shdr1.0.0CPU + GPU compute

Recorded results · 15 October 2026

Measurements you can inspect

Twenty-five completed pipeline benchmarks, six full held-out digit evaluations, and every registered test outcome. Exact tables, original outputs and downloadable vector figures sit beside the scope of each result.

25resident pipeline workloads
6 × 10,000held-out digit evaluations
0failed registered tests
250bundled shader/profile checks

Resident pipeline performance

Configuration reports: Ryzen and llvmpipe, Core i7 and Iris Xe, and Raspberry Pi 5 and V3D.

These are mean completed workload times from CMake Release builds, C++11, seed 12345, scale 1 and twenty timed repetitions on each execution path. All graphics measurements select the ES 3.1 shader profile. Timing begins after compilation, initial uploads and warmup, and ends before final downloads. Transfers inside mixed CPU/GPU pipelines remain included.

Read the renderer label. llvmpipe executes graphics work in software on the Ryzen CPU and can use several cores. Iris Xe and V3D are hardware graphics. Ordinary C++ pipelines execute serially. The asynchronous ring uses two CPU workers.

The async row measures a completed batch of 32 dispatches through three reusable slots. A neural training row measures one mini-batch gradient/SGD update, while inference measures the complete forward pipeline. The seven-stage LeNet benchmark uses resident seeded random inputs and parameters. Its trained-model evaluation has a separate timing scope below.

All 25 benchmark workloads · mean completed milliseconds
Pipeline Workload Ryzen CPU ms llvmpipe ms Core i7 CPU ms Iris Xe ms Pi 5 CPU ms V3D ms
Buffer affine/clamp/sum 16384 floats 0.300 1.374 0.139 3.928 0.516 1.956
Buffer/image blur 128 × 128 0.589 1.639 0.422 0.054 1.033 0.471
Borrowed image blur 128 × 128 0.637 1.582 0.299 0.052 0.951 0.469
Hybrid affine/clamp/sum 16384 floats 0.288 1.217 0.145 2.825 0.597 2.188
Dynamic library stages 16384 floats 0.303 1.463 0.460 2.100 0.838 1.968
RAW10/debayer/luminance 128 × 128 1.078 1.643 1.080 0.068 2.466 2.522
2D convolution 5 x 5 128 × 128 vec2 1.521 0.748 1.140 0.059 3.283 2.444
3D convolution 3 x 3 x 3 24 × 24 × 24 1.503 0.706 1.266 0.057 3.156 3.020
Forward/inverse direct DFT 32 × 32 complex 1.232 1.643 0.892 0.080 2.589 0.579
Gaussian/Sobel/Harris/NMS 128 × 128 2.379 3.138 1.841 0.089 4.571 2.345
Lossless tiled RLE round trip 65536 uints 0.625 1.367 0.462 0.248 1.256 1.253
EVT3 event decoding 65536 sixteen-bit words 2.990 2.222 2.993 0.734 4.873 6.988
Async ring 32 × 257 floats, 3 slots 1.207 7.759 0.094 0.730 0.143 1.976
EVT3 + optical flow 32 × 24 sampled plane, history 64 0.393 1.731 0.290 0.215 0.640 0.656
ANN training 32 samples, 33 parameters 0.085 0.520 0.072 0.042 0.206 0.114
ANN inference 32 samples, 33 parameters 0.004 0.156 0.002 0.015 0.006 0.037
CNN training 32 samples, 23 parameters 1.615 0.644 1.359 0.086 3.254 0.490
CNN inference 32 samples, 23 parameters 0.064 0.161 0.054 0.034 0.138 0.057
Transformer training 8 sequences, 4 tokens × 4, 134 parameters 1.324 0.770 0.972 0.284 2.404 0.729
Transformer inference 8 sequences, 4 tokens × 4, 134 parameters 0.008 0.171 0.006 0.057 0.018 0.055
FFT forward/inverse 8 × 1024 complex 3.050 10.656 2.226 0.226 5.936 4.174
FIR low-pass 8 × 2048 floats, 65 taps 2.718 0.772 1.956 0.124 6.570 5.499
Biquad low-pass 8 × 2048 floats, persistent state 0.097 0.233 0.040 0.515 0.156 0.691
FFT band-pass 8 × 1024 complex, 4-10 kHz 3.219 11.268 2.250 0.222 6.153 5.001
LeNet classifier 32 × 28 × 28, 61,706 parameters 46.521 7.622 40.790 1.756 116.491 136.617

Ryzen 5900X / llvmpipe

Ryzen 5900X / llvmpipe mean pipeline timings and CPU-to-graphics ratios
Time and ratio axes are logarithmic. A ratio above one favors the graphics path. Means are displayed to three decimal places. Ratios come from the original unrounded means, then are rounded for reporting. Download SVG
Exact timing and ratio table · Ryzen 5900X / llvmpipe
Ryzen 5900X / llvmpipe · 20 measured repetitions
Pipeline Workload CPU ms Graphics ms CPU / graphics
Buffer affine/clamp/sum 16384 floats 0.300 1.374 0.218
Buffer/image blur 128 x 128 0.589 1.639 0.359
Borrowed image blur 128 x 128 0.637 1.582 0.403
Hybrid affine/clamp/sum 16384 floats 0.288 1.217 0.237
Dynamic library stages 16384 floats 0.303 1.463 0.207
RAW10/debayer/luminance 128 x 128 1.078 1.643 0.656
2D convolution 5 x 5 128 x 128 vec2 1.521 0.748 2.034
3D convolution 3 x 3 x 3 24 x 24 x 24 1.503 0.706 2.129
Forward/inverse direct DFT 32 x 32 complex 1.232 1.643 0.750
Gaussian/Sobel/Harris/NMS 128 x 128 2.379 3.138 0.758
Lossless tiled RLE round trip 65536 uints 0.625 1.367 0.457
EVT3 event decoding 65536 sixteen-bit words 2.990 2.222 1.346
Async ring 32 x 257 floats, 3 slots 1.207 7.759 0.156
EVT3 + optical flow 32 x 24 sampled plane, history 64 0.393 1.731 0.227
ANN training 32 samples, 33 parameters 0.085 0.520 0.164
ANN inference 32 samples, 33 parameters 0.004 0.156 0.024
CNN training 32 samples, 23 parameters 1.615 0.644 2.510
CNN inference 32 samples, 23 parameters 0.064 0.161 0.397
Transformer training 8 sequences, 4 tokens x 4, 134 parameters 1.324 0.770 1.719
Transformer inference 8 sequences, 4 tokens x 4, 134 parameters 0.008 0.171 0.045
FFT forward/inverse 8 x 1024 complex 3.050 10.656 0.286
FIR low-pass 8 x 2048 floats, 65 taps 2.718 0.772 3.523
Biquad low-pass 8 x 2048 floats, persistent state 0.097 0.233 0.417
FFT band-pass 8 x 1024 complex, 4-10 kHz 3.219 11.268 0.286
LeNet classifier 32 x 28 x 28, 61,706 parameters 46.521 7.622 6.104
Complete benchmark output · Ryzen 5900X / llvmpipe

This is the recorded benchmark output. Its EVT3 wording “packed words” counts 16-bit sensor words before two words are packed into each uint.

Seed: 12345, iterations: 20, scale: 1.
GPU renderer: llvmpipe (LLVM 20.1.2, 256 bits).
Lossless RLE: 13556 payload words + 257 offset words, 4.74x size reduction.
EVT3 compact output: 223772 events from 65536 words, five stages include state reset.
Optical flow input: 610 seeded samples on a known temporal plane, six linked stages.
Neural training timings measure one mini-batch gradient/SGD update, inference measures the complete forward pipeline.
Steady-state resident timings, compilation, first uploads and final readback excluded.
Hybrid timings include transfers between CPU and GPU stages.

| Pipeline | Workload | CPU ms | GPU ms | CPU/GPU |
| --- | --- | ---: | ---: | ---: |
| Buffer affine/clamp/sum | 16384 floats | 0.300 | 1.374 | 0.218 |
| Buffer/image blur | 128 x 128 | 0.589 | 1.639 | 0.359 |
| Borrowed image blur | 128 x 128 | 0.637 | 1.582 | 0.403 |
| Hybrid affine/clamp/sum | 16384 floats | 0.288 | 1.217 | 0.237 |
| Dynamic library stages | 16384 floats | 0.303 | 1.463 | 0.207 |
| RAW10/debayer/luminance | 128 x 128 | 1.078 | 1.643 | 0.656 |
| 2D convolution 5 x 5 | 128 x 128 vec2 | 1.521 | 0.748 | 2.034 |
| 3D convolution 3 x 3 x 3 | 24 x 24 x 24 | 1.503 | 0.706 | 2.129 |
| Forward/inverse direct DFT | 32 x 32 complex | 1.232 | 1.643 | 0.750 |
| Gaussian/Sobel/Harris/NMS | 128 x 128 | 2.379 | 3.138 | 0.758 |
| Lossless tiled RLE round trip | 65536 uints | 0.625 | 1.367 | 0.457 |
| EVT3 event decoding | 65536 packed words | 2.990 | 2.222 | 1.346 |
| Async ring | 32 x 257 floats, 3 slots | 1.207 | 7.759 | 0.156 |
| EVT3 + optical flow | 32 x 24 sampled plane, history 64 | 0.393 | 1.731 | 0.227 |
| ANN training | 32 samples, 33 parameters | 0.085 | 0.520 | 0.164 |
| ANN inference | 32 samples, 33 parameters | 0.004 | 0.156 | 0.024 |
| CNN training | 32 samples, 23 parameters | 1.615 | 0.644 | 2.510 |
| CNN inference | 32 samples, 23 parameters | 0.064 | 0.161 | 0.397 |
| Transformer training | 8 sequences, 4 tokens x 4, 134 parameters | 1.324 | 0.770 | 1.719 |
| Transformer inference | 8 sequences, 4 tokens x 4, 134 parameters | 0.008 | 0.171 | 0.045 |
| FFT forward/inverse | 8 x 1024 complex | 3.050 | 10.656 | 0.286 |
| FIR low-pass | 8 x 2048 floats, 65 taps | 2.718 | 0.772 | 3.523 |
| Biquad low-pass | 8 x 2048 floats, persistent state | 0.097 | 0.233 | 0.417 |
| FFT band-pass | 8 x 1024 complex, 4-10 kHz | 3.219 | 11.268 | 0.286 |
| LeNet classifier | 32 x 28 x 28, 61,706 parameters | 46.521 | 7.622 | 6.104 |

Core i7-1260P / Iris Xe

Core i7-1260P / Iris Xe mean pipeline timings and CPU-to-graphics ratios
Time and ratio axes are logarithmic. A ratio above one favors the graphics path. Means are displayed to three decimal places. Ratios come from the original unrounded means, then are rounded for reporting. Download SVG
Exact timing and ratio table · Core i7-1260P / Iris Xe
Core i7-1260P / Iris Xe · 20 measured repetitions
Pipeline Workload CPU ms Graphics ms CPU / graphics
Buffer affine/clamp/sum 16384 floats 0.139 3.928 0.035
Buffer/image blur 128 x 128 0.422 0.054 7.838
Borrowed image blur 128 x 128 0.299 0.052 5.764
Hybrid affine/clamp/sum 16384 floats 0.145 2.825 0.051
Dynamic library stages 16384 floats 0.460 2.100 0.219
RAW10/debayer/luminance 128 x 128 1.080 0.068 15.982
2D convolution 5 x 5 128 x 128 vec2 1.140 0.059 19.428
3D convolution 3 x 3 x 3 24 x 24 x 24 1.266 0.057 22.166
Forward/inverse direct DFT 32 x 32 complex 0.892 0.080 11.134
Gaussian/Sobel/Harris/NMS 128 x 128 1.841 0.089 20.761
Lossless tiled RLE round trip 65536 uints 0.462 0.248 1.864
EVT3 event decoding 65536 sixteen-bit words 2.993 0.734 4.079
Async ring 32 x 257 floats, 3 slots 0.094 0.730 0.128
EVT3 + optical flow 32 x 24 sampled plane, history 64 0.290 0.215 1.347
ANN training 32 samples, 33 parameters 0.072 0.042 1.702
ANN inference 32 samples, 33 parameters 0.002 0.015 0.145
CNN training 32 samples, 23 parameters 1.359 0.086 15.741
CNN inference 32 samples, 23 parameters 0.054 0.034 1.570
Transformer training 8 sequences, 4 tokens x 4, 134 parameters 0.972 0.284 3.427
Transformer inference 8 sequences, 4 tokens x 4, 134 parameters 0.006 0.057 0.110
FFT forward/inverse 8 x 1024 complex 2.226 0.226 9.856
FIR low-pass 8 x 2048 floats, 65 taps 1.956 0.124 15.724
Biquad low-pass 8 x 2048 floats, persistent state 0.040 0.515 0.077
FFT band-pass 8 x 1024 complex, 4-10 kHz 2.250 0.222 10.150
LeNet classifier 32 x 28 x 28, 61,706 parameters 40.790 1.756 23.229
Complete benchmark output · Core i7-1260P / Iris Xe

This is the recorded benchmark output. Its EVT3 wording “packed words” counts 16-bit sensor words before two words are packed into each uint.

Seed: 12345, iterations: 20, scale: 1.
GPU renderer: Mesa Intel(R) Iris(R) Xe Graphics (ADL GT2).
Lossless RLE: 13556 payload words + 257 offset words, 4.74x size reduction.
EVT3 compact output: 223772 events from 65536 words, five stages include state reset.
Optical flow input: 610 seeded samples on a known temporal plane, six linked stages.
Neural training timings measure one mini-batch gradient/SGD update, inference measures the complete forward pipeline.
Steady-state resident timings, compilation, first uploads and final readback excluded.
Hybrid timings include transfers between CPU and GPU stages.

| Pipeline | Workload | CPU ms | GPU ms | CPU/GPU |
| --- | --- | ---: | ---: | ---: |
| Buffer affine/clamp/sum | 16384 floats | 0.139 | 3.928 | 0.035 |
| Buffer/image blur | 128 x 128 | 0.422 | 0.054 | 7.838 |
| Borrowed image blur | 128 x 128 | 0.299 | 0.052 | 5.764 |
| Hybrid affine/clamp/sum | 16384 floats | 0.145 | 2.825 | 0.051 |
| Dynamic library stages | 16384 floats | 0.460 | 2.100 | 0.219 |
| RAW10/debayer/luminance | 128 x 128 | 1.080 | 0.068 | 15.982 |
| 2D convolution 5 x 5 | 128 x 128 vec2 | 1.140 | 0.059 | 19.428 |
| 3D convolution 3 x 3 x 3 | 24 x 24 x 24 | 1.266 | 0.057 | 22.166 |
| Forward/inverse direct DFT | 32 x 32 complex | 0.892 | 0.080 | 11.134 |
| Gaussian/Sobel/Harris/NMS | 128 x 128 | 1.841 | 0.089 | 20.761 |
| Lossless tiled RLE round trip | 65536 uints | 0.462 | 0.248 | 1.864 |
| EVT3 event decoding | 65536 packed words | 2.993 | 0.734 | 4.079 |
| Async ring | 32 x 257 floats, 3 slots | 0.094 | 0.730 | 0.128 |
| EVT3 + optical flow | 32 x 24 sampled plane, history 64 | 0.290 | 0.215 | 1.347 |
| ANN training | 32 samples, 33 parameters | 0.072 | 0.042 | 1.702 |
| ANN inference | 32 samples, 33 parameters | 0.002 | 0.015 | 0.145 |
| CNN training | 32 samples, 23 parameters | 1.359 | 0.086 | 15.741 |
| CNN inference | 32 samples, 23 parameters | 0.054 | 0.034 | 1.570 |
| Transformer training | 8 sequences, 4 tokens x 4, 134 parameters | 0.972 | 0.284 | 3.427 |
| Transformer inference | 8 sequences, 4 tokens x 4, 134 parameters | 0.006 | 0.057 | 0.110 |
| FFT forward/inverse | 8 x 1024 complex | 2.226 | 0.226 | 9.856 |
| FIR low-pass | 8 x 2048 floats, 65 taps | 1.956 | 0.124 | 15.724 |
| Biquad low-pass | 8 x 2048 floats, persistent state | 0.040 | 0.515 | 0.077 |
| FFT band-pass | 8 x 1024 complex, 4-10 kHz | 2.250 | 0.222 | 10.150 |
| LeNet classifier | 32 x 28 x 28, 61,706 parameters | 40.790 | 1.756 | 23.229 |

Raspberry Pi 5 / V3D

Raspberry Pi 5 / V3D mean pipeline timings and CPU-to-graphics ratios
Time and ratio axes are logarithmic. A ratio above one favors the graphics path. Means are displayed to three decimal places. Ratios come from the original unrounded means, then are rounded for reporting. Download SVG
Exact timing and ratio table · Raspberry Pi 5 / V3D
Raspberry Pi 5 / V3D · 20 measured repetitions
Pipeline Workload CPU ms Graphics ms CPU / graphics
Buffer affine/clamp/sum 16384 floats 0.516 1.956 0.264
Buffer/image blur 128 x 128 1.033 0.471 2.192
Borrowed image blur 128 x 128 0.951 0.469 2.029
Hybrid affine/clamp/sum 16384 floats 0.597 2.188 0.273
Dynamic library stages 16384 floats 0.838 1.968 0.426
RAW10/debayer/luminance 128 x 128 2.466 2.522 0.978
2D convolution 5 x 5 128 x 128 vec2 3.283 2.444 1.343
3D convolution 3 x 3 x 3 24 x 24 x 24 3.156 3.020 1.045
Forward/inverse direct DFT 32 x 32 complex 2.589 0.579 4.472
Gaussian/Sobel/Harris/NMS 128 x 128 4.571 2.345 1.949
Lossless tiled RLE round trip 65536 uints 1.256 1.253 1.002
EVT3 event decoding 65536 sixteen-bit words 4.873 6.988 0.697
Async ring 32 x 257 floats, 3 slots 0.143 1.976 0.072
EVT3 + optical flow 32 x 24 sampled plane, history 64 0.640 0.656 0.976
ANN training 32 samples, 33 parameters 0.206 0.114 1.804
ANN inference 32 samples, 33 parameters 0.006 0.037 0.173
CNN training 32 samples, 23 parameters 3.254 0.490 6.636
CNN inference 32 samples, 23 parameters 0.138 0.057 2.393
Transformer training 8 sequences, 4 tokens x 4, 134 parameters 2.404 0.729 3.296
Transformer inference 8 sequences, 4 tokens x 4, 134 parameters 0.018 0.055 0.330
FFT forward/inverse 8 x 1024 complex 5.936 4.174 1.422
FIR low-pass 8 x 2048 floats, 65 taps 6.570 5.499 1.195
Biquad low-pass 8 x 2048 floats, persistent state 0.156 0.691 0.226
FFT band-pass 8 x 1024 complex, 4-10 kHz 6.153 5.001 1.230
LeNet classifier 32 x 28 x 28, 61,706 parameters 116.491 136.617 0.853
Complete benchmark output · Raspberry Pi 5 / V3D

This is the recorded benchmark output. Its EVT3 wording “packed words” counts 16-bit sensor words before two words are packed into each uint.

Seed: 12345, iterations: 20, scale: 1.
GPU renderer: V3D 7.1.7.0.
Lossless RLE: 13556 payload words + 257 offset words, 4.74x size reduction.
EVT3 compact output: 223772 events from 65536 words, five stages include state reset.
Optical flow input: 610 seeded samples on a known temporal plane, six linked stages.
Neural training timings measure one mini-batch gradient/SGD update, inference measures the complete forward pipeline.
Steady-state resident timings, compilation, first uploads and final readback excluded.
Hybrid timings include transfers between CPU and GPU stages.

| Pipeline | Workload | CPU ms | GPU ms | CPU/GPU |
| --- | --- | ---: | ---: | ---: |
| Buffer affine/clamp/sum | 16384 floats | 0.516 | 1.956 | 0.264 |
| Buffer/image blur | 128 x 128 | 1.033 | 0.471 | 2.192 |
| Borrowed image blur | 128 x 128 | 0.951 | 0.469 | 2.029 |
| Hybrid affine/clamp/sum | 16384 floats | 0.597 | 2.188 | 0.273 |
| Dynamic library stages | 16384 floats | 0.838 | 1.968 | 0.426 |
| RAW10/debayer/luminance | 128 x 128 | 2.466 | 2.522 | 0.978 |
| 2D convolution 5 x 5 | 128 x 128 vec2 | 3.283 | 2.444 | 1.343 |
| 3D convolution 3 x 3 x 3 | 24 x 24 x 24 | 3.156 | 3.020 | 1.045 |
| Forward/inverse direct DFT | 32 x 32 complex | 2.589 | 0.579 | 4.472 |
| Gaussian/Sobel/Harris/NMS | 128 x 128 | 4.571 | 2.345 | 1.949 |
| Lossless tiled RLE round trip | 65536 uints | 1.256 | 1.253 | 1.002 |
| EVT3 event decoding | 65536 packed words | 4.873 | 6.988 | 0.697 |
| Async ring | 32 x 257 floats, 3 slots | 0.143 | 1.976 | 0.072 |
| EVT3 + optical flow | 32 x 24 sampled plane, history 64 | 0.640 | 0.656 | 0.976 |
| ANN training | 32 samples, 33 parameters | 0.206 | 0.114 | 1.804 |
| ANN inference | 32 samples, 33 parameters | 0.006 | 0.037 | 0.173 |
| CNN training | 32 samples, 23 parameters | 3.254 | 0.490 | 6.636 |
| CNN inference | 32 samples, 23 parameters | 0.138 | 0.057 | 2.393 |
| Transformer training | 8 sequences, 4 tokens x 4, 134 parameters | 2.404 | 0.729 | 3.296 |
| Transformer inference | 8 sequences, 4 tokens x 4, 134 parameters | 0.018 | 0.055 | 0.330 |
| FFT forward/inverse | 8 x 1024 complex | 5.936 | 4.174 | 1.422 |
| FIR low-pass | 8 x 2048 floats, 65 taps | 6.570 | 5.499 | 1.195 |
| Biquad low-pass | 8 x 2048 floats, persistent state | 0.156 | 0.691 | 0.226 |
| FFT band-pass | 8 x 1024 complex, 4-10 kHz | 6.153 | 5.001 | 1.230 |
| LeNet classifier | 32 x 28 x 28, 61,706 parameters | 116.491 | 136.617 | 0.853 |

The EVT3 decoder emits 223772 ordered events from 65536 sensor words. Five stages include resetting decoder state. Normal flow uses 610 seeded samples from a 32×24 temporal plane and history 64. The RLE payload has 13556 words plus 257 offset words, a 4.74-fold reduction for this input. The encoder still reserves worst-case capacity. Prefix and reduction stages can dominate small graphics workloads.

The timing plots show the archived mean values at their recorded precision. Differences across machines include CPU, operating system, driver and renderer effects.

# Reproduce the resident measurement settings.
SHDR_BENCH_ITERATIONS=20 SHDR_BENCH_SCALE=1 ./build/example_benchmark --gpu es31

Complete held-out digit evaluation

The modern LeNet-style classifier produced 9778 correct predictions from all 10,000 official MNIST test images, or 97.78%, on every recorded CPU and graphics path. Every one of the 100,000 output logits was compared with the shipped independent framework reference.

The seven-stage model has 61,706 parameters, full second-convolution connectivity, ReLU, fixed average pooling and a linear ten-class head. It is a documented modern LeNet-5 adaptation. Native deployment uses plain float weights and the shdr C++/OpenGL runtime. The neural guide describes the architecture and parameter layout.

All six complete native evaluations · identical official test order
Configuration Execution path Correct Images Accuracy % Completed seconds Maximum normalized logit error
Ryzen 5900X / llvmpipe CPU 9778 10000 97.78 17.2063 4.57699e-06
Ryzen 5900X / llvmpipe llvmpipe software 9778 10000 97.78 1.76238 4.57699e-06
Core i7-1260P / Iris Xe CPU 9778 10000 97.78 13.8871 4.57699e-06
Core i7-1260P / Iris Xe Iris Xe hardware 9778 10000 97.78 1.95504 4.00418e-06
Raspberry Pi 5 / V3D CPU 9778 10000 97.78 35.0628 4.00418e-06
Raspberry Pi 5 / V3D V3D hardware 9778 10000 97.78 41.5901 4.57699e-06
Accuracy and completed time for all six full MNIST evaluations
Every path has 97.78% accuracy. These evaluation times include bounded batch construction, input normalization, program setup, native inference, result reads and reference comparisons. Timing begins after the initial dataset reads. The resident benchmarks above measure their own dispatch intervals. Download SVG

Normalized logit error is abs(actual-reference)/(1+abs(reference)), maximized over all 100,000 scores. The model was prepared on the separate 60,000-image training split for three complete epochs with Adam, learning rate 0.001, batch size 128 and seed 12345. Independent framework evaluation also recorded 97.78%. The data guide records public input provenance and preprocessing.

# Prepare all official test images explicitly before a complete evaluation.
python3 scripts/prepare_datasets.py --full-mnist
# Evaluate every held-out digit on the selected native backend.
SHDR_FULL_MNIST_DIR=build/datasets ./build/example_real_data --gpu es31

Reference confusion counts

This matrix counts the independent framework predictions from the shipped reference logits and all original test labels. The native evaluations above record total accuracy and the maximum normalized error across all 100,000 logits.

Actual independent reference confusion counts for all official test digits
Rows are official labels and columns are independent framework predictions. Cell annotations are exact counts. Positive color values use a logarithmic scale, and zero-count cells remain white. Download SVG
Exact reference confusion table · 10000 official test images
Independent framework reference counts · true class by predicted class
True digit / predicted 0 1 2 3 4 5 6 7 8 9
0 969 0 1 0 0 1 3 1 4 1
1 0 1121 2 1 0 0 1 1 9 0
2 6 1 1009 4 0 0 1 4 7 0
3 1 0 3 982 0 14 0 4 6 0
4 1 2 4 0 936 0 3 3 2 31
5 1 0 1 1 0 884 2 1 2 0
6 5 3 0 1 4 6 936 0 3 0
7 1 1 5 3 0 0 0 1009 3 6
8 3 0 2 4 1 5 0 1 956 2
9 4 5 0 6 2 6 0 2 8 976

Small-model training on real zero/one images

The ANN, CNN and transformer examples fit one balanced 32-image mini-batch from the official training split. They use the original binary digit labels with documented feature transformations. The numbers below are training loss before and after the recorded updates. These measurements describe optimization on the training images. The LeNet evaluation above reports held-out classification quality. All six complete evaluation logs print the same values at the recorded precision.

Recorded observations on one 32-image training mini-batch
Model SGD updates Initial mean binary cross entropy Final mean binary cross entropy
ANN 32 0.694159 0.20248
CNN 32 0.699285 0.446509
Transformer 12 0.61123 0.360663
Recorded initial and final mini-batch losses for ANN, CNN and transformer
This figure compares the two recorded observations: initial and final mini-batch loss. ANN and CNN have 32 updates, and the transformer has 12. Download SVG

Complete real-data functional outputs

Each run includes linked and mixed resources, asynchronous submission, dynamic libraries, image storage, all Bayer packing/phase combinations, corners, direct DFT and reconstruction, 3D convolution, FFT and windows, five FIR modes, four biquad modes, exact EVT3 decoding and normal flow, small-model training, and the full deeper classifier. Satellite corners, event flow and ECG processing use numerical reference checks and backend comparisons in these fixtures.

Complete public-data evaluation log · Ryzen 5900X / llvmpipe / CPU
NASA linked affine/clamp/reduction: verified, normalized CPU/GPU error 0
NASA mixed CPU/GPU resources: verified, normalized CPU/GPU error 0
NASA asynchronous resource submission: verified
NASA dynamically loaded shader pipeline: verified, normalized CPU/GPU error 0
NASA buffer/image/box-filter pipeline: verified, normalized CPU/GPU error 0
NASA Gaussian/Sobel/Harris/nonmaximum suppression: verified, normalized CPU/GPU error 0
NASA selected Harris peaks: 61
NASA 2D convolution/DFT/inverse: verified, normalized CPU/GPU error 0
MNIST image-stack 3D convolution: verified, normalized CPU/GPU error 0
NASA derived Bayer packing 0 phase 0: verified, normalized CPU/GPU error 0
NASA derived Bayer packing 0 phase 1: verified, normalized CPU/GPU error 0
NASA derived Bayer packing 0 phase 2: verified, normalized CPU/GPU error 0
NASA derived Bayer packing 0 phase 3: verified, normalized CPU/GPU error 0
NASA derived Bayer packing 1 phase 0: verified, normalized CPU/GPU error 0
NASA derived Bayer packing 1 phase 1: verified, normalized CPU/GPU error 0
NASA derived Bayer packing 1 phase 2: verified, normalized CPU/GPU error 0
NASA derived Bayer packing 1 phase 3: verified, normalized CPU/GPU error 0
NASA derived Bayer packing 2 phase 0: verified, normalized CPU/GPU error 0
NASA derived Bayer packing 2 phase 1: verified, normalized CPU/GPU error 0
NASA derived Bayer packing 2 phase 2: verified, normalized CPU/GPU error 0
NASA derived Bayer packing 2 phase 3: verified, normalized CPU/GPU error 0
NASA derived Bayer packing 3 phase 0: verified, normalized CPU/GPU error 0
NASA derived Bayer packing 3 phase 1: verified, normalized CPU/GPU error 0
NASA derived Bayer packing 3 phase 2: verified, normalized CPU/GPU error 0
NASA derived Bayer packing 3 phase 3: verified, normalized CPU/GPU error 0
NASA tiled lossless RLE image roundtrip: verified, normalized CPU/GPU error 0
NASA RLE encoded words: 15996 from 12288 input words
PhysioNet ECG 1D FFT/inverse: verified, normalized CPU/GPU error 0
PhysioNet ECG FFT filter 0: verified, normalized CPU/GPU error 0
PhysioNet ECG FFT filter 1: verified, normalized CPU/GPU error 0
PhysioNet ECG FFT filter 2: verified, normalized CPU/GPU error 0
PhysioNet ECG FFT filter 3: verified, normalized CPU/GPU error 0
PhysioNet ECG spectral window 0: verified, normalized CPU/GPU error 0
PhysioNet ECG spectral window 1: verified, normalized CPU/GPU error 0
PhysioNet ECG spectral window 2: verified, normalized CPU/GPU error 0
PhysioNet ECG spectral window 3: verified, normalized CPU/GPU error 0
PhysioNet ECG FIR filter 0: verified, normalized CPU/GPU error 0
PhysioNet ECG FIR filter 1: verified, normalized CPU/GPU error 0
PhysioNet ECG FIR filter 2: verified, normalized CPU/GPU error 0
PhysioNet ECG FIR filter 3: verified, normalized CPU/GPU error 0
PhysioNet ECG FIR filter 4: verified, normalized CPU/GPU error 0
PhysioNet ECG biquad filter 0: verified, normalized CPU/GPU error 0
PhysioNet ECG biquad filter 1: verified, normalized CPU/GPU error 0
PhysioNet ECG biquad filter 2: verified, normalized CPU/GPU error 0
PhysioNet ECG biquad filter 3: verified, normalized CPU/GPU error 0
Prophesee real EVT3 decode/normal-flow pipeline: verified, normalized CPU/GPU error 0
Prophesee recording: 21950 decoded events, 21950 compact CD words, 500 accepted normal-flow fits
MNIST two-feature ANN training/inference: verified, normalized CPU/GPU error 0
MNIST two-feature ANN mini-batch loss: 0.694159 -> 0.20248
MNIST compact CNN training/inference: verified, normalized CPU/GPU error 0
MNIST compact CNN mini-batch loss: 0.699285 -> 0.446509
MNIST patch-token transformer training/inference: verified, normalized CPU/GPU error 0
MNIST patch-token transformer mini-batch loss: 0.61123 -> 0.360663
LeNet-style classifier: all seven layers and 61706 parameters verified against a double oracle
LeNet MNIST held-out: 9778/10000 correct, accuracy 97.78%, normalized reference error 4.57699e-06, completed seconds 17.2063
All public real-data functional cases passed
Complete public-data evaluation log · Ryzen 5900X / llvmpipe / ES 3.1 graphics
NASA linked affine/clamp/reduction: verified, normalized CPU/GPU error 0
NASA mixed CPU/GPU resources: verified, normalized CPU/GPU error 0
NASA asynchronous resource submission: verified
NASA dynamically loaded shader pipeline: verified, normalized CPU/GPU error 0
NASA buffer/image/box-filter pipeline: verified, normalized CPU/GPU error 0
NASA Gaussian/Sobel/Harris/nonmaximum suppression: verified, normalized CPU/GPU error 0
NASA selected Harris peaks: 61
NASA 2D convolution/DFT/inverse: verified, normalized CPU/GPU error 9.86406e-07
MNIST image-stack 3D convolution: verified, normalized CPU/GPU error 0
NASA derived Bayer packing 0 phase 0: verified, normalized CPU/GPU error 6.86873e-08
NASA derived Bayer packing 0 phase 1: verified, normalized CPU/GPU error 4.38723e-08
NASA derived Bayer packing 0 phase 2: verified, normalized CPU/GPU error 4.4295e-08
NASA derived Bayer packing 0 phase 3: verified, normalized CPU/GPU error 4.36838e-08
NASA derived Bayer packing 1 phase 0: verified, normalized CPU/GPU error 6.86873e-08
NASA derived Bayer packing 1 phase 1: verified, normalized CPU/GPU error 4.38723e-08
NASA derived Bayer packing 1 phase 2: verified, normalized CPU/GPU error 4.4295e-08
NASA derived Bayer packing 1 phase 3: verified, normalized CPU/GPU error 4.36838e-08
NASA derived Bayer packing 2 phase 0: verified, normalized CPU/GPU error 3.99307e-08
NASA derived Bayer packing 2 phase 1: verified, normalized CPU/GPU error 4.23175e-08
NASA derived Bayer packing 2 phase 2: verified, normalized CPU/GPU error 6.92246e-08
NASA derived Bayer packing 2 phase 3: verified, normalized CPU/GPU error 4.29885e-08
NASA derived Bayer packing 3 phase 0: verified, normalized CPU/GPU error 4.28234e-08
NASA derived Bayer packing 3 phase 1: verified, normalized CPU/GPU error 4.13282e-08
NASA derived Bayer packing 3 phase 2: verified, normalized CPU/GPU error 7.13851e-08
NASA derived Bayer packing 3 phase 3: verified, normalized CPU/GPU error 4.44882e-08
NASA tiled lossless RLE image roundtrip: verified, normalized CPU/GPU error 0
NASA RLE encoded words: 15996 from 12288 input words
PhysioNet ECG 1D FFT/inverse: verified, normalized CPU/GPU error 0
PhysioNet ECG FFT filter 0: verified, normalized CPU/GPU error 2.3254e-08
PhysioNet ECG FFT filter 1: verified, normalized CPU/GPU error 3.29144e-08
PhysioNet ECG FFT filter 2: verified, normalized CPU/GPU error 3.84523e-08
PhysioNet ECG FFT filter 3: verified, normalized CPU/GPU error 4.5632e-08
PhysioNet ECG spectral window 0: verified, normalized CPU/GPU error 0
PhysioNet ECG spectral window 1: verified, normalized CPU/GPU error 3.69377e-06
PhysioNet ECG spectral window 2: verified, normalized CPU/GPU error 3.20483e-06
PhysioNet ECG spectral window 3: verified, normalized CPU/GPU error 4.3956e-06
PhysioNet ECG FIR filter 0: verified, normalized CPU/GPU error 0
PhysioNet ECG FIR filter 1: verified, normalized CPU/GPU error 0
PhysioNet ECG FIR filter 2: verified, normalized CPU/GPU error 0
PhysioNet ECG FIR filter 3: verified, normalized CPU/GPU error 0
PhysioNet ECG FIR filter 4: verified, normalized CPU/GPU error 0
PhysioNet ECG biquad filter 0: verified, normalized CPU/GPU error 1.18087e-07
PhysioNet ECG biquad filter 1: verified, normalized CPU/GPU error 1.16945e-07
PhysioNet ECG biquad filter 2: verified, normalized CPU/GPU error 0
PhysioNet ECG biquad filter 3: verified, normalized CPU/GPU error 4.75941e-08
Prophesee real EVT3 decode/normal-flow pipeline: verified, normalized CPU/GPU error 0
Prophesee recording: 21950 decoded events, 21950 compact CD words, 500 accepted normal-flow fits
MNIST two-feature ANN training/inference: verified, normalized CPU/GPU error 1.59604e-07
MNIST two-feature ANN mini-batch loss: 0.694159 -> 0.20248
MNIST compact CNN training/inference: verified, normalized CPU/GPU error 1.43349e-07
MNIST compact CNN mini-batch loss: 0.699285 -> 0.446509
MNIST patch-token transformer training/inference: verified, normalized CPU/GPU error 9.37875e-08
MNIST patch-token transformer mini-batch loss: 0.61123 -> 0.360663
LeNet-style classifier: all seven layers and 61706 parameters verified against a double oracle
LeNet MNIST held-out: 9778/10000 correct, accuracy 97.78%, normalized reference error 4.57699e-06, completed seconds 1.76238
All public real-data functional cases passed
Complete public-data evaluation log · Core i7-1260P / Iris Xe / CPU
NASA linked affine/clamp/reduction: verified, normalized CPU/GPU error 0
NASA mixed CPU/GPU resources: verified, normalized CPU/GPU error 0
NASA asynchronous resource submission: verified
NASA dynamically loaded shader pipeline: verified, normalized CPU/GPU error 0
NASA buffer/image/box-filter pipeline: verified, normalized CPU/GPU error 0
NASA Gaussian/Sobel/Harris/nonmaximum suppression: verified, normalized CPU/GPU error 0
NASA selected Harris peaks: 61
NASA 2D convolution/DFT/inverse: verified, normalized CPU/GPU error 0
MNIST image-stack 3D convolution: verified, normalized CPU/GPU error 0
NASA derived Bayer packing 0 phase 0: verified, normalized CPU/GPU error 0
NASA derived Bayer packing 0 phase 1: verified, normalized CPU/GPU error 0
NASA derived Bayer packing 0 phase 2: verified, normalized CPU/GPU error 0
NASA derived Bayer packing 0 phase 3: verified, normalized CPU/GPU error 0
NASA derived Bayer packing 1 phase 0: verified, normalized CPU/GPU error 0
NASA derived Bayer packing 1 phase 1: verified, normalized CPU/GPU error 0
NASA derived Bayer packing 1 phase 2: verified, normalized CPU/GPU error 0
NASA derived Bayer packing 1 phase 3: verified, normalized CPU/GPU error 0
NASA derived Bayer packing 2 phase 0: verified, normalized CPU/GPU error 0
NASA derived Bayer packing 2 phase 1: verified, normalized CPU/GPU error 0
NASA derived Bayer packing 2 phase 2: verified, normalized CPU/GPU error 0
NASA derived Bayer packing 2 phase 3: verified, normalized CPU/GPU error 0
NASA derived Bayer packing 3 phase 0: verified, normalized CPU/GPU error 0
NASA derived Bayer packing 3 phase 1: verified, normalized CPU/GPU error 0
NASA derived Bayer packing 3 phase 2: verified, normalized CPU/GPU error 0
NASA derived Bayer packing 3 phase 3: verified, normalized CPU/GPU error 0
NASA tiled lossless RLE image roundtrip: verified, normalized CPU/GPU error 0
NASA RLE encoded words: 15996 from 12288 input words
PhysioNet ECG 1D FFT/inverse: verified, normalized CPU/GPU error 0
PhysioNet ECG FFT filter 0: verified, normalized CPU/GPU error 0
PhysioNet ECG FFT filter 1: verified, normalized CPU/GPU error 0
PhysioNet ECG FFT filter 2: verified, normalized CPU/GPU error 0
PhysioNet ECG FFT filter 3: verified, normalized CPU/GPU error 0
PhysioNet ECG spectral window 0: verified, normalized CPU/GPU error 0
PhysioNet ECG spectral window 1: verified, normalized CPU/GPU error 0
PhysioNet ECG spectral window 2: verified, normalized CPU/GPU error 0
PhysioNet ECG spectral window 3: verified, normalized CPU/GPU error 0
PhysioNet ECG FIR filter 0: verified, normalized CPU/GPU error 0
PhysioNet ECG FIR filter 1: verified, normalized CPU/GPU error 0
PhysioNet ECG FIR filter 2: verified, normalized CPU/GPU error 0
PhysioNet ECG FIR filter 3: verified, normalized CPU/GPU error 0
PhysioNet ECG FIR filter 4: verified, normalized CPU/GPU error 0
PhysioNet ECG biquad filter 0: verified, normalized CPU/GPU error 0
PhysioNet ECG biquad filter 1: verified, normalized CPU/GPU error 0
PhysioNet ECG biquad filter 2: verified, normalized CPU/GPU error 0
PhysioNet ECG biquad filter 3: verified, normalized CPU/GPU error 0
Prophesee real EVT3 decode/normal-flow pipeline: verified, normalized CPU/GPU error 0
Prophesee recording: 21950 decoded events, 21950 compact CD words, 500 accepted normal-flow fits
MNIST two-feature ANN training/inference: verified, normalized CPU/GPU error 0
MNIST two-feature ANN mini-batch loss: 0.694159 -> 0.20248
MNIST compact CNN training/inference: verified, normalized CPU/GPU error 0
MNIST compact CNN mini-batch loss: 0.699285 -> 0.446509
MNIST patch-token transformer training/inference: verified, normalized CPU/GPU error 0
MNIST patch-token transformer mini-batch loss: 0.61123 -> 0.360663
LeNet-style classifier: all seven layers and 61706 parameters verified against a double oracle
LeNet MNIST held-out: 9778/10000 correct, accuracy 97.78%, normalized reference error 4.57699e-06, completed seconds 13.8871
All public real-data functional cases passed
Complete public-data evaluation log · Core i7-1260P / Iris Xe / ES 3.1 graphics
NASA linked affine/clamp/reduction: verified, normalized CPU/GPU error 5.94309e-08
NASA mixed CPU/GPU resources: verified, normalized CPU/GPU error 3.97048e-08
NASA asynchronous resource submission: verified
NASA dynamically loaded shader pipeline: verified, normalized CPU/GPU error 5.94309e-08
NASA buffer/image/box-filter pipeline: verified, normalized CPU/GPU error 3.96951e-08
NASA Gaussian/Sobel/Harris/nonmaximum suppression: verified, normalized CPU/GPU error 7.93092e-08
NASA selected Harris peaks: 61
NASA 2D convolution/DFT/inverse: verified, normalized CPU/GPU error 1.80282e-06
MNIST image-stack 3D convolution: verified, normalized CPU/GPU error 1.16552e-07
NASA derived Bayer packing 0 phase 0: verified, normalized CPU/GPU error 1.03031e-07
NASA derived Bayer packing 0 phase 1: verified, normalized CPU/GPU error 7.81543e-08
NASA derived Bayer packing 0 phase 2: verified, normalized CPU/GPU error 1.018e-07
NASA derived Bayer packing 0 phase 3: verified, normalized CPU/GPU error 7.94091e-08
NASA derived Bayer packing 1 phase 0: verified, normalized CPU/GPU error 7.03978e-08
NASA derived Bayer packing 1 phase 1: verified, normalized CPU/GPU error 7.08066e-08
NASA derived Bayer packing 1 phase 2: verified, normalized CPU/GPU error 4.4786e-08
NASA derived Bayer packing 1 phase 3: verified, normalized CPU/GPU error 6.23303e-08
NASA derived Bayer packing 2 phase 0: verified, normalized CPU/GPU error 7.20829e-08
NASA derived Bayer packing 2 phase 1: verified, normalized CPU/GPU error 7.39612e-08
NASA derived Bayer packing 2 phase 2: verified, normalized CPU/GPU error 6.92246e-08
NASA derived Bayer packing 2 phase 3: verified, normalized CPU/GPU error 7.3392e-08
NASA derived Bayer packing 3 phase 0: verified, normalized CPU/GPU error 7.86475e-08
NASA derived Bayer packing 3 phase 1: verified, normalized CPU/GPU error 7.70446e-08
NASA derived Bayer packing 3 phase 2: verified, normalized CPU/GPU error 7.74885e-08
NASA derived Bayer packing 3 phase 3: verified, normalized CPU/GPU error 1.01541e-07
NASA tiled lossless RLE image roundtrip: verified, normalized CPU/GPU error 0
NASA RLE encoded words: 15996 from 12288 input words
PhysioNet ECG 1D FFT/inverse: verified, normalized CPU/GPU error 3.6866e-06
PhysioNet ECG FFT filter 0: verified, normalized CPU/GPU error 7.13594e-08
PhysioNet ECG FFT filter 1: verified, normalized CPU/GPU error 1.47238e-07
PhysioNet ECG FFT filter 2: verified, normalized CPU/GPU error 1.1783e-07
PhysioNet ECG FFT filter 3: verified, normalized CPU/GPU error 9.9671e-08
PhysioNet ECG spectral window 0: verified, normalized CPU/GPU error 3.6866e-06
PhysioNet ECG spectral window 1: verified, normalized CPU/GPU error 4.82402e-05
PhysioNet ECG spectral window 2: verified, normalized CPU/GPU error 4.10839e-05
PhysioNet ECG spectral window 3: verified, normalized CPU/GPU error 6.12265e-05
PhysioNet ECG FIR filter 0: verified, normalized CPU/GPU error 6.89868e-08
PhysioNet ECG FIR filter 1: verified, normalized CPU/GPU error 6.79252e-08
PhysioNet ECG FIR filter 2: verified, normalized CPU/GPU error 6.75995e-08
PhysioNet ECG FIR filter 3: verified, normalized CPU/GPU error 7.03995e-08
PhysioNet ECG FIR filter 4: verified, normalized CPU/GPU error 9.13858e-08
PhysioNet ECG biquad filter 0: verified, normalized CPU/GPU error 1.02826e-07
PhysioNet ECG biquad filter 1: verified, normalized CPU/GPU error 1.51945e-07
PhysioNet ECG biquad filter 2: verified, normalized CPU/GPU error 4.99365e-08
PhysioNet ECG biquad filter 3: verified, normalized CPU/GPU error 8.46699e-08
Prophesee real EVT3 decode/normal-flow pipeline: verified, normalized CPU/GPU error 2.94724e-07
Prophesee recording: 21950 decoded events, 21950 compact CD words, 500 accepted normal-flow fits
MNIST two-feature ANN training/inference: verified, normalized CPU/GPU error 2.2301e-07
MNIST two-feature ANN mini-batch loss: 0.694159 -> 0.20248
MNIST compact CNN training/inference: verified, normalized CPU/GPU error 1.21958e-07
MNIST compact CNN mini-batch loss: 0.699285 -> 0.446509
MNIST patch-token transformer training/inference: verified, normalized CPU/GPU error 1.28923e-07
MNIST patch-token transformer mini-batch loss: 0.61123 -> 0.360663
LeNet-style classifier: all seven layers and 61706 parameters verified against a double oracle
LeNet MNIST held-out: 9778/10000 correct, accuracy 97.78%, normalized reference error 4.00418e-06, completed seconds 1.95504
All public real-data functional cases passed
Complete public-data evaluation log · Raspberry Pi 5 / V3D / CPU
NASA linked affine/clamp/reduction: verified, normalized CPU/GPU error 0
NASA mixed CPU/GPU resources: verified, normalized CPU/GPU error 0
NASA asynchronous resource submission: verified
NASA dynamically loaded shader pipeline: verified, normalized CPU/GPU error 0
NASA buffer/image/box-filter pipeline: verified, normalized CPU/GPU error 0
NASA Gaussian/Sobel/Harris/nonmaximum suppression: verified, normalized CPU/GPU error 0
NASA selected Harris peaks: 61
NASA 2D convolution/DFT/inverse: verified, normalized CPU/GPU error 0
MNIST image-stack 3D convolution: verified, normalized CPU/GPU error 0
NASA derived Bayer packing 0 phase 0: verified, normalized CPU/GPU error 0
NASA derived Bayer packing 0 phase 1: verified, normalized CPU/GPU error 0
NASA derived Bayer packing 0 phase 2: verified, normalized CPU/GPU error 0
NASA derived Bayer packing 0 phase 3: verified, normalized CPU/GPU error 0
NASA derived Bayer packing 1 phase 0: verified, normalized CPU/GPU error 0
NASA derived Bayer packing 1 phase 1: verified, normalized CPU/GPU error 0
NASA derived Bayer packing 1 phase 2: verified, normalized CPU/GPU error 0
NASA derived Bayer packing 1 phase 3: verified, normalized CPU/GPU error 0
NASA derived Bayer packing 2 phase 0: verified, normalized CPU/GPU error 0
NASA derived Bayer packing 2 phase 1: verified, normalized CPU/GPU error 0
NASA derived Bayer packing 2 phase 2: verified, normalized CPU/GPU error 0
NASA derived Bayer packing 2 phase 3: verified, normalized CPU/GPU error 0
NASA derived Bayer packing 3 phase 0: verified, normalized CPU/GPU error 0
NASA derived Bayer packing 3 phase 1: verified, normalized CPU/GPU error 0
NASA derived Bayer packing 3 phase 2: verified, normalized CPU/GPU error 0
NASA derived Bayer packing 3 phase 3: verified, normalized CPU/GPU error 0
NASA tiled lossless RLE image roundtrip: verified, normalized CPU/GPU error 0
NASA RLE encoded words: 15996 from 12288 input words
PhysioNet ECG 1D FFT/inverse: verified, normalized CPU/GPU error 0
PhysioNet ECG FFT filter 0: verified, normalized CPU/GPU error 0
PhysioNet ECG FFT filter 1: verified, normalized CPU/GPU error 0
PhysioNet ECG FFT filter 2: verified, normalized CPU/GPU error 0
PhysioNet ECG FFT filter 3: verified, normalized CPU/GPU error 0
PhysioNet ECG spectral window 0: verified, normalized CPU/GPU error 0
PhysioNet ECG spectral window 1: verified, normalized CPU/GPU error 0
PhysioNet ECG spectral window 2: verified, normalized CPU/GPU error 0
PhysioNet ECG spectral window 3: verified, normalized CPU/GPU error 0
PhysioNet ECG FIR filter 0: verified, normalized CPU/GPU error 0
PhysioNet ECG FIR filter 1: verified, normalized CPU/GPU error 0
PhysioNet ECG FIR filter 2: verified, normalized CPU/GPU error 0
PhysioNet ECG FIR filter 3: verified, normalized CPU/GPU error 0
PhysioNet ECG FIR filter 4: verified, normalized CPU/GPU error 0
PhysioNet ECG biquad filter 0: verified, normalized CPU/GPU error 0
PhysioNet ECG biquad filter 1: verified, normalized CPU/GPU error 0
PhysioNet ECG biquad filter 2: verified, normalized CPU/GPU error 0
PhysioNet ECG biquad filter 3: verified, normalized CPU/GPU error 0
Prophesee real EVT3 decode/normal-flow pipeline: verified, normalized CPU/GPU error 0
Prophesee recording: 21950 decoded events, 21950 compact CD words, 500 accepted normal-flow fits
MNIST two-feature ANN training/inference: verified, normalized CPU/GPU error 0
MNIST two-feature ANN mini-batch loss: 0.694159 -> 0.20248
MNIST compact CNN training/inference: verified, normalized CPU/GPU error 0
MNIST compact CNN mini-batch loss: 0.699285 -> 0.446509
MNIST patch-token transformer training/inference: verified, normalized CPU/GPU error 0
MNIST patch-token transformer mini-batch loss: 0.61123 -> 0.360663
LeNet-style classifier: all seven layers and 61706 parameters verified against a double oracle
LeNet MNIST held-out: 9778/10000 correct, accuracy 97.78%, normalized reference error 4.00418e-06, completed seconds 35.0628
All public real-data functional cases passed
Complete public-data evaluation log · Raspberry Pi 5 / V3D / ES 3.1 graphics
NASA linked affine/clamp/reduction: verified, normalized CPU/GPU error 0
NASA mixed CPU/GPU resources: verified, normalized CPU/GPU error 0
NASA asynchronous resource submission: verified
NASA dynamically loaded shader pipeline: verified, normalized CPU/GPU error 0
NASA buffer/image/box-filter pipeline: verified, normalized CPU/GPU error 3.96539e-08
NASA Gaussian/Sobel/Harris/nonmaximum suppression: verified, normalized CPU/GPU error 7.36937e-08
NASA selected Harris peaks: 61
NASA 2D convolution/DFT/inverse: verified, normalized CPU/GPU error 3.6358e-05
MNIST image-stack 3D convolution: verified, normalized CPU/GPU error 1.16552e-07
NASA derived Bayer packing 0 phase 0: verified, normalized CPU/GPU error 7.92272e-08
NASA derived Bayer packing 0 phase 1: verified, normalized CPU/GPU error 1.06714e-07
NASA derived Bayer packing 0 phase 2: verified, normalized CPU/GPU error 1.00736e-07
NASA derived Bayer packing 0 phase 3: verified, normalized CPU/GPU error 1.08238e-07
NASA derived Bayer packing 1 phase 0: verified, normalized CPU/GPU error 6.72781e-08
NASA derived Bayer packing 1 phase 1: verified, normalized CPU/GPU error 4.70995e-08
NASA derived Bayer packing 1 phase 2: verified, normalized CPU/GPU error 7.92794e-08
NASA derived Bayer packing 1 phase 3: verified, normalized CPU/GPU error 7.11205e-08
NASA derived Bayer packing 2 phase 0: verified, normalized CPU/GPU error 7.52445e-08
NASA derived Bayer packing 2 phase 1: verified, normalized CPU/GPU error 6.91206e-08
NASA derived Bayer packing 2 phase 2: verified, normalized CPU/GPU error 7.10627e-08
NASA derived Bayer packing 2 phase 3: verified, normalized CPU/GPU error 4.36114e-08
NASA derived Bayer packing 3 phase 0: verified, normalized CPU/GPU error 7.962e-08
NASA derived Bayer packing 3 phase 1: verified, normalized CPU/GPU error 1.06704e-07
NASA derived Bayer packing 3 phase 2: verified, normalized CPU/GPU error 7.94e-08
NASA derived Bayer packing 3 phase 3: verified, normalized CPU/GPU error 7.94582e-08
NASA tiled lossless RLE image roundtrip: verified, normalized CPU/GPU error 0
NASA RLE encoded words: 15996 from 12288 input words
PhysioNet ECG 1D FFT/inverse: verified, normalized CPU/GPU error 3.68661e-06
PhysioNet ECG FFT filter 0: verified, normalized CPU/GPU error 7.08652e-08
PhysioNet ECG FFT filter 1: verified, normalized CPU/GPU error 1.17658e-07
PhysioNet ECG FFT filter 2: verified, normalized CPU/GPU error 1.18611e-07
PhysioNet ECG FFT filter 3: verified, normalized CPU/GPU error 8.98478e-08
PhysioNet ECG spectral window 0: verified, normalized CPU/GPU error 3.68661e-06
PhysioNet ECG spectral window 1: verified, normalized CPU/GPU error 7.06065e-06
PhysioNet ECG spectral window 2: verified, normalized CPU/GPU error 6.78297e-06
PhysioNet ECG spectral window 3: verified, normalized CPU/GPU error 1.05561e-05
PhysioNet ECG FIR filter 0: verified, normalized CPU/GPU error 6.89868e-08
PhysioNet ECG FIR filter 1: verified, normalized CPU/GPU error 6.79252e-08
PhysioNet ECG FIR filter 2: verified, normalized CPU/GPU error 6.75995e-08
PhysioNet ECG FIR filter 3: verified, normalized CPU/GPU error 7.03995e-08
PhysioNet ECG FIR filter 4: verified, normalized CPU/GPU error 9.13858e-08
PhysioNet ECG biquad filter 0: verified, normalized CPU/GPU error 1.18087e-07
PhysioNet ECG biquad filter 1: verified, normalized CPU/GPU error 1.38721e-07
PhysioNet ECG biquad filter 2: verified, normalized CPU/GPU error 4.99365e-08
PhysioNet ECG biquad filter 3: verified, normalized CPU/GPU error 8.81661e-08
Prophesee real EVT3 decode/normal-flow pipeline: verified, normalized CPU/GPU error 2.89683e-07
Prophesee recording: 21950 decoded events, 21950 compact CD words, 500 accepted normal-flow fits
MNIST two-feature ANN training/inference: verified, normalized CPU/GPU error 2.36301e-07
MNIST two-feature ANN mini-batch loss: 0.694159 -> 0.20248
MNIST compact CNN training/inference: verified, normalized CPU/GPU error 1.14864e-07
MNIST compact CNN mini-batch loss: 0.699285 -> 0.446509
MNIST patch-token transformer training/inference: verified, normalized CPU/GPU error 9.5935e-08
MNIST patch-token transformer mini-batch loss: 0.61123 -> 0.360663
LeNet-style classifier: all seven layers and 61706 parameters verified against a double oracle
LeNet MNIST held-out: 9778/10000 correct, accuracy 97.78%, normalized reference error 4.57699e-06, completed seconds 41.5901
All public real-data functional cases passed

Every registered test outcome

The table below matches cases by test name. A skipped context records a requested graphics profile outside the tested driver configuration. Iris Xe has 108 registered cases. Ryzen and Raspberry Pi also include the optional glslang check, for 109 cases.

Final frozen runtime and integration test totals
Configuration Passed Skipped Failed Registered cases
Ryzen 5900X / llvmpipe 92 17 0 109
Core i7-1260P / Iris Xe 108 0 0 108
Raspberry Pi 5 / V3D 41 68 0 109
Passed, skipped and failed registered test cases on all three platforms
The counts refer to registered CTest cases. Some cases contain many individual assertions or compiler checks. The 250 shader/profile combinations are counted separately below. Download SVG
Profile coverage differs by driver. Iris Xe executed ES 3.1, ES 3.2, GL 4.3, GL 4.5 and GL 4.6 requests. llvmpipe executed the first four and skipped GL 4.6. The tested V3D driver executed ES 3.1 and skipped ES 3.2 and desktop GL requests.

Math checks cover vector and matrix operations, integer and packing functions, and alias-safe swizzles. Runtime checks cover resource authority, linked dispatches, composition, snapshots and error handling. GPU checks cover numerical parity and context/resource lifetimes. Async checks cover bounded CPU queues, independent snapshots, nonblocking fence polling, and explicit waits. Shader examples include independent references and recorded edge cases.

Named CTest records

All 109 named case records with status and duration
Exact registered test names and reported durations
Test name Ryzen result Ryzen seconds Iris Xe result Iris Xe seconds Pi 5 result Pi 5 seconds
example_pipeline_cpu Passed 0.00 Passed 0.00 Passed 0.00
example_pipeline_es31 Passed 0.08 Passed 0.04 Passed 0.08
example_pipeline_es32 Passed 0.07 Passed 0.03 Skipped 0.05
example_pipeline_gl43 Passed 0.07 Passed 0.03 Skipped 0.05
example_pipeline_gl45 Passed 0.06 Passed 0.03 Skipped 0.05
example_pipeline_gl46 Skipped 0.03 Passed 0.02 Skipped 0.05
example_images_cpu Passed 0.00 Passed 0.00 Passed 0.00
example_images_es31 Passed 0.06 Passed 0.03 Passed 0.08
example_images_es32 Passed 0.07 Passed 0.03 Skipped 0.05
example_images_gl43 Passed 0.08 Passed 0.03 Skipped 0.05
example_images_gl45 Passed 0.08 Passed 0.03 Skipped 0.06
example_images_gl46 Skipped 0.03 Passed 0.04 Skipped 0.05
example_hybrid_cpu Passed 0.00 Passed 0.00 Passed 0.00
example_hybrid_es31 Passed 0.06 Passed 0.03 Passed 0.08
example_hybrid_es32 Passed 0.07 Passed 0.03 Skipped 0.05
example_hybrid_gl43 Passed 0.06 Passed 0.03 Skipped 0.05
example_hybrid_gl45 Passed 0.07 Passed 0.03 Skipped 0.05
example_hybrid_gl46 Skipped 0.03 Passed 0.03 Skipped 0.05
example_async_cpu Passed 0.00 Passed 0.00 Passed 0.00
example_async_es31 Passed 0.07 Passed 0.03 Passed 0.08
example_async_es32 Passed 0.07 Passed 0.03 Skipped 0.05
example_async_gl43 Passed 0.07 Passed 0.03 Skipped 0.05
example_async_gl45 Passed 0.07 Passed 0.03 Skipped 0.05
example_async_gl46 Skipped 0.03 Passed 0.03 Skipped 0.05
example_spectral_cpu Passed 0.00 Passed 0.00 Passed 0.00
example_spectral_es31 Passed 0.08 Passed 0.03 Passed 0.08
example_spectral_es32 Passed 0.08 Passed 0.03 Skipped 0.05
example_spectral_gl43 Passed 0.08 Passed 0.03 Skipped 0.05
example_spectral_gl45 Passed 0.07 Passed 0.03 Skipped 0.05
example_spectral_gl46 Skipped 0.03 Passed 0.04 Skipped 0.05
example_vision_cpu Passed 0.00 Passed 0.00 Passed 0.00
example_vision_es31 Passed 0.08 Passed 0.03 Passed 0.08
example_vision_es32 Passed 0.09 Passed 0.03 Skipped 0.05
example_vision_gl43 Passed 0.08 Passed 0.03 Skipped 0.05
example_vision_gl45 Passed 0.08 Passed 0.03 Skipped 0.05
example_vision_gl46 Skipped 0.04 Passed 0.04 Skipped 0.05
example_imaging_cpu Passed 0.00 Passed 0.00 Passed 0.00
example_imaging_es31 Passed 0.08 Passed 0.04 Passed 0.08
example_imaging_es32 Passed 0.08 Passed 0.03 Skipped 0.05
example_imaging_gl43 Passed 0.08 Passed 0.04 Skipped 0.05
example_imaging_gl45 Passed 0.09 Passed 0.04 Skipped 0.05
example_imaging_gl46 Skipped 0.03 Passed 0.04 Skipped 0.05
example_evt3_cpu Passed 0.01 Passed 0.01 Passed 0.02
example_evt3_es31 Passed 1.97 Passed 0.83 Passed 0.61
example_evt3_es32 Passed 2.04 Passed 0.82 Skipped 0.05
example_evt3_gl43 Passed 2.05 Passed 0.80 Skipped 0.05
example_evt3_gl45 Passed 2.01 Passed 0.82 Skipped 0.05
example_evt3_gl46 Skipped 0.03 Passed 0.80 Skipped 0.05
example_neural_cpu Passed 0.02 Passed 0.02 Passed 0.03
example_neural_es31 Passed 0.17 Passed 0.06 Passed 0.12
example_neural_es32 Passed 0.17 Passed 0.05 Skipped 0.05
example_neural_gl43 Passed 0.16 Passed 0.05 Skipped 0.05
example_neural_gl45 Passed 0.17 Passed 0.05 Skipped 0.05
example_neural_gl46 Skipped 0.03 Passed 0.05 Skipped 0.05
example_cnn_cpu Passed 0.32 Passed 0.20 Passed 0.59
example_cnn_es31 Passed 0.45 Passed 0.30 Passed 0.80
example_cnn_es32 Passed 0.60 Passed 0.29 Skipped 0.05
example_cnn_gl43 Passed 0.54 Passed 0.27 Skipped 0.05
example_cnn_gl45 Passed 0.58 Passed 0.27 Skipped 0.05
example_cnn_gl46 Skipped 0.04 Passed 0.29 Skipped 0.05
example_transformer_cpu Passed 0.06 Passed 0.04 Passed 0.09
example_transformer_es31 Passed 0.18 Passed 0.12 Passed 0.21
example_transformer_es32 Passed 0.23 Passed 0.12 Skipped 0.05
example_transformer_gl43 Passed 0.21 Passed 0.12 Skipped 0.05
example_transformer_gl45 Passed 0.16 Passed 0.12 Skipped 0.05
example_transformer_gl46 Skipped 0.04 Passed 0.12 Skipped 0.05
example_signal_cpu Passed 0.02 Passed 0.01 Passed 0.03
example_signal_es31 Passed 0.40 Passed 0.13 Passed 0.20
example_signal_es32 Passed 0.38 Passed 0.13 Skipped 0.05
example_signal_gl43 Passed 0.39 Passed 0.12 Skipped 0.05
example_signal_gl45 Passed 0.37 Passed 0.13 Skipped 0.05
example_signal_gl46 Skipped 0.04 Passed 0.12 Skipped 0.05
example_dynamic_cpu Passed 0.00 Passed 0.00 Passed 0.00
example_dynamic_es31 Passed 0.10 Passed 0.03 Passed 0.08
example_dynamic_es32 Passed 0.09 Passed 0.03 Skipped 0.05
example_dynamic_gl43 Passed 0.07 Passed 0.03 Skipped 0.05
example_dynamic_gl45 Passed 0.07 Passed 0.03 Skipped 0.05
example_dynamic_gl46 Skipped 0.03 Passed 0.03 Skipped 0.05
runtime_cpu Passed 0.00 Passed 0.00 Passed 0.00
async_cpu Passed 0.00 Passed 0.00 Passed 0.00
runtime_es31 Passed 0.14 Passed 0.04 Passed 0.10
runtime_es32 Passed 0.15 Passed 0.04 Skipped 0.05
runtime_gl43 Passed 0.14 Passed 0.04 Skipped 0.05
runtime_gl45 Passed 0.14 Passed 0.04 Skipped 0.05
runtime_gl46 Skipped 0.04 Passed 0.04 Skipped 0.05
test_math_gpu_cpu Passed 0.00 Passed 0.00 Passed 0.00
test_math_gpu_es31 Passed 0.07 Passed 0.03 Passed 0.08
test_math_gpu_es32 Passed 0.07 Passed 0.03 Skipped 0.05
test_math_gpu_gl43 Passed 0.07 Passed 0.03 Skipped 0.05
test_math_gpu_gl45 Passed 0.07 Passed 0.03 Skipped 0.05
test_math_gpu_gl46 Skipped 0.03 Passed 0.03 Skipped 0.05
math_cpu Passed 0.00 Passed 0.00 Passed 0.00
example_benchmark_cpu Passed 0.35 Passed 0.26 Passed 0.70
example_benchmark_es31 Passed 0.83 Passed 0.49 Passed 1.56
example_benchmark_es32 Passed 0.89 Passed 0.48 Skipped 0.05
example_benchmark_gl43 Passed 0.81 Passed 0.48 Skipped 0.05
example_benchmark_gl45 Passed 0.86 Passed 0.47 Skipped 0.05
example_benchmark_gl46 Skipped 0.04 Passed 0.46 Skipped 0.05
example_real_data_cpu Passed 1.48 Passed 0.72 Passed 2.11
example_real_data_es31 Passed 0.82 Passed 0.39 Passed 2.84
example_real_data_es32 Passed 0.77 Passed 0.41 Skipped 0.05
example_real_data_gl43 Passed 0.74 Passed 0.42 Skipped 0.05
example_real_data_gl45 Passed 0.74 Passed 0.42 Skipped 0.05
example_real_data_gl46 Skipped 0.03 Passed 0.43 Skipped 0.05
glsl_compilation Passed 33.10 Not configured — Passed 30.55
header_is_current Passed 0.22 Passed 0.03 Passed 0.05
fixture_integrity Passed 0.31 Passed 0.04 Passed 0.10
shader_frontend Passed 33.38 Passed 8.78 Passed 27.01
transformer_shaders_are_current Passed 0.22 Passed 0.02 Passed 0.05
Complete CTest log · Ryzen 5900X / llvmpipe

The two build-directory lines use <build>. Every test name, status, duration, summary and skip record is preserved.

Internal ctest changing into directory: <build>
Test project <build>
        Start   1: example_pipeline_cpu
  1/109 Test   #1: example_pipeline_cpu ..............   Passed    0.00 sec
        Start   2: example_pipeline_es31
  2/109 Test   #2: example_pipeline_es31 .............   Passed    0.08 sec
        Start   3: example_pipeline_es32
  3/109 Test   #3: example_pipeline_es32 .............   Passed    0.07 sec
        Start   4: example_pipeline_gl43
  4/109 Test   #4: example_pipeline_gl43 .............   Passed    0.07 sec
        Start   5: example_pipeline_gl45
  5/109 Test   #5: example_pipeline_gl45 .............   Passed    0.06 sec
        Start   6: example_pipeline_gl46
  6/109 Test   #6: example_pipeline_gl46 .............***Skipped   0.03 sec
        Start   7: example_images_cpu
  7/109 Test   #7: example_images_cpu ................   Passed    0.00 sec
        Start   8: example_images_es31
  8/109 Test   #8: example_images_es31 ...............   Passed    0.06 sec
        Start   9: example_images_es32
  9/109 Test   #9: example_images_es32 ...............   Passed    0.07 sec
        Start  10: example_images_gl43
 10/109 Test  #10: example_images_gl43 ...............   Passed    0.08 sec
        Start  11: example_images_gl45
 11/109 Test  #11: example_images_gl45 ...............   Passed    0.08 sec
        Start  12: example_images_gl46
 12/109 Test  #12: example_images_gl46 ...............***Skipped   0.03 sec
        Start  13: example_hybrid_cpu
 13/109 Test  #13: example_hybrid_cpu ................   Passed    0.00 sec
        Start  14: example_hybrid_es31
 14/109 Test  #14: example_hybrid_es31 ...............   Passed    0.06 sec
        Start  15: example_hybrid_es32
 15/109 Test  #15: example_hybrid_es32 ...............   Passed    0.07 sec
        Start  16: example_hybrid_gl43
 16/109 Test  #16: example_hybrid_gl43 ...............   Passed    0.06 sec
        Start  17: example_hybrid_gl45
 17/109 Test  #17: example_hybrid_gl45 ...............   Passed    0.07 sec
        Start  18: example_hybrid_gl46
 18/109 Test  #18: example_hybrid_gl46 ...............***Skipped   0.03 sec
        Start  19: example_async_cpu
 19/109 Test  #19: example_async_cpu .................   Passed    0.00 sec
        Start  20: example_async_es31
 20/109 Test  #20: example_async_es31 ................   Passed    0.07 sec
        Start  21: example_async_es32
 21/109 Test  #21: example_async_es32 ................   Passed    0.07 sec
        Start  22: example_async_gl43
 22/109 Test  #22: example_async_gl43 ................   Passed    0.07 sec
        Start  23: example_async_gl45
 23/109 Test  #23: example_async_gl45 ................   Passed    0.07 sec
        Start  24: example_async_gl46
 24/109 Test  #24: example_async_gl46 ................***Skipped   0.03 sec
        Start  25: example_spectral_cpu
 25/109 Test  #25: example_spectral_cpu ..............   Passed    0.00 sec
        Start  26: example_spectral_es31
 26/109 Test  #26: example_spectral_es31 .............   Passed    0.08 sec
        Start  27: example_spectral_es32
 27/109 Test  #27: example_spectral_es32 .............   Passed    0.08 sec
        Start  28: example_spectral_gl43
 28/109 Test  #28: example_spectral_gl43 .............   Passed    0.08 sec
        Start  29: example_spectral_gl45
 29/109 Test  #29: example_spectral_gl45 .............   Passed    0.07 sec
        Start  30: example_spectral_gl46
 30/109 Test  #30: example_spectral_gl46 .............***Skipped   0.03 sec
        Start  31: example_vision_cpu
 31/109 Test  #31: example_vision_cpu ................   Passed    0.00 sec
        Start  32: example_vision_es31
 32/109 Test  #32: example_vision_es31 ...............   Passed    0.08 sec
        Start  33: example_vision_es32
 33/109 Test  #33: example_vision_es32 ...............   Passed    0.09 sec
        Start  34: example_vision_gl43
 34/109 Test  #34: example_vision_gl43 ...............   Passed    0.08 sec
        Start  35: example_vision_gl45
 35/109 Test  #35: example_vision_gl45 ...............   Passed    0.08 sec
        Start  36: example_vision_gl46
 36/109 Test  #36: example_vision_gl46 ...............***Skipped   0.04 sec
        Start  37: example_imaging_cpu
 37/109 Test  #37: example_imaging_cpu ...............   Passed    0.00 sec
        Start  38: example_imaging_es31
 38/109 Test  #38: example_imaging_es31 ..............   Passed    0.08 sec
        Start  39: example_imaging_es32
 39/109 Test  #39: example_imaging_es32 ..............   Passed    0.08 sec
        Start  40: example_imaging_gl43
 40/109 Test  #40: example_imaging_gl43 ..............   Passed    0.08 sec
        Start  41: example_imaging_gl45
 41/109 Test  #41: example_imaging_gl45 ..............   Passed    0.09 sec
        Start  42: example_imaging_gl46
 42/109 Test  #42: example_imaging_gl46 ..............***Skipped   0.03 sec
        Start  43: example_evt3_cpu
 43/109 Test  #43: example_evt3_cpu ..................   Passed    0.01 sec
        Start  44: example_evt3_es31
 44/109 Test  #44: example_evt3_es31 .................   Passed    1.97 sec
        Start  45: example_evt3_es32
 45/109 Test  #45: example_evt3_es32 .................   Passed    2.04 sec
        Start  46: example_evt3_gl43
 46/109 Test  #46: example_evt3_gl43 .................   Passed    2.05 sec
        Start  47: example_evt3_gl45
 47/109 Test  #47: example_evt3_gl45 .................   Passed    2.01 sec
        Start  48: example_evt3_gl46
 48/109 Test  #48: example_evt3_gl46 .................***Skipped   0.03 sec
        Start  49: example_neural_cpu
 49/109 Test  #49: example_neural_cpu ................   Passed    0.02 sec
        Start  50: example_neural_es31
 50/109 Test  #50: example_neural_es31 ...............   Passed    0.17 sec
        Start  51: example_neural_es32
 51/109 Test  #51: example_neural_es32 ...............   Passed    0.17 sec
        Start  52: example_neural_gl43
 52/109 Test  #52: example_neural_gl43 ...............   Passed    0.16 sec
        Start  53: example_neural_gl45
 53/109 Test  #53: example_neural_gl45 ...............   Passed    0.17 sec
        Start  54: example_neural_gl46
 54/109 Test  #54: example_neural_gl46 ...............***Skipped   0.03 sec
        Start  55: example_cnn_cpu
 55/109 Test  #55: example_cnn_cpu ...................   Passed    0.32 sec
        Start  56: example_cnn_es31
 56/109 Test  #56: example_cnn_es31 ..................   Passed    0.45 sec
        Start  57: example_cnn_es32
 57/109 Test  #57: example_cnn_es32 ..................   Passed    0.60 sec
        Start  58: example_cnn_gl43
 58/109 Test  #58: example_cnn_gl43 ..................   Passed    0.54 sec
        Start  59: example_cnn_gl45
 59/109 Test  #59: example_cnn_gl45 ..................   Passed    0.58 sec
        Start  60: example_cnn_gl46
 60/109 Test  #60: example_cnn_gl46 ..................***Skipped   0.04 sec
        Start  61: example_transformer_cpu
 61/109 Test  #61: example_transformer_cpu ...........   Passed    0.06 sec
        Start  62: example_transformer_es31
 62/109 Test  #62: example_transformer_es31 ..........   Passed    0.18 sec
        Start  63: example_transformer_es32
 63/109 Test  #63: example_transformer_es32 ..........   Passed    0.23 sec
        Start  64: example_transformer_gl43
 64/109 Test  #64: example_transformer_gl43 ..........   Passed    0.21 sec
        Start  65: example_transformer_gl45
 65/109 Test  #65: example_transformer_gl45 ..........   Passed    0.16 sec
        Start  66: example_transformer_gl46
 66/109 Test  #66: example_transformer_gl46 ..........***Skipped   0.04 sec
        Start  67: example_signal_cpu
 67/109 Test  #67: example_signal_cpu ................   Passed    0.02 sec
        Start  68: example_signal_es31
 68/109 Test  #68: example_signal_es31 ...............   Passed    0.40 sec
        Start  69: example_signal_es32
 69/109 Test  #69: example_signal_es32 ...............   Passed    0.38 sec
        Start  70: example_signal_gl43
 70/109 Test  #70: example_signal_gl43 ...............   Passed    0.39 sec
        Start  71: example_signal_gl45
 71/109 Test  #71: example_signal_gl45 ...............   Passed    0.37 sec
        Start  72: example_signal_gl46
 72/109 Test  #72: example_signal_gl46 ...............***Skipped   0.04 sec
        Start  73: example_dynamic_cpu
 73/109 Test  #73: example_dynamic_cpu ...............   Passed    0.00 sec
        Start  74: example_dynamic_es31
 74/109 Test  #74: example_dynamic_es31 ..............   Passed    0.10 sec
        Start  75: example_dynamic_es32
 75/109 Test  #75: example_dynamic_es32 ..............   Passed    0.09 sec
        Start  76: example_dynamic_gl43
 76/109 Test  #76: example_dynamic_gl43 ..............   Passed    0.07 sec
        Start  77: example_dynamic_gl45
 77/109 Test  #77: example_dynamic_gl45 ..............   Passed    0.07 sec
        Start  78: example_dynamic_gl46
 78/109 Test  #78: example_dynamic_gl46 ..............***Skipped   0.03 sec
        Start  79: runtime_cpu
 79/109 Test  #79: runtime_cpu .......................   Passed    0.00 sec
        Start  80: async_cpu
 80/109 Test  #80: async_cpu .........................   Passed    0.00 sec
        Start  81: runtime_es31
 81/109 Test  #81: runtime_es31 ......................   Passed    0.14 sec
        Start  82: runtime_es32
 82/109 Test  #82: runtime_es32 ......................   Passed    0.15 sec
        Start  83: runtime_gl43
 83/109 Test  #83: runtime_gl43 ......................   Passed    0.14 sec
        Start  84: runtime_gl45
 84/109 Test  #84: runtime_gl45 ......................   Passed    0.14 sec
        Start  85: runtime_gl46
 85/109 Test  #85: runtime_gl46 ......................***Skipped   0.04 sec
        Start  86: test_math_gpu_cpu
 86/109 Test  #86: test_math_gpu_cpu .................   Passed    0.00 sec
        Start  87: test_math_gpu_es31
 87/109 Test  #87: test_math_gpu_es31 ................   Passed    0.07 sec
        Start  88: test_math_gpu_es32
 88/109 Test  #88: test_math_gpu_es32 ................   Passed    0.07 sec
        Start  89: test_math_gpu_gl43
 89/109 Test  #89: test_math_gpu_gl43 ................   Passed    0.07 sec
        Start  90: test_math_gpu_gl45
 90/109 Test  #90: test_math_gpu_gl45 ................   Passed    0.07 sec
        Start  91: test_math_gpu_gl46
 91/109 Test  #91: test_math_gpu_gl46 ................***Skipped   0.03 sec
        Start  92: math_cpu
 92/109 Test  #92: math_cpu ..........................   Passed    0.00 sec
        Start  93: example_benchmark_cpu
 93/109 Test  #93: example_benchmark_cpu .............   Passed    0.35 sec
        Start  94: example_benchmark_es31
 94/109 Test  #94: example_benchmark_es31 ............   Passed    0.83 sec
        Start  95: example_benchmark_es32
 95/109 Test  #95: example_benchmark_es32 ............   Passed    0.89 sec
        Start  96: example_benchmark_gl43
 96/109 Test  #96: example_benchmark_gl43 ............   Passed    0.81 sec
        Start  97: example_benchmark_gl45
 97/109 Test  #97: example_benchmark_gl45 ............   Passed    0.86 sec
        Start  98: example_benchmark_gl46
 98/109 Test  #98: example_benchmark_gl46 ............***Skipped   0.04 sec
        Start  99: example_real_data_cpu
 99/109 Test  #99: example_real_data_cpu .............   Passed    1.48 sec
        Start 100: example_real_data_es31
100/109 Test #100: example_real_data_es31 ............   Passed    0.82 sec
        Start 101: example_real_data_es32
101/109 Test #101: example_real_data_es32 ............   Passed    0.77 sec
        Start 102: example_real_data_gl43
102/109 Test #102: example_real_data_gl43 ............   Passed    0.74 sec
        Start 103: example_real_data_gl45
103/109 Test #103: example_real_data_gl45 ............   Passed    0.74 sec
        Start 104: example_real_data_gl46
104/109 Test #104: example_real_data_gl46 ............***Skipped   0.03 sec
        Start 105: glsl_compilation
105/109 Test #105: glsl_compilation ..................   Passed   33.10 sec
        Start 106: header_is_current
106/109 Test #106: header_is_current .................   Passed    0.22 sec
        Start 107: fixture_integrity
107/109 Test #107: fixture_integrity .................   Passed    0.31 sec
        Start 108: shader_frontend
108/109 Test #108: shader_frontend ...................   Passed   33.38 sec
        Start 109: transformer_shaders_are_current
109/109 Test #109: transformer_shaders_are_current ...   Passed    0.22 sec

100% tests passed, 0 tests failed out of 109

Total Test time (real) =  93.14 sec

The following tests did not run:
	  6 - example_pipeline_gl46 (Skipped)
	 12 - example_images_gl46 (Skipped)
	 18 - example_hybrid_gl46 (Skipped)
	 24 - example_async_gl46 (Skipped)
	 30 - example_spectral_gl46 (Skipped)
	 36 - example_vision_gl46 (Skipped)
	 42 - example_imaging_gl46 (Skipped)
	 48 - example_evt3_gl46 (Skipped)
	 54 - example_neural_gl46 (Skipped)
	 60 - example_cnn_gl46 (Skipped)
	 66 - example_transformer_gl46 (Skipped)
	 72 - example_signal_gl46 (Skipped)
	 78 - example_dynamic_gl46 (Skipped)
	 85 - runtime_gl46 (Skipped)
	 91 - test_math_gpu_gl46 (Skipped)
	 98 - example_benchmark_gl46 (Skipped)
	104 - example_real_data_gl46 (Skipped)
Complete CTest log · Core i7-1260P / Iris Xe

The two build-directory lines use <build>. Every test name, status, duration, summary and skip record is preserved.

Internal ctest changing into directory: <build>
Test project <build>
        Start   1: example_pipeline_cpu
  1/108 Test   #1: example_pipeline_cpu ..............   Passed    0.00 sec
        Start   2: example_pipeline_es31
  2/108 Test   #2: example_pipeline_es31 .............   Passed    0.04 sec
        Start   3: example_pipeline_es32
  3/108 Test   #3: example_pipeline_es32 .............   Passed    0.03 sec
        Start   4: example_pipeline_gl43
  4/108 Test   #4: example_pipeline_gl43 .............   Passed    0.03 sec
        Start   5: example_pipeline_gl45
  5/108 Test   #5: example_pipeline_gl45 .............   Passed    0.03 sec
        Start   6: example_pipeline_gl46
  6/108 Test   #6: example_pipeline_gl46 .............   Passed    0.02 sec
        Start   7: example_images_cpu
  7/108 Test   #7: example_images_cpu ................   Passed    0.00 sec
        Start   8: example_images_es31
  8/108 Test   #8: example_images_es31 ...............   Passed    0.03 sec
        Start   9: example_images_es32
  9/108 Test   #9: example_images_es32 ...............   Passed    0.03 sec
        Start  10: example_images_gl43
 10/108 Test  #10: example_images_gl43 ...............   Passed    0.03 sec
        Start  11: example_images_gl45
 11/108 Test  #11: example_images_gl45 ...............   Passed    0.03 sec
        Start  12: example_images_gl46
 12/108 Test  #12: example_images_gl46 ...............   Passed    0.04 sec
        Start  13: example_hybrid_cpu
 13/108 Test  #13: example_hybrid_cpu ................   Passed    0.00 sec
        Start  14: example_hybrid_es31
 14/108 Test  #14: example_hybrid_es31 ...............   Passed    0.03 sec
        Start  15: example_hybrid_es32
 15/108 Test  #15: example_hybrid_es32 ...............   Passed    0.03 sec
        Start  16: example_hybrid_gl43
 16/108 Test  #16: example_hybrid_gl43 ...............   Passed    0.03 sec
        Start  17: example_hybrid_gl45
 17/108 Test  #17: example_hybrid_gl45 ...............   Passed    0.03 sec
        Start  18: example_hybrid_gl46
 18/108 Test  #18: example_hybrid_gl46 ...............   Passed    0.03 sec
        Start  19: example_async_cpu
 19/108 Test  #19: example_async_cpu .................   Passed    0.00 sec
        Start  20: example_async_es31
 20/108 Test  #20: example_async_es31 ................   Passed    0.03 sec
        Start  21: example_async_es32
 21/108 Test  #21: example_async_es32 ................   Passed    0.03 sec
        Start  22: example_async_gl43
 22/108 Test  #22: example_async_gl43 ................   Passed    0.03 sec
        Start  23: example_async_gl45
 23/108 Test  #23: example_async_gl45 ................   Passed    0.03 sec
        Start  24: example_async_gl46
 24/108 Test  #24: example_async_gl46 ................   Passed    0.03 sec
        Start  25: example_spectral_cpu
 25/108 Test  #25: example_spectral_cpu ..............   Passed    0.00 sec
        Start  26: example_spectral_es31
 26/108 Test  #26: example_spectral_es31 .............   Passed    0.03 sec
        Start  27: example_spectral_es32
 27/108 Test  #27: example_spectral_es32 .............   Passed    0.03 sec
        Start  28: example_spectral_gl43
 28/108 Test  #28: example_spectral_gl43 .............   Passed    0.03 sec
        Start  29: example_spectral_gl45
 29/108 Test  #29: example_spectral_gl45 .............   Passed    0.03 sec
        Start  30: example_spectral_gl46
 30/108 Test  #30: example_spectral_gl46 .............   Passed    0.04 sec
        Start  31: example_vision_cpu
 31/108 Test  #31: example_vision_cpu ................   Passed    0.00 sec
        Start  32: example_vision_es31
 32/108 Test  #32: example_vision_es31 ...............   Passed    0.03 sec
        Start  33: example_vision_es32
 33/108 Test  #33: example_vision_es32 ...............   Passed    0.03 sec
        Start  34: example_vision_gl43
 34/108 Test  #34: example_vision_gl43 ...............   Passed    0.03 sec
        Start  35: example_vision_gl45
 35/108 Test  #35: example_vision_gl45 ...............   Passed    0.03 sec
        Start  36: example_vision_gl46
 36/108 Test  #36: example_vision_gl46 ...............   Passed    0.04 sec
        Start  37: example_imaging_cpu
 37/108 Test  #37: example_imaging_cpu ...............   Passed    0.00 sec
        Start  38: example_imaging_es31
 38/108 Test  #38: example_imaging_es31 ..............   Passed    0.04 sec
        Start  39: example_imaging_es32
 39/108 Test  #39: example_imaging_es32 ..............   Passed    0.03 sec
        Start  40: example_imaging_gl43
 40/108 Test  #40: example_imaging_gl43 ..............   Passed    0.04 sec
        Start  41: example_imaging_gl45
 41/108 Test  #41: example_imaging_gl45 ..............   Passed    0.04 sec
        Start  42: example_imaging_gl46
 42/108 Test  #42: example_imaging_gl46 ..............   Passed    0.04 sec
        Start  43: example_evt3_cpu
 43/108 Test  #43: example_evt3_cpu ..................   Passed    0.01 sec
        Start  44: example_evt3_es31
 44/108 Test  #44: example_evt3_es31 .................   Passed    0.83 sec
        Start  45: example_evt3_es32
 45/108 Test  #45: example_evt3_es32 .................   Passed    0.82 sec
        Start  46: example_evt3_gl43
 46/108 Test  #46: example_evt3_gl43 .................   Passed    0.80 sec
        Start  47: example_evt3_gl45
 47/108 Test  #47: example_evt3_gl45 .................   Passed    0.82 sec
        Start  48: example_evt3_gl46
 48/108 Test  #48: example_evt3_gl46 .................   Passed    0.80 sec
        Start  49: example_neural_cpu
 49/108 Test  #49: example_neural_cpu ................   Passed    0.02 sec
        Start  50: example_neural_es31
 50/108 Test  #50: example_neural_es31 ...............   Passed    0.06 sec
        Start  51: example_neural_es32
 51/108 Test  #51: example_neural_es32 ...............   Passed    0.05 sec
        Start  52: example_neural_gl43
 52/108 Test  #52: example_neural_gl43 ...............   Passed    0.05 sec
        Start  53: example_neural_gl45
 53/108 Test  #53: example_neural_gl45 ...............   Passed    0.05 sec
        Start  54: example_neural_gl46
 54/108 Test  #54: example_neural_gl46 ...............   Passed    0.05 sec
        Start  55: example_cnn_cpu
 55/108 Test  #55: example_cnn_cpu ...................   Passed    0.20 sec
        Start  56: example_cnn_es31
 56/108 Test  #56: example_cnn_es31 ..................   Passed    0.30 sec
        Start  57: example_cnn_es32
 57/108 Test  #57: example_cnn_es32 ..................   Passed    0.29 sec
        Start  58: example_cnn_gl43
 58/108 Test  #58: example_cnn_gl43 ..................   Passed    0.27 sec
        Start  59: example_cnn_gl45
 59/108 Test  #59: example_cnn_gl45 ..................   Passed    0.27 sec
        Start  60: example_cnn_gl46
 60/108 Test  #60: example_cnn_gl46 ..................   Passed    0.29 sec
        Start  61: example_transformer_cpu
 61/108 Test  #61: example_transformer_cpu ...........   Passed    0.04 sec
        Start  62: example_transformer_es31
 62/108 Test  #62: example_transformer_es31 ..........   Passed    0.12 sec
        Start  63: example_transformer_es32
 63/108 Test  #63: example_transformer_es32 ..........   Passed    0.12 sec
        Start  64: example_transformer_gl43
 64/108 Test  #64: example_transformer_gl43 ..........   Passed    0.12 sec
        Start  65: example_transformer_gl45
 65/108 Test  #65: example_transformer_gl45 ..........   Passed    0.12 sec
        Start  66: example_transformer_gl46
 66/108 Test  #66: example_transformer_gl46 ..........   Passed    0.12 sec
        Start  67: example_signal_cpu
 67/108 Test  #67: example_signal_cpu ................   Passed    0.01 sec
        Start  68: example_signal_es31
 68/108 Test  #68: example_signal_es31 ...............   Passed    0.13 sec
        Start  69: example_signal_es32
 69/108 Test  #69: example_signal_es32 ...............   Passed    0.13 sec
        Start  70: example_signal_gl43
 70/108 Test  #70: example_signal_gl43 ...............   Passed    0.12 sec
        Start  71: example_signal_gl45
 71/108 Test  #71: example_signal_gl45 ...............   Passed    0.13 sec
        Start  72: example_signal_gl46
 72/108 Test  #72: example_signal_gl46 ...............   Passed    0.12 sec
        Start  73: example_dynamic_cpu
 73/108 Test  #73: example_dynamic_cpu ...............   Passed    0.00 sec
        Start  74: example_dynamic_es31
 74/108 Test  #74: example_dynamic_es31 ..............   Passed    0.03 sec
        Start  75: example_dynamic_es32
 75/108 Test  #75: example_dynamic_es32 ..............   Passed    0.03 sec
        Start  76: example_dynamic_gl43
 76/108 Test  #76: example_dynamic_gl43 ..............   Passed    0.03 sec
        Start  77: example_dynamic_gl45
 77/108 Test  #77: example_dynamic_gl45 ..............   Passed    0.03 sec
        Start  78: example_dynamic_gl46
 78/108 Test  #78: example_dynamic_gl46 ..............   Passed    0.03 sec
        Start  79: runtime_cpu
 79/108 Test  #79: runtime_cpu .......................   Passed    0.00 sec
        Start  80: async_cpu
 80/108 Test  #80: async_cpu .........................   Passed    0.00 sec
        Start  81: runtime_es31
 81/108 Test  #81: runtime_es31 ......................   Passed    0.04 sec
        Start  82: runtime_es32
 82/108 Test  #82: runtime_es32 ......................   Passed    0.04 sec
        Start  83: runtime_gl43
 83/108 Test  #83: runtime_gl43 ......................   Passed    0.04 sec
        Start  84: runtime_gl45
 84/108 Test  #84: runtime_gl45 ......................   Passed    0.04 sec
        Start  85: runtime_gl46
 85/108 Test  #85: runtime_gl46 ......................   Passed    0.04 sec
        Start  86: test_math_gpu_cpu
 86/108 Test  #86: test_math_gpu_cpu .................   Passed    0.00 sec
        Start  87: test_math_gpu_es31
 87/108 Test  #87: test_math_gpu_es31 ................   Passed    0.03 sec
        Start  88: test_math_gpu_es32
 88/108 Test  #88: test_math_gpu_es32 ................   Passed    0.03 sec
        Start  89: test_math_gpu_gl43
 89/108 Test  #89: test_math_gpu_gl43 ................   Passed    0.03 sec
        Start  90: test_math_gpu_gl45
 90/108 Test  #90: test_math_gpu_gl45 ................   Passed    0.03 sec
        Start  91: test_math_gpu_gl46
 91/108 Test  #91: test_math_gpu_gl46 ................   Passed    0.03 sec
        Start  92: math_cpu
 92/108 Test  #92: math_cpu ..........................   Passed    0.00 sec
        Start  93: example_benchmark_cpu
 93/108 Test  #93: example_benchmark_cpu .............   Passed    0.26 sec
        Start  94: example_benchmark_es31
 94/108 Test  #94: example_benchmark_es31 ............   Passed    0.49 sec
        Start  95: example_benchmark_es32
 95/108 Test  #95: example_benchmark_es32 ............   Passed    0.48 sec
        Start  96: example_benchmark_gl43
 96/108 Test  #96: example_benchmark_gl43 ............   Passed    0.48 sec
        Start  97: example_benchmark_gl45
 97/108 Test  #97: example_benchmark_gl45 ............   Passed    0.47 sec
        Start  98: example_benchmark_gl46
 98/108 Test  #98: example_benchmark_gl46 ............   Passed    0.46 sec
        Start  99: example_real_data_cpu
 99/108 Test  #99: example_real_data_cpu .............   Passed    0.72 sec
        Start 100: example_real_data_es31
100/108 Test #100: example_real_data_es31 ............   Passed    0.39 sec
        Start 101: example_real_data_es32
101/108 Test #101: example_real_data_es32 ............   Passed    0.41 sec
        Start 102: example_real_data_gl43
102/108 Test #102: example_real_data_gl43 ............   Passed    0.42 sec
        Start 103: example_real_data_gl45
103/108 Test #103: example_real_data_gl45 ............   Passed    0.42 sec
        Start 104: example_real_data_gl46
104/108 Test #104: example_real_data_gl46 ............   Passed    0.43 sec
        Start 105: header_is_current
105/108 Test #105: header_is_current .................   Passed    0.03 sec
        Start 106: fixture_integrity
106/108 Test #106: fixture_integrity .................   Passed    0.04 sec
        Start 107: shader_frontend
107/108 Test #107: shader_frontend ...................   Passed    8.78 sec
        Start 108: transformer_shaders_are_current
108/108 Test #108: transformer_shaders_are_current ...   Passed    0.02 sec

100% tests passed, 0 tests failed out of 108

Total Test time (real) =  23.20 sec
Complete CTest log · Raspberry Pi 5 / V3D

The two build-directory lines use <build>. Every test name, status, duration, summary and skip record is preserved.

Internal ctest changing into directory: <build>
Test project <build>
        Start   1: example_pipeline_cpu
  1/109 Test   #1: example_pipeline_cpu ..............   Passed    0.00 sec
        Start   2: example_pipeline_es31
  2/109 Test   #2: example_pipeline_es31 .............   Passed    0.08 sec
        Start   3: example_pipeline_es32
  3/109 Test   #3: example_pipeline_es32 .............***Skipped   0.05 sec
        Start   4: example_pipeline_gl43
  4/109 Test   #4: example_pipeline_gl43 .............***Skipped   0.05 sec
        Start   5: example_pipeline_gl45
  5/109 Test   #5: example_pipeline_gl45 .............***Skipped   0.05 sec
        Start   6: example_pipeline_gl46
  6/109 Test   #6: example_pipeline_gl46 .............***Skipped   0.05 sec
        Start   7: example_images_cpu
  7/109 Test   #7: example_images_cpu ................   Passed    0.00 sec
        Start   8: example_images_es31
  8/109 Test   #8: example_images_es31 ...............   Passed    0.08 sec
        Start   9: example_images_es32
  9/109 Test   #9: example_images_es32 ...............***Skipped   0.05 sec
        Start  10: example_images_gl43
 10/109 Test  #10: example_images_gl43 ...............***Skipped   0.05 sec
        Start  11: example_images_gl45
 11/109 Test  #11: example_images_gl45 ...............***Skipped   0.06 sec
        Start  12: example_images_gl46
 12/109 Test  #12: example_images_gl46 ...............***Skipped   0.05 sec
        Start  13: example_hybrid_cpu
 13/109 Test  #13: example_hybrid_cpu ................   Passed    0.00 sec
        Start  14: example_hybrid_es31
 14/109 Test  #14: example_hybrid_es31 ...............   Passed    0.08 sec
        Start  15: example_hybrid_es32
 15/109 Test  #15: example_hybrid_es32 ...............***Skipped   0.05 sec
        Start  16: example_hybrid_gl43
 16/109 Test  #16: example_hybrid_gl43 ...............***Skipped   0.05 sec
        Start  17: example_hybrid_gl45
 17/109 Test  #17: example_hybrid_gl45 ...............***Skipped   0.05 sec
        Start  18: example_hybrid_gl46
 18/109 Test  #18: example_hybrid_gl46 ...............***Skipped   0.05 sec
        Start  19: example_async_cpu
 19/109 Test  #19: example_async_cpu .................   Passed    0.00 sec
        Start  20: example_async_es31
 20/109 Test  #20: example_async_es31 ................   Passed    0.08 sec
        Start  21: example_async_es32
 21/109 Test  #21: example_async_es32 ................***Skipped   0.05 sec
        Start  22: example_async_gl43
 22/109 Test  #22: example_async_gl43 ................***Skipped   0.05 sec
        Start  23: example_async_gl45
 23/109 Test  #23: example_async_gl45 ................***Skipped   0.05 sec
        Start  24: example_async_gl46
 24/109 Test  #24: example_async_gl46 ................***Skipped   0.05 sec
        Start  25: example_spectral_cpu
 25/109 Test  #25: example_spectral_cpu ..............   Passed    0.00 sec
        Start  26: example_spectral_es31
 26/109 Test  #26: example_spectral_es31 .............   Passed    0.08 sec
        Start  27: example_spectral_es32
 27/109 Test  #27: example_spectral_es32 .............***Skipped   0.05 sec
        Start  28: example_spectral_gl43
 28/109 Test  #28: example_spectral_gl43 .............***Skipped   0.05 sec
        Start  29: example_spectral_gl45
 29/109 Test  #29: example_spectral_gl45 .............***Skipped   0.05 sec
        Start  30: example_spectral_gl46
 30/109 Test  #30: example_spectral_gl46 .............***Skipped   0.05 sec
        Start  31: example_vision_cpu
 31/109 Test  #31: example_vision_cpu ................   Passed    0.00 sec
        Start  32: example_vision_es31
 32/109 Test  #32: example_vision_es31 ...............   Passed    0.08 sec
        Start  33: example_vision_es32
 33/109 Test  #33: example_vision_es32 ...............***Skipped   0.05 sec
        Start  34: example_vision_gl43
 34/109 Test  #34: example_vision_gl43 ...............***Skipped   0.05 sec
        Start  35: example_vision_gl45
 35/109 Test  #35: example_vision_gl45 ...............***Skipped   0.05 sec
        Start  36: example_vision_gl46
 36/109 Test  #36: example_vision_gl46 ...............***Skipped   0.05 sec
        Start  37: example_imaging_cpu
 37/109 Test  #37: example_imaging_cpu ...............   Passed    0.00 sec
        Start  38: example_imaging_es31
 38/109 Test  #38: example_imaging_es31 ..............   Passed    0.08 sec
        Start  39: example_imaging_es32
 39/109 Test  #39: example_imaging_es32 ..............***Skipped   0.05 sec
        Start  40: example_imaging_gl43
 40/109 Test  #40: example_imaging_gl43 ..............***Skipped   0.05 sec
        Start  41: example_imaging_gl45
 41/109 Test  #41: example_imaging_gl45 ..............***Skipped   0.05 sec
        Start  42: example_imaging_gl46
 42/109 Test  #42: example_imaging_gl46 ..............***Skipped   0.05 sec
        Start  43: example_evt3_cpu
 43/109 Test  #43: example_evt3_cpu ..................   Passed    0.02 sec
        Start  44: example_evt3_es31
 44/109 Test  #44: example_evt3_es31 .................   Passed    0.61 sec
        Start  45: example_evt3_es32
 45/109 Test  #45: example_evt3_es32 .................***Skipped   0.05 sec
        Start  46: example_evt3_gl43
 46/109 Test  #46: example_evt3_gl43 .................***Skipped   0.05 sec
        Start  47: example_evt3_gl45
 47/109 Test  #47: example_evt3_gl45 .................***Skipped   0.05 sec
        Start  48: example_evt3_gl46
 48/109 Test  #48: example_evt3_gl46 .................***Skipped   0.05 sec
        Start  49: example_neural_cpu
 49/109 Test  #49: example_neural_cpu ................   Passed    0.03 sec
        Start  50: example_neural_es31
 50/109 Test  #50: example_neural_es31 ...............   Passed    0.12 sec
        Start  51: example_neural_es32
 51/109 Test  #51: example_neural_es32 ...............***Skipped   0.05 sec
        Start  52: example_neural_gl43
 52/109 Test  #52: example_neural_gl43 ...............***Skipped   0.05 sec
        Start  53: example_neural_gl45
 53/109 Test  #53: example_neural_gl45 ...............***Skipped   0.05 sec
        Start  54: example_neural_gl46
 54/109 Test  #54: example_neural_gl46 ...............***Skipped   0.05 sec
        Start  55: example_cnn_cpu
 55/109 Test  #55: example_cnn_cpu ...................   Passed    0.59 sec
        Start  56: example_cnn_es31
 56/109 Test  #56: example_cnn_es31 ..................   Passed    0.80 sec
        Start  57: example_cnn_es32
 57/109 Test  #57: example_cnn_es32 ..................***Skipped   0.05 sec
        Start  58: example_cnn_gl43
 58/109 Test  #58: example_cnn_gl43 ..................***Skipped   0.05 sec
        Start  59: example_cnn_gl45
 59/109 Test  #59: example_cnn_gl45 ..................***Skipped   0.05 sec
        Start  60: example_cnn_gl46
 60/109 Test  #60: example_cnn_gl46 ..................***Skipped   0.05 sec
        Start  61: example_transformer_cpu
 61/109 Test  #61: example_transformer_cpu ...........   Passed    0.09 sec
        Start  62: example_transformer_es31
 62/109 Test  #62: example_transformer_es31 ..........   Passed    0.21 sec
        Start  63: example_transformer_es32
 63/109 Test  #63: example_transformer_es32 ..........***Skipped   0.05 sec
        Start  64: example_transformer_gl43
 64/109 Test  #64: example_transformer_gl43 ..........***Skipped   0.05 sec
        Start  65: example_transformer_gl45
 65/109 Test  #65: example_transformer_gl45 ..........***Skipped   0.05 sec
        Start  66: example_transformer_gl46
 66/109 Test  #66: example_transformer_gl46 ..........***Skipped   0.05 sec
        Start  67: example_signal_cpu
 67/109 Test  #67: example_signal_cpu ................   Passed    0.03 sec
        Start  68: example_signal_es31
 68/109 Test  #68: example_signal_es31 ...............   Passed    0.20 sec
        Start  69: example_signal_es32
 69/109 Test  #69: example_signal_es32 ...............***Skipped   0.05 sec
        Start  70: example_signal_gl43
 70/109 Test  #70: example_signal_gl43 ...............***Skipped   0.05 sec
        Start  71: example_signal_gl45
 71/109 Test  #71: example_signal_gl45 ...............***Skipped   0.05 sec
        Start  72: example_signal_gl46
 72/109 Test  #72: example_signal_gl46 ...............***Skipped   0.05 sec
        Start  73: example_dynamic_cpu
 73/109 Test  #73: example_dynamic_cpu ...............   Passed    0.00 sec
        Start  74: example_dynamic_es31
 74/109 Test  #74: example_dynamic_es31 ..............   Passed    0.08 sec
        Start  75: example_dynamic_es32
 75/109 Test  #75: example_dynamic_es32 ..............***Skipped   0.05 sec
        Start  76: example_dynamic_gl43
 76/109 Test  #76: example_dynamic_gl43 ..............***Skipped   0.05 sec
        Start  77: example_dynamic_gl45
 77/109 Test  #77: example_dynamic_gl45 ..............***Skipped   0.05 sec
        Start  78: example_dynamic_gl46
 78/109 Test  #78: example_dynamic_gl46 ..............***Skipped   0.05 sec
        Start  79: runtime_cpu
 79/109 Test  #79: runtime_cpu .......................   Passed    0.00 sec
        Start  80: async_cpu
 80/109 Test  #80: async_cpu .........................   Passed    0.00 sec
        Start  81: runtime_es31
 81/109 Test  #81: runtime_es31 ......................   Passed    0.10 sec
        Start  82: runtime_es32
 82/109 Test  #82: runtime_es32 ......................***Skipped   0.05 sec
        Start  83: runtime_gl43
 83/109 Test  #83: runtime_gl43 ......................***Skipped   0.05 sec
        Start  84: runtime_gl45
 84/109 Test  #84: runtime_gl45 ......................***Skipped   0.05 sec
        Start  85: runtime_gl46
 85/109 Test  #85: runtime_gl46 ......................***Skipped   0.05 sec
        Start  86: test_math_gpu_cpu
 86/109 Test  #86: test_math_gpu_cpu .................   Passed    0.00 sec
        Start  87: test_math_gpu_es31
 87/109 Test  #87: test_math_gpu_es31 ................   Passed    0.08 sec
        Start  88: test_math_gpu_es32
 88/109 Test  #88: test_math_gpu_es32 ................***Skipped   0.05 sec
        Start  89: test_math_gpu_gl43
 89/109 Test  #89: test_math_gpu_gl43 ................***Skipped   0.05 sec
        Start  90: test_math_gpu_gl45
 90/109 Test  #90: test_math_gpu_gl45 ................***Skipped   0.05 sec
        Start  91: test_math_gpu_gl46
 91/109 Test  #91: test_math_gpu_gl46 ................***Skipped   0.05 sec
        Start  92: math_cpu
 92/109 Test  #92: math_cpu ..........................   Passed    0.00 sec
        Start  93: example_benchmark_cpu
 93/109 Test  #93: example_benchmark_cpu .............   Passed    0.70 sec
        Start  94: example_benchmark_es31
 94/109 Test  #94: example_benchmark_es31 ............   Passed    1.56 sec
        Start  95: example_benchmark_es32
 95/109 Test  #95: example_benchmark_es32 ............***Skipped   0.05 sec
        Start  96: example_benchmark_gl43
 96/109 Test  #96: example_benchmark_gl43 ............***Skipped   0.05 sec
        Start  97: example_benchmark_gl45
 97/109 Test  #97: example_benchmark_gl45 ............***Skipped   0.05 sec
        Start  98: example_benchmark_gl46
 98/109 Test  #98: example_benchmark_gl46 ............***Skipped   0.05 sec
        Start  99: example_real_data_cpu
 99/109 Test  #99: example_real_data_cpu .............   Passed    2.11 sec
        Start 100: example_real_data_es31
100/109 Test #100: example_real_data_es31 ............   Passed    2.84 sec
        Start 101: example_real_data_es32
101/109 Test #101: example_real_data_es32 ............***Skipped   0.05 sec
        Start 102: example_real_data_gl43
102/109 Test #102: example_real_data_gl43 ............***Skipped   0.05 sec
        Start 103: example_real_data_gl45
103/109 Test #103: example_real_data_gl45 ............***Skipped   0.05 sec
        Start 104: example_real_data_gl46
104/109 Test #104: example_real_data_gl46 ............***Skipped   0.05 sec
        Start 105: glsl_compilation
105/109 Test #105: glsl_compilation ..................   Passed   30.55 sec
        Start 106: header_is_current
106/109 Test #106: header_is_current .................   Passed    0.05 sec
        Start 107: fixture_integrity
107/109 Test #107: fixture_integrity .................   Passed    0.10 sec
        Start 108: shader_frontend
108/109 Test #108: shader_frontend ...................   Passed   27.01 sec
        Start 109: transformer_shaders_are_current
109/109 Test #109: transformer_shaders_are_current ...   Passed    0.05 sec

100% tests passed, 0 tests failed out of 109

Total Test time (real) =  71.98 sec

The following tests did not run:
	  3 - example_pipeline_es32 (Skipped)
	  4 - example_pipeline_gl43 (Skipped)
	  5 - example_pipeline_gl45 (Skipped)
	  6 - example_pipeline_gl46 (Skipped)
	  9 - example_images_es32 (Skipped)
	 10 - example_images_gl43 (Skipped)
	 11 - example_images_gl45 (Skipped)
	 12 - example_images_gl46 (Skipped)
	 15 - example_hybrid_es32 (Skipped)
	 16 - example_hybrid_gl43 (Skipped)
	 17 - example_hybrid_gl45 (Skipped)
	 18 - example_hybrid_gl46 (Skipped)
	 21 - example_async_es32 (Skipped)
	 22 - example_async_gl43 (Skipped)
	 23 - example_async_gl45 (Skipped)
	 24 - example_async_gl46 (Skipped)
	 27 - example_spectral_es32 (Skipped)
	 28 - example_spectral_gl43 (Skipped)
	 29 - example_spectral_gl45 (Skipped)
	 30 - example_spectral_gl46 (Skipped)
	 33 - example_vision_es32 (Skipped)
	 34 - example_vision_gl43 (Skipped)
	 35 - example_vision_gl45 (Skipped)
	 36 - example_vision_gl46 (Skipped)
	 39 - example_imaging_es32 (Skipped)
	 40 - example_imaging_gl43 (Skipped)
	 41 - example_imaging_gl45 (Skipped)
	 42 - example_imaging_gl46 (Skipped)
	 45 - example_evt3_es32 (Skipped)
	 46 - example_evt3_gl43 (Skipped)
	 47 - example_evt3_gl45 (Skipped)
	 48 - example_evt3_gl46 (Skipped)
	 51 - example_neural_es32 (Skipped)
	 52 - example_neural_gl43 (Skipped)
	 53 - example_neural_gl45 (Skipped)
	 54 - example_neural_gl46 (Skipped)
	 57 - example_cnn_es32 (Skipped)
	 58 - example_cnn_gl43 (Skipped)
	 59 - example_cnn_gl45 (Skipped)
	 60 - example_cnn_gl46 (Skipped)
	 63 - example_transformer_es32 (Skipped)
	 64 - example_transformer_gl43 (Skipped)
	 65 - example_transformer_gl45 (Skipped)
	 66 - example_transformer_gl46 (Skipped)
	 69 - example_signal_es32 (Skipped)
	 70 - example_signal_gl43 (Skipped)
	 71 - example_signal_gl45 (Skipped)
	 72 - example_signal_gl46 (Skipped)
	 75 - example_dynamic_es32 (Skipped)
	 76 - example_dynamic_gl43 (Skipped)
	 77 - example_dynamic_gl45 (Skipped)
	 78 - example_dynamic_gl46 (Skipped)
	 82 - runtime_es32 (Skipped)
	 83 - runtime_gl43 (Skipped)
	 84 - runtime_gl45 (Skipped)
	 85 - runtime_gl46 (Skipped)
	 88 - test_math_gpu_es32 (Skipped)
	 89 - test_math_gpu_gl43 (Skipped)
	 90 - test_math_gpu_gl45 (Skipped)
	 91 - test_math_gpu_gl46 (Skipped)
	 95 - example_benchmark_es32 (Skipped)
	 96 - example_benchmark_gl43 (Skipped)
	 97 - example_benchmark_gl45 (Skipped)
	 98 - example_benchmark_gl46 (Skipped)
	101 - example_real_data_es32 (Skipped)
	102 - example_real_data_gl43 (Skipped)
	103 - example_real_data_gl45 (Skipped)
	104 - example_real_data_gl46 (Skipped)

Frontend and offline shader compilation

Eight frontend regression tests passed. They check lexical declaration handling, language-subset diagnostics, generated C++ semantics, numeric promotions across profiles, private global initialization per invocation, fixed arrays and helper parameters, descriptor defaults and push constants, and constant-shadowing behavior.

Frontend regression output · eight tests

The named registered shader_frontend case above runs these regression groups.

........
----------------------------------------------------------------------
Ran 8 tests in 23.273s

OK

All 250 bundled shader/profile compiler checks passed where glslang was configured: 50 native modules × five language profiles. These offline checks establish language compatibility. The runtime records above describe execution on each available graphics context. The complete public collection is a separate 664-record compiler report, with 258 inputs accepted by C++11 and all five generated profiles.

All 50 bundled modules and their five compiler profiles
Aggregate successful offline run resolved by native module and profile
Native module ES 3.10 ES 3.20 GLSL 4.30 GLSL 4.50 GLSL 4.60
affine.cpp Passed Passed Passed Passed Passed
ann_gradient.cpp Passed Passed Passed Passed Passed
ann_infer.cpp Passed Passed Passed Passed Passed
bayer_unpack.cpp Passed Passed Passed Passed Passed
blur.cpp Passed Passed Passed Passed Passed
buffer_to_image.cpp Passed Passed Passed Passed Passed
clamp.cpp Passed Passed Passed Passed Passed
cnn_gradient.cpp Passed Passed Passed Passed Passed
cnn_infer.cpp Passed Passed Passed Passed Passed
convolution2d.cpp Passed Passed Passed Passed Passed
convolution3d.cpp Passed Passed Passed Passed Passed
debayer.cpp Passed Passed Passed Passed Passed
dft_columns.cpp Passed Passed Passed Passed Passed
dft_rows.cpp Passed Passed Passed Passed Passed
evt3_emit.cpp Passed Passed Passed Passed Passed
evt3_flow.cpp Passed Passed Passed Passed Passed
evt3_reset.cpp Passed Passed Passed Passed Passed
evt3_seed.cpp Passed Passed Passed Passed Passed
evt3_summary.cpp Passed Passed Passed Passed Passed
fft_bit_reverse.cpp Passed Passed Passed Passed Passed
fft_butterfly.cpp Passed Passed Passed Passed Passed
gaussian_horizontal.cpp Passed Passed Passed Passed Passed
gaussian_vertical.cpp Passed Passed Passed Passed Passed
harris_response.cpp Passed Passed Passed Passed Passed
harris_tensor.cpp Passed Passed Passed Passed Passed
image_to_buffer.cpp Passed Passed Passed Passed Passed
lenet_conv.cpp Passed Passed Passed Passed Passed
lenet_dense.cpp Passed Passed Passed Passed Passed
lenet_pool.cpp Passed Passed Passed Passed Passed
luminance.cpp Passed Passed Passed Passed Passed
math.cpp Passed Passed Passed Passed Passed
nn_binary_loss.cpp Passed Passed Passed Passed Passed
nn_sgd.cpp Passed Passed Passed Passed Passed
nonmax.cpp Passed Passed Passed Passed Passed
rgba_float_image.cpp Passed Passed Passed Passed Passed
rgba_int_image.cpp Passed Passed Passed Passed Passed
rgba_uint_image.cpp Passed Passed Passed Passed Passed
rle_count.cpp Passed Passed Passed Passed Passed
rle_decode.cpp Passed Passed Passed Passed Passed
rle_encode.cpp Passed Passed Passed Passed Passed
rle_offsets.cpp Passed Passed Passed Passed Passed
scalar_images.cpp Passed Passed Passed Passed Passed
signal_biquad.cpp Passed Passed Passed Passed Passed
signal_fir.cpp Passed Passed Passed Passed Passed
signal_frequency_mask.cpp Passed Passed Passed Passed Passed
signal_window.cpp Passed Passed Passed Passed Passed
sobel.cpp Passed Passed Passed Passed Passed
sum.cpp Passed Passed Passed Passed Passed
transformer_gradient.cpp Passed Passed Passed Passed Passed
transformer_infer.cpp Passed Passed Passed Passed Passed

Actual public inputs

These images and waveforms come from the prepared files used by the examples. The ECG panel labels an independently calculated 65-tap Hamming-sinc low-pass reference and its real Fourier spectrum. The input source, units and transformations are documented in Public datasets.

Actual NASA image, PhysioNet ECG with independent filter and Fourier references, and official digit samples
NASA Blue Marble RGB preview, PhysioNet MIT-BIH record 100 MLII at 360 Hz in millivolts, and the first shipped instance of each MNIST class. FIR filtering is causal with a zero prefix and 40 Hz cutoff. The digit labels come from the official annotations. Credits and source terms are in the data guide. Download SVG

Methods, fingerprints and reproducibility

Complete archived model, dataset and evaluation metadata
schema_version1
measurement_date2026-10-15
header_sha2560555b3742fa5646d094747786a9d90bbda334bf0dfbf35c639b359224f571416
frontend_sha2569aedc85bb057bf22411fc0ae0adc115e57b38640b1fdad4de9838208566d783e
shader_profilees31
model
architectureModern LeNet-5-style with full C3 connectivity, ReLU, fixed average pooling and linear logits
parameters61706
weights_sha256327dd259df961e6ed9b8f9fc82d2cd56b1334ca7525f92403de9c6495d1fe8de
independent_reference_sha256e3623237df15d5671dbb5cfd04c072d58f3d14a31bc04fd6b80c1d29026195e5
dataset
nameOfficial MNIST held-out test split
images10000
image_archive_sha2568d422c7b0a1c1c79245a5bcf07fe86e33eeafee792b84584aec276f5a2dbc4e6
label_archive_sha256f7ae60f92e00ec6debd23a6088c31dbd2371eca3ffa0defaefb259924204aec6
input_scalebyte / 255 with a two-pixel zero border
sample_orderAll original test images in their original order
method
batch_limit128
normalized_errorabs(actual-reference)/(1+abs(reference)), maximum across all 100000 class scores
evaluation_time_scopeBatch construction, input normalization, program setup, native inference, result reads and reference comparisons. Initial dataset file reads are excluded.
resident_benchmark_is_separatetrue
all_public_data_cases_passedtrue
bundled_adapters_unchanged_after_frontend_updatetrue
measurements
  1. configurationamd
    cpu_modelAMD Ryzen 9 5900X
    rendererllvmpipe (LLVM 20.1.2, 256 bits)
    graphics_kindsoftware on CPU
    labelRyzen 5900X / llvmpipe
    cpu
    correct9778
    images10000
    accuracy_percent97.78
    maximum_normalized_logit_error4.57699e-06
    completed_evaluation_seconds17.2063
    gpu
    correct9778
    images10000
    accuracy_percent97.78
    maximum_normalized_logit_error4.57699e-06
    completed_evaluation_seconds1.76238
  2. configurationintel
    cpu_modelIntel Core i7-1260P
    rendererMesa Intel Iris Xe Graphics (ADL GT2)
    graphics_kindhardware
    labelCore i7-1260P / Iris Xe
    cpu
    correct9778
    images10000
    accuracy_percent97.78
    maximum_normalized_logit_error4.57699e-06
    completed_evaluation_seconds13.8871
    gpu
    correct9778
    images10000
    accuracy_percent97.78
    maximum_normalized_logit_error4.00418e-06
    completed_evaluation_seconds1.95504
  3. configurationpi
    cpu_modelRaspberry Pi 5, four Cortex-A76 cores
    rendererBroadcom V3D 7.1.7.0
    graphics_kindhardware
    labelPi 5 / V3D
    cpu
    correct9778
    images10000
    accuracy_percent97.78
    maximum_normalized_logit_error4.00418e-06
    completed_evaluation_seconds35.0628
    gpu
    correct9778
    images10000
    accuracy_percent97.78
    maximum_normalized_logit_error4.57699e-06
    completed_evaluation_seconds41.5901
measurement_frontend_sha2562273a940ee6c6796660d4507977769cf471bb29ad6010b8178bda0905e0a590a

Measurements were recorded on 15 October 2026. Benchmark means retain the original three-decimal output. Evaluation values retain their printed precision. Exact values accompany each figure.

The figures show the recorded mean timings, exact independent-framework confusion counts, and initial and final training losses. Hardware and software graphics paths are labeled explicitly.

Recorded implementation, model and dataset identities
Record Value
Measured runtime header SHA-256 0555b3742fa5646d094747786a9d90bbda334bf0dfbf35c639b359224f571416
Recorded corpus/frontend SHA-256 9aedc85bb057bf22411fc0ae0adc115e57b38640b1fdad4de9838208566d783e
Timing/evaluation frontend SHA-256 2273a940ee6c6796660d4507977769cf471bb29ad6010b8178bda0905e0a590a
Publication frontend SHA-256 d50c74eee1d0742c96ef2695279c3ac3d4b4035e503bf31912e97a244237ac2c
Model weights SHA-256 327dd259df961e6ed9b8f9fc82d2cd56b1334ca7525f92403de9c6495d1fe8de
Independent reference logits SHA-256 e3623237df15d5671dbb5cfd04c072d58f3d14a31bc04fd6b80c1d29026195e5
Official test-image archive SHA-256 8d422c7b0a1c1c79245a5bcf07fe86e33eeafee792b84584aec276f5a2dbc4e6
Official test-label archive SHA-256 f7ae60f92e00ec6debd23a6088c31dbd2371eca3ffa0defaefb259924204aec6
Input normalization byte / 255 with a two-pixel zero border
Test sample order All original test images in their original order
Evaluation batch limit 128
Normalized error definition abs(actual-reference)/(1+abs(reference)), maximum across all 100000 class scores

Inspect the library and example sources, public input metadata, and corpus compiler provenance for the inputs and tools behind these checks.

schema version: 1

measurement date: 2026-10-15

shader profile: es31