Jetson Video Benchmark
Purpose
Measure codec-stage FPS and megapixels/second for an evidenced target and exact
workload. Supported live routes are encode, decode, P4/P5 comparison, and a
strictly increasing worker-capacity sweep. A separate no-media path calculates
clearly labeled SDK-documentation estimates; it never claims a measurement.
Terminal gates
- For a request solely for PSNR or SSIM, state that objective quality is
outside this skill and needs a separately authorized workflow, then stop.
Return only that scope response. Do not append an alternative benchmark or
next step, name another tool, request media, probe, or install anything.
- Resolve codec direction before probing or calculating. Explicit encode or
decode wording wins. Otherwise, quality, preset, bitrate, rate control,
recording, or compressed-output wording implies encode; explicitly
compressed input, ingest, playback, or IP/RTSP input implies decode. Treat
generic camera-count wording such as "connect" or "handle", and a codec name
on a camera or device, as direction-neutral. Set neutral wording aside and
resolve direction on the remaining cues; a neutral cue never creates a
conflict. For an encode-led camera estimate, state that NVDEC applies only
when cameras already emit the named compressed codec, keep encode primary,
and ask the user to confirm direction after giving the planning scenarios.
If non-neutral cues conflict or are absent, present both interpretations,
state that a unique encode estimate also needs an exact preset and measured
validation needs representative input, then ask which direction applies and
stop at that gate.
Transcoding consumes separate decode and encode budgets: measure it as the
supported decode and encode routes and report each budget separately, never
as one combined FPS.
- An explicit live/run/real/actual measurement requires representative input
for every requested direction: raw frames with format, geometry, and frame
count for encode; a compressed elementary stream or container for decode.
Each input must be one exact target-local path or user-supplied HTTP(S) URL.
If any is absent, return only
input_required and ask for the missing
item(s), including a separate exact path or URL for every requested
direction, before reading workflow references or sibling skills, probing,
authentication, recipe work, or workspace creation. Never benchmark the
setup smoke fixture or substitute catalog media.
- A no-media planning/expected/indicative question follows
documented-performance-estimates.md.
It requires the exact documented row and, for a scaled estimate, clock
provenance; it sets
measurement_performed: false. Preserve that reference
byte-for-byte. Omitted preset, per-stream FPS, or stream mix does not block
planning: enumerate the reference's bounded documented candidates and
scenarios, disclose every assumption, and ask for the omitted values.
Live preflight
Read benchmark-workflow.md completely once,
then apply its preflight section. Preserve explicit
native, pynvc, and both; map delegated selection to auto. Obtain fresh
read-only readiness from jetson-video-setup, passing any exact interpreter
already supplied or established in this conversation. Setup otherwise checks
the conventional profile path. For auto, zero eligible surfaces blocks, one
runs, and two returns selection_required. Python encode and comparison need a
full-samples environment; decode performance may use pynvc-smoke.
Encode, compare, and encode-capacity also require a validated schema-2 recipe
from jetson-video-recipe. A missing dependency stops that branch with
dependency_required; an eligible peer may continue as partial.
Measure
- Follow that same reference for sample allowlists, builds, direct argument
lists, frame accounting, option checks, marker grammar, comparisons, and
worker sweeps.
- Show the dry-run plan, then use a fresh mode-0700 workspace. Authenticate
each launcher immediately before use and launch its literal argument list.
- Run one excluded whole-process warmup and at least three new measured
processes per variant and surface, each with a finite timeout and without
-loop.
- Accept the sample-reported FPS only when its positive marker and processed
frame count match. Apply every identity, workload, marker, and statistics
check in benchmark-output-contract.md.
Report
Retain each launch's unedited output and write the compact result described by
benchmark-output-contract.md. Report
every repetition, recomputed mean/minimum/maximum, MP/s only when dimensions
are sample-bound, partial branches, limitations, and retry reason. Label worker
sweeps as codec-stage capacity bounds and preset comparisons as throughput-only.
Limitations
- Results apply only to the evidenced target, release, clocks, sample, content,
frame range, and controls; they are not a portable product ceiling.
- Do not interpolate undocumented presets or scale across format, bit depth,
chroma, codec, rate control, or tuning. Resolution scaling is allowed only as
the estimate reference's explicitly labeled pixel-area heuristic.
- Keep documented estimates per engine; use an all-engine multiplier only
under the explicit aggregate-planning rule in the estimate reference.
- The PyNvVideoCodec 2.1 encode-performance helper caps each worker at 1,000
frames; follow the common-frame rule in the workflow.
- Preserve
input_required, selection_required, dependency_required,
blocked, partial, and failed; do not promote a peer's success.
1---2name: jetson-video-benchmark3license: "Apache-2.0"4description: >-5 Use when measuring Jetson Video Codec SDK or PyNvVideoCodec encode/decode6 throughput, comparing presets or surfaces, testing codec-worker capacity7 with authenticated samples and user media, or producing a clearly labeled8 documentation-derived planning estimate when representative media is absent.9 Also use for a video request limited to PSNR or SSIM, to apply the terminal10 scope response.11metadata:12 source: https://skillmd.com/skills/nvidia/jetson-video-benchmark13 author: "Vinit Bansal <vinitkumarb@nvidia.com>"14 tags: [jetson, video-codec-sdk, pynvvideocodec, benchmark, nvenc, nvdec]15 languages: [markdown]16 data-classification: public17---1819# Jetson Video Benchmark2021## Purpose2223Measure codec-stage FPS and megapixels/second for an evidenced target and exact24workload. Supported live routes are encode, decode, P4/P5 comparison, and a25strictly increasing worker-capacity sweep. A separate no-media path calculates26clearly labeled SDK-documentation estimates; it never claims a measurement.2728## Terminal gates29301. For a request solely for PSNR or SSIM, state that objective quality is31 outside this skill and needs a separately authorized workflow, then stop.32 Return only that scope response. Do not append an alternative benchmark or33 next step, name another tool, request media, probe, or install anything.342. Resolve codec direction before probing or calculating. Explicit encode or35 decode wording wins. Otherwise, quality, preset, bitrate, rate control,36 recording, or compressed-output wording implies encode; explicitly37 compressed input, ingest, playback, or IP/RTSP input implies decode. Treat38 generic camera-count wording such as "connect" or "handle", and a codec name39 on a camera or device, as direction-neutral. Set neutral wording aside and40 resolve direction on the remaining cues; a neutral cue never creates a41 conflict. For an encode-led camera estimate, state that NVDEC applies only42 when cameras already emit the named compressed codec, keep encode primary,43 and ask the user to confirm direction after giving the planning scenarios.44 If non-neutral cues conflict or are absent, present both interpretations,45 state that a unique encode estimate also needs an exact preset and measured46 validation needs representative input, then ask which direction applies and47 stop at that gate.48 Transcoding consumes separate decode and encode budgets: measure it as the49 supported decode and encode routes and report each budget separately, never50 as one combined FPS.513. An explicit live/run/real/actual measurement requires representative input52 for every requested direction: raw frames with format, geometry, and frame53 count for encode; a compressed elementary stream or container for decode.54 Each input must be one exact target-local path or user-supplied HTTP(S) URL.55 If any is absent, return only `input_required` and ask for the missing56 item(s), including a separate exact path or URL for every requested57 direction, before reading workflow references or sibling skills, probing,58 authentication, recipe work, or workspace creation. Never benchmark the59 setup smoke fixture or substitute catalog media.604. A no-media planning/expected/indicative question follows61 [documented-performance-estimates.md](references/documented-performance-estimates.md).62 It requires the exact documented row and, for a scaled estimate, clock63 provenance; it sets `measurement_performed: false`. Preserve that reference64 byte-for-byte. Omitted preset, per-stream FPS, or stream mix does not block65 planning: enumerate the reference's bounded documented candidates and66 scenarios, disclose every assumption, and ask for the omitted values.6768## Live preflight6970Read [benchmark-workflow.md](references/benchmark-workflow.md) completely once,71then apply its preflight section. Preserve explicit72`native`, `pynvc`, and `both`; map delegated selection to `auto`. Obtain fresh73read-only readiness from `jetson-video-setup`, passing any exact interpreter74already supplied or established in this conversation. Setup otherwise checks75the conventional profile path. For `auto`, zero eligible surfaces blocks, one76runs, and two returns `selection_required`. Python encode and comparison need a77`full-samples` environment; decode performance may use `pynvc-smoke`.7879Encode, compare, and encode-capacity also require a validated schema-2 recipe80from `jetson-video-recipe`. A missing dependency stops that branch with81`dependency_required`; an eligible peer may continue as `partial`.8283## Measure84851. Follow that same reference for sample allowlists, builds, direct argument86 lists, frame accounting, option checks, marker grammar, comparisons, and87 worker sweeps.882. Show the dry-run plan, then use a fresh mode-0700 workspace. Authenticate89 each launcher immediately before use and launch its literal argument list.903. Run one excluded whole-process warmup and at least three new measured91 processes per variant and surface, each with a finite timeout and without92 `-loop`.934. Accept the sample-reported FPS only when its positive marker and processed94 frame count match. Apply every identity, workload, marker, and statistics95 check in [benchmark-output-contract.md](references/benchmark-output-contract.md).9697## Report9899Retain each launch's unedited output and write the compact result described by100[benchmark-output-contract.md](references/benchmark-output-contract.md). Report101every repetition, recomputed mean/minimum/maximum, MP/s only when dimensions102are sample-bound, partial branches, limitations, and retry reason. Label worker103sweeps as codec-stage capacity bounds and preset comparisons as throughput-only.104105## Limitations106107- Results apply only to the evidenced target, release, clocks, sample, content,108 frame range, and controls; they are not a portable product ceiling.109- Do not interpolate undocumented presets or scale across format, bit depth,110 chroma, codec, rate control, or tuning. Resolution scaling is allowed only as111 the estimate reference's explicitly labeled pixel-area heuristic.112- Keep documented estimates per engine; use an all-engine multiplier only113 under the explicit aggregate-planning rule in the estimate reference.114- The PyNvVideoCodec 2.1 encode-performance helper caps each worker at 1,000115 frames; follow the common-frame rule in the workflow.116- Preserve `input_required`, `selection_required`, `dependency_required`,117 `blocked`, `partial`, and `failed`; do not promote a peer's success.