Direct RGBA receive: fixed source encoder
Configuration and method
These tests use an instrumented research fixture. Noisedeck A sends 1920 × 1080 RGBA packets through Sync to Noisedeck B.
Native camera publication remains active. B receives direct RGBA through an HTTP BYOB reader.
SCALAR is the parent fixture. ENCODER adds a private fixed-descriptor encoder and retained packet views to the source sink.
The parent decoder and ACK scalar changes remain in both arms. Both arms use the same native executable. Receiver credit remains one.
Each source packet contains 8,294,400 pixel bytes and a 64-byte header. Both source limits are 16,588,928 bytes.
The encoder still copies each payload into the packet. It does not remove that full-frame copy or the source readback.
The private encoder checks sequence and timestamp values on each call. Descriptor changes rebuild its retained state.
The source sink retains its pressure checks, actual send-return accounting and synchronous buffer ownership.
The tests use GLSL and synth.testPattern. Timing uses pattern 0 with gridSize 4. Pixel controls use pattern 5 with gridSize 16.
The source, compiled pattern and uniforms passed their existing identity checks.
Retained host metadata identifies an Apple M2, eight logical CPUs, 24 GiB of memory and Darwin 23.6.0.
This metadata is not an in-run thermal or frequency measurement.
All recorded power samples show AC power. The samples do not prove continuous AC power between reads.
Measurement results
The report includes 16 valid runs and one rejected startup attempt. Each arm has web and desktop positive, poison-negative, five-second and three-minute runs.
The eight pixel controls passed. Each accepted observation refers to the exact controls for its arm and context.
The four long observations contain 720 complete seconds and four partial tails. The four short observations contain 20 complete seconds and their tails.
Input and final markers must match within the same render. Rates use each complete measurement interval, including its partial tail.
SCALAR web has no complete seconds below 50. ENCODER web has six, including one second with zero input and final pairs.
The ENCODER web cells starting at seconds 65 through 69 contain 14, 0, 19, 14 and 16 pairs.
The cell starting at second 149 contains 47 pairs. These cells remain in the totals and diagrams.
SCALAR desktop has five complete seconds below 50. ENCODER desktop has two. This does not establish sustained 60 frames/s.
Open full-size diagram
B clocks remain separate. Input: dashed orange. Final: solid blue. All 720 complete seconds use a 0–66 scale. Dotted lines mark 50 and 60. Partial tails remain in the table.Source pressure
The source checks its byte limit at admission, before encoding and after encoding.
A check rejects additional bytes above the limit. It also rejects observed queued bytes above the limit minus those additional bytes.
The remaining allowance is 8,294,464 bytes in both arms.
The recorder stores positive checks and actual drops. It does not store timed buffer values for passed checks.
SCALAR recorded 19 web drops and 128 desktop drops. ENCODER recorded 56 web drops and 101 desktop drops.
ENCODER web recorded 37 admission drops and 19 before encoding. Desktop recorded 64 and 37 in those phases.
No measured run recorded a drop after encoding. The tables retain all three phases.
A positive check and its actual drop are separate records for one decision. Drop counts include only actual drop records.
The inclusive byte thresholds describe observed positive checks. They do not predict delivery under another limit.
Each pressure diagram uses its own A clock. Each delivery diagram uses its own B clock.
A frame sequence can join identities across processes. It does not establish latency or simultaneous timing.
Open full-size diagram
A clocks remain separate. Points show positive checks only. Shapes identify phases. Dashed lines mark each remaining allowance. Dotted lines mark each source limit. There is no continuous buffer trace.Open full-size diagram
A clocks remain separate. Each bar includes actual drops in that second or partial tail. Colors separate phases. Positive-check records do not add another drop.Startup and diagnostic work
The raw data retains startup, warmup and cleanup events. These events do not enter the measured drop counts.
Startup generation history remains in the data. Measurement intervals use each source binding.
All accepted pressure journals have zero unread gaps. Ring overwrite counts remain explicit.
Each long observation has 21 pressure cuts. Each cut has a 250 ms deadline for capture, pressure drains and pressure-journal writes.
Cut durations include controller waiting and I/O. They are not CPU costs or measured effects on rendering.
Existing source checks and source-journal writes occur outside that pressure-cut deadline.
The recorder retains 393,312 bytes of numeric storage. Positive checks and drops add clock reads.
Constructor-count tests check the targeted encoder changes. They do not measure total allocation, retained heap or garbage collection.
Rejected attempt
SCALAR desktop-observe180-scalarv1-a1 failed before native startup or measurement. It has no accepted rate or pressure interval.
The harness requested debugger port 65535. Electron 42.3.0 selected port 49154 because its source excludes that upper boundary.
All owned browser processes exited. The unchanged desktop repeat passed and remains a separate result.
The startup failure is not a product performance measurement. ENCODER completed all eight gates on their first attempts.
Native timing and retained events
Native elapsed times cover setup, measurement and cleanup. They are not CPU times.
The SCALAR desktop ring retained 256 of 462 slow events. It overwrote 206 individual records.
ENCODER retained all 110 web slow events and all 198 desktop slow events.
Aggregate stage counts, totals and maxima remain available. They do not reconstruct the lost individual events.
Native clocks have no measured alignment to either page clock. These maxima cannot identify a plotted B gap.
Limitations
GPU contention on the busy test machine is a working hypothesis.
These tests did not measure GPU scheduling or contention.
This sequential comparison does not establish causation, repeatability or a performance ranking. It does not estimate statistical uncertainty.
Both arms use the same source limit. These observations do not establish an effect on end-to-end latency.
There is no uninstrumented comparison with the same scheduling conditions. The recorder does not measure continuous socket occupancy.
There are no in-run thermal or frequency measurements. These tests do not establish garbage collection as the cause of pressure.
The zero-pair cell does not identify a stop caller, a native stall or a GC event.
These measurements do not establish released-product performance, external camera delivery or hours-long operation.
The earlier reports remain unchanged.
Source references
The numeric data includes result, control, source-manifest, native-receipt and journal digests.
SHA256SUMS identifies every public file. The exact private inputs remain outside the report.
Numeric data · Text report · Checksums
Earlier decoder and ACK report