Files
vllm-omni/benchmarks/edge_harness/results/profile_calib_w8a16.json
T
tzhouamandClaude Opus 5 9264ff81ab Include the baselines, comparisons and study documents, not just the fork
The packaging commit shipped the code and a summary of the numbers. It left out
everything the numbers are measured *against*, which is most of what makes them
checkable.

Added:

  baselines/                 llama.cpp runs (thread sweeps, NUMA sweeps, the
                             interleaved A/B passes behind the headline decode
                             figure) and the quality references: each build
                             against its own bf16 over 12 greedy prompts.
  benchmarks/.../results/    102 artifacts backing the published figures, each
                             with a provenance block naming the model,
                             quantization, thread set and build SHAs.
  docs/engines/              the engine analyses -- llama.cpp, ExecuTorch, MNN,
                             ncnn, vLLM, vLLM-Omni -- and the cross-engine and
                             device-simulation comparisons.
  docs/analysis/             the full study documents the summaries distil.
  docs/edge/comparisons.md   where this fork sits against llama.cpp and why:
                             decode is llama.cpp's, prefill is ours by ~2x, and
                             since neither engine exceeds 39% of the measured
                             222 GB/s ceiling the decode gap is the execution
                             model rather than the kernels.
  docs/edge/environment.md   pins and machine, and the correction below.

On "the modified vllm": there isn't one, and the doc says so plainly because it
is the obvious assumption. The vLLM checkout is clean upstream (0 commits
ahead, 0 uncommitted) and the installed wheel is stock 0.28.0+cpu. This fork
changes vLLM's CPU *behaviour* through vLLM's own extension points at import
time -- three CPU patches in vllm_omni/patch.py, the quantization-config
registry, the IR op registry, platform hooks -- specifically so the changes
survive a wheel upgrade rather than being lost on it. One of those patches is
why "KV sizing is not a memory lever" looked true for months: CpuPlatform
overwrote kv_cache_memory_bytes unconditionally, so the setting had no effect.
There is no patch series to apply; pin the version and let patch.py do the rest.

.gitignore gains explicit exceptions: upstream's blanket `*.json` and `results/`
rules were silently eating the measurement artifacts, which in this fork are the
substance rather than build output. Binary captures stay ignored.

No vLLM or llama.cpp source is redistributed -- only output produced with them.
NOTICE records both.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01L1ax5wYLA6XbqEL1i8L9gZ
2026-09-14 14:01:12 +08:00

1 line
9.6 KiB
JSON

{"execution_summary": {"estimated_inference_time": 803, "estimated_inference_peak_memory": 138977280, "first_load_time": 104796, "first_load_peak_memory": 113455104, "warm_load_time": 105370, "warm_load_peak_memory": 105713664, "compile_time": 0, "compile_peak_memory": 0, "compile_memory_increase_range": [0, 0], "compile_memory_peak_range": [0, 0], "first_load_memory_increase_range": [3270000, 12452832], "first_load_memory_peak_range": [11902976, 21085808], "warm_load_memory_increase_range": [26110560, 36053536], "warm_load_memory_peak_range": [15630336, 25573312], "inference_memory_increase_range": [0, 1065008], "inference_memory_peak_range": [0, 9549376], "all_compile_times": [], "all_first_load_times": [104796], "all_warm_load_times": [105370], "all_inference_times": [1533, 870, 840, 852, 838, 813, 812, 808, 811, 814, 821, 973, 861, 843, 852, 837, 835, 843, 833, 824, 807, 808, 815, 812, 816, 822, 837, 835, 815, 807, 815, 808, 815, 806, 805, 805, 806, 857, 826, 805, 805, 804, 804, 810, 803, 804, 806, 804, 816, 805, 806, 811, 812, 818, 810, 805, 806, 806, 818, 807, 805, 806, 806, 819, 807, 806, 807, 848, 815, 1201, 987, 806, 805, 813, 805, 805, 805, 810, 804, 807, 804, 805, 806, 805, 813, 804, 810, 866, 837, 819, 803, 803, 806, 804, 812, 805, 803, 806, 804, 815]}, "execution_detail": [{"name": "Input", "type": "Input", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 38815}, {"name": "/layers.0/Pow", "type": "/layers.0/Pow", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 54240}, {"name": "/layers.0/ReduceMean", "type": "/layers.0/ReduceMean", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 10302}, {"name": "/layers.0/Add", "type": "/layers.0/Add", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 11317}, {"name": "/layers.0/Sqrt", "type": "/layers.0/Sqrt", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "/layers.0/Div", "type": "/layers.0/Div", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 72747}, {"name": "/layers.0/Mul", "type": "/layers.0/Mul", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 87612}, {"name": "/layers.0/Mul_1", "type": "/layers.0/Mul", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 28026}, {"name": "/layers.0/q_k_v_proj/MatMul", "type": "/layers.0/q", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "/layers.0/q_k_v_proj/MatMul_post_reshape", "type": "/layers.0/q", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "/layers.0/g_proj/MatMul", "type": "/layers.0/g", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 10746}, {"name": "/layers.0/g_proj/MatMul_post_reshape", "type": "/layers.0/g", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "node_Slice_647", "type": "node", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 100071}, {"name": "node_Slice_650", "type": "node", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 20997}, {"name": "node_Slice_653", "type": "node", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 7438}, {"name": "/layers.0/Sigmoid", "type": "/layers.0/Sigmoid", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "/layers.0/Reshape", "type": "/layers.0/Reshape", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "/layers.0/Reshape_1", "type": "/layers.0/Reshape", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 302}, {"name": "/layers.0/Reshape_2", "type": "/layers.0/Reshape", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 6351}, {"name": "/layers.0/Reshape_3", "type": "/layers.0/Reshape", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 16354}, {"name": "node_Slice_1165", "type": "node", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 8720}, {"name": "node_Slice_1168", "type": "node", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 1829}, {"name": "/layers.0/Mul_2", "type": "/layers.0/Mul", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 16759}, {"name": "node_Slice_1388", "type": "node", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 2646}, {"name": "node_Slice_1391", "type": "node", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "/layers.0/Mul_4", "type": "/layers.0/Mul", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 8958}, {"name": "/layers.0/Concat_3", "type": "/layers.0/Concat", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 22033}, {"name": "/layers.0/Neg", "type": "/layers.0/Neg", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 2967}, {"name": "/layers.0/Neg_1", "type": "/layers.0/Neg", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 11087}, {"name": "/layers.0/Concat", "type": "/layers.0/Concat", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 1292}, {"name": "/layers.0/Concat_1", "type": "/layers.0/Concat", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "/layers.0/Mul_3", "type": "/layers.0/Mul", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 11157}, {"name": "/layers.0/Mul_5", "type": "/layers.0/Mul", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 14091}, {"name": "/layers.0/Add_1", "type": "/layers.0/Add", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 3547}, {"name": "/layers.0/Add_2", "type": "/layers.0/Add", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 1063}, {"name": "/layers.0/Reshape_4", "type": "/layers.0/Reshape", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 248}, {"name": "/layers.0/Concat_2", "type": "/layers.0/Concat", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 17585}, {"name": "/layers.0/Transpose", "type": "/layers.0/Transpose", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "/layers.0/MatMul", "type": "/layers.0/MatMul", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 197582}, {"name": "/layers.0/Mul_6", "type": "/layers.0/Mul", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 2115}, {"name": "/layers.0/Add_3", "type": "/layers.0/Add", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 37708}, {"name": "/layers.0/Softmax", "type": "/layers.0/Softmax", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 16817}, {"name": "/layers.0/MatMul_1", "type": "/layers.0/MatMul", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 59003}, {"name": "/layers.0/Reshape_5", "type": "/layers.0/Reshape", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "/layers.0/Mul_7", "type": "/layers.0/Mul", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 21553}, {"name": "/layers.0/out_proj/MatMul", "type": "/layers.0/out", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 43593}, {"name": "/layers.0/out_proj/MatMul_post_reshape", "type": "/layers.0/out", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "/layers.0/Add_4", "type": "/layers.0/Add", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 7006}, {"name": "/layers.0/Pow_1", "type": "/layers.0/Pow", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 11195}, {"name": "/layers.0/ReduceMean_1", "type": "/layers.0/ReduceMean", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 1481}, {"name": "/layers.0/Add_5", "type": "/layers.0/Add", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 186}, {"name": "/layers.0/Sqrt_1", "type": "/layers.0/Sqrt", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "/layers.0/Div_1", "type": "/layers.0/Div", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 58619}, {"name": "/layers.0/Mul_8", "type": "/layers.0/Mul", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 37518}, {"name": "/layers.0/Mul_9", "type": "/layers.0/Mul", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 12218}, {"name": "/layers.0/gate_proj/MatMul", "type": "/layers.0/gate", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "/layers.0/gate_proj/MatMul_post_reshape", "type": "/layers.0/gate", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "/layers.0/up_proj/MatMul", "type": "/layers.0/up", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 456167}, {"name": "/layers.0/up_proj/MatMul_post_reshape", "type": "/layers.0/up", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "_elementwiseneuron_0", "type": "elementwiseneuron", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 427975}, {"name": "/layers.0/Mul_12", "type": "/layers.0/Mul", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 19765}, {"name": "/layers.0/down_proj/MatMul", "type": "/layers.0/down", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 432305}, {"name": "/layers.0/down_proj/MatMul_post_reshape", "type": "/layers.0/down", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "/layers.0/Add_7", "type": "/layers.0/Add", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 5846}, {"name": "/layers.0/Add_2_identity", "type": "/layers.0/Add", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "/layers.0/Reshape_2_identity", "type": "/layers.0/Reshape", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "Output", "type": "Output", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 9779}]}