over the measured loop only, divided by frames. Frame buffers are freed between frames; that free is inside wall_ms and outside frame_ms. Outputs: Virtual-1@0,0 1280x720 scale 1. No GPU on this machine; dma-buf was not used and is not claimed. This process does not move the pointer and does not start or stop a session; the pixels are whatever the compositor is already showing. Last warmup frame nonzero bytes: 3673876."
},
{
"route": "grim-rs",
"description": "grim-rs crate (Rust), long-lived process",
"cell": "all-raw",
"run": 2,
"command": "/var/lib/hermes/worktrees/hhs-capture-speed/spikes/capture-speed/001a-grim-rs/target/release/grimrs-capture --all --frames 200 --warmup 10 --mode raw --json",
"rc": 0,
"elapsed_s": 15.4,
"language": "rust",
"protocol": "ext-image-copy-capture-v1 (auto)",
"output": "all",
"geometry": [
1280,
720
],
"mode": "raw",
"warmup": 10,
"frames": 200,
"frame_ms": {
"min": 42.327,
"p50": 74.965,
"mean": 71.806,
"p95": 82.012,
"max": 98.162
},
"fps_sustained": 13.927,
"wall_ms": 14361.366,
"cpu_ms_per_frame": 1.721,
"bytes_p50": null,
"persistent": true,
"rss_peak_mb": 745.63,
"rss_peak_kb_raw": 763520,
"system_cpu_ms_per_frame": 143.5,
"compositor_cpu_ms_per_frame": 46.9,
"machine_cpu_pct": 199.8,
"frames_bytes_allocated_p50": 3686400,
"notes": "grim-rs 0.2.0 via Grim::new() (auto-detect); copy target is a wl_shm buffer created inside each capture_* call, not a dma-buf the producer imports. Cursor not overlaid (the frozen argv has no cursor flag; hhstream-live's grim command passes -c). JPEG is Grim::to_jpeg, quality 80, the crate default and the live-stream setting. Raw timer stops when capture_* returns, which is after the crate copies the mapping into a Vec and swizzles XRGB/ARGB to RGBA (and applies any output transform or y-invert). The public API does not expose the mapped buffer before that, so raw includes work the contract says to leave outside the timer. Percentiles are linear interpolation on the sorted samples. cpu_ms_per_frame is getrusage(RUSAGE_SELF) utime+stime over the measured loop only, divided by frames. Frame buffers are freed between frames; that free is inside wall_ms and outside frame_ms. Outputs: Virtual-1@0,0 1280x720 scale 1. No GPU on this machine; dma-buf was not used and is not claimed. This process does not move the pointer and does not start or stop a session; the pixels are whatever the compositor is already showing. Last warmup frame nonzero bytes: 3673876."
},
{
"route": "grim-rs",
"description": "grim-rs crate (Rust), long-lived process",
"cell": "all-jpeg",
"run": 1,
"command": "/var/lib/hermes/worktrees/hhs-capture-speed/spikes/capture-speed/001a-grim-rs/target/release/grimrs-capture --all --frames 200 --warmup 10 --mode jpeg --json",
"rc": 0,
"elapsed_s": 16.4,
"language": "rust",
"protocol": "ext-image-copy-capture-v1 (auto)",
"output": "all",
"geometry": [
1280,
720
],
"mode": "jpeg",
"warmup": 10,
"frames": 200,
"frame_ms": {
"min": 43.383,
"p50": 78.211,
"mean": 76.591,
"p95": 91.117,
"max": 120.327
},
"fps_sustained": 13.056,
"wall_ms": 15318.39,
"cpu_ms_per_frame": 7.657,
"bytes_p50": 218563.0,
"persistent": true,
"rss_peak_mb": 745.62,
"rss_peak_kb_raw": 763516,
"system_cpu_ms_per_frame": 153.05,
"compositor_cpu_ms_per_frame": 48.15,
"machine_cpu_pct": 199.8,
"frames_bytes_allocated_p50": 3686400,
"notes": "grim-rs 0.2.0 via Grim::new() (auto-detect); copy target is a wl_shm buffer created inside each capture_* call, not a dma-buf the producer imports. Cursor not overlaid (the frozen argv has no cursor flag; hhstream-live's grim command passes -c). JPEG is Grim::to_jpeg, quality 80, the crate default and the live-stream setting. Raw timer stops when capture_* returns, which is after the crate copies the mapping into a Vec and swizzles XRGB/ARGB to RGBA (and applies any output transform or y-invert). The public API does not expose the mapped buffer before that, so raw includes work the contract says to leave outside the timer. Percentiles are linear interpolation on the sorted samples. cpu_ms_per_frame is getrusage(RUSAGE_SELF) utime+stime over the measured loop only, divided by frames. Frame buffers are freed between frames; that free is inside wall_ms and outside frame_ms. Outputs: Virtual-1@0,0 1280x720 scale 1. No GPU on this machine; dma-buf was not used and is not claimed. This process does not move the pointer and does not start or stop a session; the pixels are whatever the compositor is already showing. Last warmup frame nonzero bytes: 217104."
},
{
"route": "grim-rs",
"description": "grim-rs crate (Rust), long-lived process",
"cell": "all-jpeg",
"run": 2,
"command": "/var/lib/hermes/worktrees/hhs-capture-speed/spikes/capture-speed/001a-grim-rs/target/release/grimrs-capture --all --frames 200 --warmup 10 --mode jpeg --json",
"rc": 0,
"elapsed_s": 15.8,
"language": "rust",
"protocol": "ext-image-copy-capture-v1 (auto)",
"output": "all",
"geometry": [
1280,
720
],
"mode": "jpeg",
"warmup": 10,
"frames": 200,
"frame_ms": {
"min": 41.516,
"p50": 75.597,
"mean": 74.024,
"p95": 92.187,
"max": 126.722
},
"fps_sustained": 13.509,
"wall_ms": 14805.03,
"cpu_ms_per_frame": 7.682,
"bytes_p50": 218563.0,
"persistent": true,
"rss_peak_mb": 745.52,
"rss_peak_kb_raw": 763416,
"system_cpu_ms_per_frame": 147.85,
"compositor_cpu_ms_per_frame": 47.5,
"machine_cpu_pct": 199.7,
"frames_bytes_allocated_p50": 3686400,
"notes": "grim-rs 0.2.0 via Grim::new() (auto-detect); copy target is a wl_shm buffer created inside each capture_* call, not a dma-buf the producer imports. Cursor not overlaid (the frozen argv has no cursor flag; hhstream-live's grim command passes -c). JPEG is Grim::to_jpeg, quality 80, the crate default and the live-stream setting. Raw timer stops when capture_* returns, which is after the crate copies the mapping into a Vec and swizzles XRGB/ARGB to RGBA (and applies any output transform or y-invert). The public API does not expose the mapped buffer before that, so raw includes work the contract says to leave outside the timer. Percentiles are linear interpolation on the sorted samples. cpu_ms_per_frame is getrusage(RUSAGE_SELF) utime+stime over the measured loop only, divided by frames. Frame buffers are freed between frames; that free is inside wall_ms and outside frame_ms. Outputs: Virtual-1@0,0 1280x720 scale 1. No GPU on this machine; dma-buf was not used and is not claimed. This process does not move the pointer and does not start or stop a session; the pixels are whatever the compositor is already showing. Last warmup frame nonzero bytes: 217104."
}
]
+50
View File
@@ -0,0 +1,50 @@
#!/usr/bin/env bash
# The decision-relevant matrix: one session's container (1280x720) per pass, one
# session at a time.
#
# Why it is narrower than the first attempt: the `--all` cells changed meaning
# whenever a second session existed (2560x720 became 3840x720), and running two
# sessions plus four routes on a two-core software-rendered box twice crashed the
# compositor mid-run. The wire-relevant case for live streaming and agent eyes is
# one session's window, so that is what this measures -- and it runs the sessions
# sequentially to keep the load off the compositor.
set -u
cd "$(dirname "$0")"
start_session() {
local page="$1" profile="$2"
hhstream session start chromium --idle-timeout 3h --wait-ready 45s -- \
--ozone-platform=wayland --no-first-run --no-default-browser-check \
--user-data-dir="/var/lib/hermes/hhstream-sessions/$profile" \
--window-size=1280,720 "--app=$page" >/dev/null 2>&1
sleep 6
hhstream session list --json 2>/dev/null | python3 -c "
import json,sys
d=json.load(sys.stdin)
s=d['sessions'][-1]
print(s['container']['output'])"
}
echo "########## PASS A: moving screen ##########"
A=$(start_session "http://127.0.0.1/spin.html" chromium-bench-fin)
echo "target: $A"
if [ -n "$A" ]; then
nix-shell shell.nix --run "python3 bench.py --output $A --cells output-raw --cells output-jpeg \
--frames 200 --warmup 10 --runs 2 --control-seconds 10 --resume --results results/final-motion"
hhstream session kill "$(hhstream session list --json | python3 -c 'import json,sys;print(json.load(sys.stdin)[\"sessions\"][-1][\"id\"])')" >/dev/null 2>&1
else
echo "no session: compositor may be down"
fi
echo "########## PASS B: still screen ##########"
B=$(start_session "http://127.0.0.1/static.html" chromium-bench-still)
echo "target: $B"
if [ -n "$B" ]; then
nix-shell shell.nix --run "python3 bench.py --output $B --cells output-raw --cells output-jpeg \
--frames 200 --warmup 10 --runs 2 --control-seconds 10 --resume --results results/final-still"
fi
echo "########## compositor survived? ##########"
systemctl is-active hyprland-headless.service
journalctl --since "30 minutes ago" --no-pager 2>/dev/null | grep -c "dumped core" | sed 's/^/core dumps in the last 30 min: /'
echo "########## final matrix complete ##########"