← GPU latency results

gpu/routine_full_command_wall

profile: nvidia-gpuno machine-readable summary

Legacy report: ingestion provenance is unrecorded; no raw artifact is inferred.

Original report

GPU command-wall observations

Wall time includes preparation, validation, sampling and reporting. These are end-to-end workflow costs; operation-performance ratios are not inferred. Missing phase instrumentation is displayed as unavailable, not zero. Filtered or unsuccessful runs do not establish full suite coverage.

gpu/dense / 20261014_005047

Raw run: data/results/nvidia-gpu/gpu/dense/20261014_005047.

tenferro-rs 9655b4e0daced17e0aa35a30d35710b069e10644; harness 25f190048045c01cb363384ba45b2a0b0b640db8. GPU: NVIDIA GeForce RTX 3060 (GPU-a78d5217-eba3-72c2-3d5b-8ae496ebbc2e).

Command through this suite: 238.737 s; collection including build/release delays: 231.161 s; report generation/copy: 0.162 s.

Scope: script clock-read through this suite report copy; devcontainer CLI startup and final wall-record write excluded.

Backend Build s Process s (startup/setup/checks/samples)
tenferro-cuda-trace 67.081 0.423
tenferro-cuda-eager 0.000 24.742
pytorch-cuda 0.000 29.626
jax-cuda 0.000 42.167
cublaslt 0.000 21.659
cutlass 0.000 20.882
cusolver 0.000 9.977

Case phases below exclude process startup/JSONL writing. Preparation includes input generation, transfers and reference preparation; lazy initialization can occur in that phase. Sampler walls include untimed batch preparation/cleanup. Command totals are cumulative across suites in one invocation, so do not sum those totals.

Problem Backend Status Backend init s Prepare s Compile s Probe/init s Warmup s Calibration s Sampling s Validation s
dense_matmul_f64_3072 tenferro-cuda-trace unsupported — — — — — — — —
dense_batched_matmul_f64_b1024_256 tenferro-cuda-trace unsupported — — — — — — — —
dense_qr_f64_1536 tenferro-cuda-trace unsupported — — — — — — — —
dense_solve_f64_512_rhs16 tenferro-cuda-trace unsupported — — — — — — — —
dense_solve_f64_1024_rhs16 tenferro-cuda-trace unsupported — — — — — — — —
dense_solve_f64_2048_rhs16 tenferro-cuda-trace unsupported — — — — — — — —
dense_solve_f64_4096_rhs16 tenferro-cuda-trace unsupported — — — — — — — —
dense_solve_f64_2048_rhs128 tenferro-cuda-trace unsupported — — — — — — — —
dense_svd_f64_256 tenferro-cuda-trace unsupported — — — — — — — —
dense_eigh_f64_1024 tenferro-cuda-trace unsupported — — — — — — — —
dense_batched_matmul_f64_b16_32 tenferro-cuda-trace unsupported — — — — — — — —
dense_matmul_f64_3072 tenferro-cuda-eager ok 0.000 2.336 — 0.342 0.314 0.314 0.941 2.825
dense_batched_matmul_f64_b1024_256 tenferro-cuda-eager ok 0.000 1.982 — 0.200 0.186 0.186 0.559 5.048
dense_qr_f64_1536 tenferro-cuda-eager ok 0.000 0.043 — 0.340 0.096 0.096 0.286 0.624
dense_solve_f64_512_rhs16 tenferro-cuda-eager ok 0.000 0.255 — 0.400 0.004 0.004 0.012 0.002
dense_solve_f64_1024_rhs16 tenferro-cuda-eager ok 0.000 0.051 — 0.015 0.011 0.011 0.033 0.007
dense_solve_f64_2048_rhs16 tenferro-cuda-eager ok 0.000 0.094 — 0.051 0.048 0.048 0.145 0.073
dense_solve_f64_4096_rhs16 tenferro-cuda-eager ok 0.000 0.310 — 0.290 0.288 0.288 0.864 0.251
dense_solve_f64_2048_rhs128 tenferro-cuda-eager ok 0.000 0.074 — 0.057 0.055 0.055 0.164 0.092
dense_svd_f64_256 tenferro-cuda-eager ok 0.000 0.007 — 0.133 0.082 0.082 0.245 0.006
dense_eigh_f64_1024 tenferro-cuda-eager ok 0.000 2.023 — 0.085 0.079 0.079 0.225 0.201
dense_batched_matmul_f64_b16_32 tenferro-cuda-eager ok 0.000 0.007 — 0.000 0.000 0.006 0.009 0.000
dense_matmul_f64_3072 pytorch-cuda ok 1.759 1.763 0.000 0.353 0.314 0.314 0.941 1.055
dense_batched_matmul_f64_b1024_256 pytorch-cuda ok 0.000 7.388 0.000 0.194 0.186 0.187 0.563 5.369
dense_qr_f64_1536 pytorch-cuda ok 0.000 0.078 0.000 0.132 0.101 0.096 0.287 0.847
dense_solve_f64_512_rhs16 pytorch-cuda ok 0.000 0.015 0.000 0.019 0.003 0.003 0.009 0.005
dense_solve_f64_1024_rhs16 pytorch-cuda ok 0.000 0.032 0.000 0.011 0.011 0.011 0.032 0.015
dense_solve_f64_2048_rhs16 pytorch-cuda ok 0.000 0.156 0.000 0.048 0.048 0.048 0.145 0.064
dense_solve_f64_4096_rhs16 pytorch-cuda ok 0.000 0.838 0.000 0.289 0.289 0.289 0.868 0.368
dense_solve_f64_2048_rhs128 pytorch-cuda ok 0.000 0.163 0.000 0.054 0.054 0.054 0.161 0.118
dense_svd_f64_256 pytorch-cuda ok 0.000 0.003 0.000 0.087 0.084 0.084 0.253 0.096
dense_eigh_f64_1024 pytorch-cuda ok 0.000 0.069 0.000 0.085 0.075 0.075 0.225 0.404
dense_batched_matmul_f64_b16_32 pytorch-cuda ok 0.000 0.001 0.000 0.000 0.000 0.008 0.010 0.001
dense_matmul_f64_3072 jax-cuda ok 0.948 0.883 5.769 0.274 0.272 0.272 0.815 1.436
dense_batched_matmul_f64_b1024_256 jax-cuda ok 0.000 7.628 4.296 0.165 0.167 0.161 0.484 5.943
dense_qr_f64_1536 jax-cuda ok 0.000 0.085 0.183 0.109 0.098 0.097 0.291 1.057
dense_solve_f64_512_rhs16 jax-cuda ok 0.000 0.018 0.211 0.009 0.003 0.003 0.010 0.008
dense_solve_f64_1024_rhs16 jax-cuda ok 0.000 0.063 0.241 0.012 0.011 0.011 0.033 0.025
dense_solve_f64_2048_rhs16 jax-cuda ok 0.000 0.260 0.235 0.050 0.049 0.049 0.147 0.129
dense_solve_f64_4096_rhs16 jax-cuda ok 0.000 1.067 0.238 0.291 0.290 0.290 0.870 0.539
dense_solve_f64_2048_rhs128 jax-cuda ok 0.000 0.250 0.213 0.055 0.054 0.054 0.163 0.215
dense_svd_f64_256 jax-cuda ok 0.000 0.004 0.289 0.088 0.083 0.083 0.248 0.102
dense_eigh_f64_1024 jax-cuda ok 0.000 0.107 0.209 0.085 0.075 0.076 0.226 0.458
dense_batched_matmul_f64_b16_32 jax-cuda ok 0.000 0.018 0.390 0.001 0.000 0.006 0.008 0.002
dense_matmul_f64_3072 cublaslt ok 0.273 1.754 0.000 0.356 0.314 0.314 0.941 1.062
dense_batched_matmul_f64_b1024_256 cublaslt ok 0.000 7.374 0.000 0.198 0.186 0.186 0.559 5.398
dense_qr_f64_1536 cublaslt unsupported — — — — — — — —
dense_solve_f64_512_rhs16 cublaslt unsupported — — — — — — — —
dense_solve_f64_1024_rhs16 cublaslt unsupported — — — — — — — —
dense_solve_f64_2048_rhs16 cublaslt unsupported — — — — — — — —
dense_solve_f64_4096_rhs16 cublaslt unsupported — — — — — — — —
dense_solve_f64_2048_rhs128 cublaslt unsupported — — — — — — — —
dense_svd_f64_256 cublaslt unsupported — — — — — — — —
dense_eigh_f64_1024 cublaslt unsupported — — — — — — — —
dense_batched_matmul_f64_b16_32 cublaslt ok 0.000 0.002 0.000 0.000 0.000 0.008 0.009 0.001
dense_matmul_f64_3072 cutlass ok 1.929 1.753 — 0.334 0.320 0.319 0.959 0.939
dense_batched_matmul_f64_b1024_256 cutlass ok 0.000 7.402 — 0.200 0.188 0.188 0.563 4.586
dense_qr_f64_1536 cutlass unsupported — — — — — — — —
dense_solve_f64_512_rhs16 cutlass unsupported — — — — — — — —
dense_solve_f64_1024_rhs16 cutlass unsupported — — — — — — — —
dense_solve_f64_2048_rhs16 cutlass unsupported — — — — — — — —
dense_solve_f64_4096_rhs16 cutlass unsupported — — — — — — — —
dense_solve_f64_2048_rhs128 cutlass unsupported — — — — — — — —
dense_svd_f64_256 cutlass unsupported — — — — — — — —
dense_eigh_f64_1024 cutlass unsupported — — — — — — — —
dense_batched_matmul_f64_b16_32 cutlass ok 0.000 0.002 — 0.000 0.000 0.006 0.009 0.001
dense_matmul_f64_3072 cusolver unsupported — — — — — — — —
dense_batched_matmul_f64_b1024_256 cusolver unsupported — — — — — — — —
dense_qr_f64_1536 cusolver ok 0.268 0.084 0.000 0.160 0.095 0.095 0.284 0.869
dense_solve_f64_512_rhs16 cusolver ok 0.000 0.015 0.000 0.011 0.003 0.003 0.009 0.004
dense_solve_f64_1024_rhs16 cusolver ok 0.000 0.037 0.000 0.011 0.011 0.011 0.032 0.015
dense_solve_f64_2048_rhs16 cusolver ok 0.000 0.158 0.000 0.048 0.048 0.048 0.143 0.064
dense_solve_f64_4096_rhs16 cusolver ok 0.000 0.850 0.000 0.287 0.286 0.286 0.858 0.361
dense_solve_f64_2048_rhs128 cusolver ok 0.000 0.163 0.000 0.053 0.053 0.053 0.160 0.117
dense_svd_f64_256 cusolver ok 0.000 0.003 0.000 0.036 0.027 0.027 0.082 0.039
dense_eigh_f64_1024 cusolver ok 0.000 0.075 0.000 0.084 0.074 0.074 0.223 0.416
dense_batched_matmul_f64_b16_32 cusolver unsupported — — — — — — — —
## gpu/einsum / 20261014_005047

Raw run: data/results/nvidia-gpu/gpu/einsum/20261014_005047.

tenferro-rs 9655b4e0daced17e0aa35a30d35710b069e10644; harness 25f190048045c01cb363384ba45b2a0b0b640db8. GPU: NVIDIA GeForce RTX 3060 (GPU-a78d5217-eba3-72c2-3d5b-8ae496ebbc2e).

Command through this suite: 302.395 s; collection including build/release delays: 56.715 s; report generation/copy: 0.138 s.

Scope: script clock-read through this suite report copy; devcontainer CLI startup and final wall-record write excluded.

Backend Build s Process s (startup/setup/checks/samples)
tenferro-cuda-trace 0.000 0.370
tenferro-cuda-eager 0.000 7.460
pytorch-cuda 0.000 8.355
jax-cuda 0.000 11.648
cublaslt 0.000 8.330
cutlass 0.000 8.026

Case phases below exclude process startup/JSONL writing. Preparation includes input generation, transfers and reference preparation; lazy initialization can occur in that phase. Sampler walls include untimed batch preparation/cleanup. Command totals are cumulative across suites in one invocation, so do not sum those totals.

Problem Backend Status Backend init s Prepare s Compile s Probe/init s Warmup s Calibration s Sampling s Validation s
einsum_bin_matmul_3072_f64 tenferro-cuda-trace unsupported — — — — — — — —
einsum_bin_matmul_3072_f64 tenferro-cuda-eager ok 0.000 2.329 — 0.340 0.312 0.312 0.936 2.761
einsum_bin_matmul_3072_f64 pytorch-cuda ok 1.747 1.768 0.000 0.355 0.312 0.312 0.936 1.971
einsum_bin_matmul_3072_f64 jax-cuda ok 1.005 0.886 5.820 0.271 0.269 0.269 0.806 1.421
einsum_bin_matmul_3072_f64 cublaslt ok 0.238 1.769 0.000 0.349 0.312 0.312 0.937 1.999
einsum_bin_matmul_3072_f64 cutlass ok 1.916 1.766 — 0.332 0.318 0.318 0.954 1.534
## gpu/sparse / 20261014_005047

Raw run: data/results/nvidia-gpu/gpu/sparse/20261014_005047.

tenferro-rs 9655b4e0daced17e0aa35a30d35710b069e10644; harness 25f190048045c01cb363384ba45b2a0b0b640db8. GPU: NVIDIA GeForce RTX 3060 (GPU-a78d5217-eba3-72c2-3d5b-8ae496ebbc2e).

Command through this suite: 548.099 s; collection including build/release delays: 238.680 s; report generation/copy: 0.152 s.

Scope: script clock-read through this suite report copy; devcontainer CLI startup and final wall-record write excluded.

Backend Build s Process s (startup/setup/checks/samples)
pytorch-cuda 0.000 85.345
cusparse 0.000 85.919
ginkgo 0.000 61.116

Case phases below exclude process startup/JSONL writing. Preparation includes input generation, transfers and reference preparation; lazy initialization can occur in that phase. Sampler walls include untimed batch preparation/cleanup. Command totals are cumulative across suites in one invocation, so do not sum those totals.

Problem Backend Status Backend init s Prepare s Compile s Probe/init s Warmup s Calibration s Sampling s Validation s
sparse_synthetic_4m_64m_spmv_f64 pytorch-cuda ok 1.769 63.425 0.000 0.041 0.020 0.020 0.059 0.351
sparse_synthetic_64k_4m_spmm_f64_rhs1024 pytorch-cuda ok 0.000 12.779 0.000 0.122 0.108 0.108 0.325 5.404
sparse_synthetic_4m_64m_spmv_f64 cusparse ok 0.268 63.983 0.000 0.042 0.020 0.020 0.059 0.351
sparse_synthetic_64k_4m_spmm_f64_rhs1024 cusparse ok 0.000 12.826 0.000 0.123 0.108 0.108 0.325 5.390
sparse_synthetic_4m_64m_spmv_f64 ginkgo ok 1.897 33.419 — 0.038 0.021 0.021 0.063 0.250
sparse_synthetic_64k_4m_spmm_f64_rhs1024 ginkgo ok 0.000 11.421 — 1.176 1.175 1.175 3.526 4.934
## gpu/tensornetwork / 20261014_005047

Raw run: data/results/nvidia-gpu/gpu/tensornetwork/20261014_005047.

tenferro-rs 9655b4e0daced17e0aa35a30d35710b069e10644; harness 25f190048045c01cb363384ba45b2a0b0b640db8. GPU: NVIDIA GeForce RTX 3060 (GPU-a78d5217-eba3-72c2-3d5b-8ae496ebbc2e).

Command through this suite: 865.537 s; collection including build/release delays: 310.621 s; report generation/copy: 0.140 s.

Scope: script clock-read through this suite report copy; devcontainer CLI startup and final wall-record write excluded.

Backend Build s Process s (startup/setup/checks/samples)
tenferro-cuda-trace 0.000 0.371
tenferro-cuda-eager 0.000 6.736
pytorch-cuda 0.000 5.210
jax-cuda 0.000 7.723
jax-cuda-tree-jit 0.000 51.554
jax-cuda-eager 0.000 226.511

Case phases below exclude process startup/JSONL writing. Preparation includes input generation, transfers and reference preparation; lazy initialization can occur in that phase. Sampler walls include untimed batch preparation/cleanup. Command totals are cumulative across suites in one invocation, so do not sum those totals.

Problem Backend Status Backend init s Prepare s Compile s Probe/init s Warmup s Calibration s Sampling s Validation s
tensornetwork_permutation_optimized_f32 tenferro-cuda-trace unsupported — — — — — — — —
tensornetwork_permutation_optimized_f32 tenferro-cuda-eager ok 0.000 2.121 — 3.500 0.136 0.122 0.353 0.002
tensornetwork_permutation_optimized_f32 pytorch-cuda ok 1.706 0.045 0.000 0.436 0.337 0.336 1.010 0.341
tensornetwork_permutation_optimized_f32 jax-cuda ok 0.984 0.092 3.344 0.457 0.286 0.286 0.858 0.385
tensornetwork_permutation_optimized_f32 jax-cuda-tree-jit ok 0.952 0.090 46.512 0.380 0.360 0.362 1.088 0.459
tensornetwork_permutation_optimized_f32 jax-cuda-eager ok 0.945 0.106 0.000 217.946 0.820 0.831 2.517 0.946