anodyine/vllm-omni
1
0 Fork 0
Code Issues 4 Pull Requests 2 Actions Packages Projects Releases Wiki Activity
Files
ab2f2a3d658db5be5ec96a950da55cb74b48e7aa
vllm-omni/benchmarks/edge_harness/results/profile_block4_fp16.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
35 KiB
JSON
Raw Blame History

{"execution_summary": {"estimated_inference_time": 6103, "estimated_inference_peak_memory": 154943488, "first_load_time": 173117, "first_load_peak_memory": 113643520, "warm_load_time": 176691, "warm_load_peak_memory": 124940288, "compile_time": 0, "compile_peak_memory": 0, "compile_memory_increase_range": [0, 0], "compile_memory_peak_range": [0, 0], "first_load_memory_increase_range": [0, 2398992], "first_load_memory_peak_range": [1228800, 10888560], "warm_load_memory_increase_range": [0, 2296256], "warm_load_memory_peak_range": [335872, 10494528], "inference_memory_increase_range": [0, 6340656], "inference_memory_peak_range": [6332416, 15690592], "all_compile_times": [], "all_first_load_times": [173117], "all_warm_load_times": [176691], "all_inference_times": [8046, 6381, 6308, 6330, 6422, 6444, 6621, 6554, 6463, 7236, 6207, 6835, 6499, 6191, 6175, 6205, 6177, 6197, 6160, 6560, 6171, 6167, 6156, 6177, 6149, 6226, 6159, 6146, 6144, 6144, 6362, 6186, 6130, 6141, 6193, 6135, 6392, 6330, 6145, 6146, 6141, 6123, 6151, 6191, 7266, 6796, 6240, 6180, 6170, 6179, 6157, 6303, 6285, 6554, 6221, 6585, 6215, 6153, 6157, 6133, 6271, 6167, 6194, 6285, 6137, 6142, 6153, 6253, 6159, 6185, 6195, 6229, 7027, 6126, 6135, 6129, 6135, 6155, 6499, 6235, 6134, 6145, 6112, 6214, 6160, 6168, 6912, 6158, 6122, 6273, 6404, 6111, 6152, 6135, 6123, 6121, 6107, 6123, 6103, 6181]}, "execution_detail": [{"name": "Input", "type": "Input", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 160867}, {"name": "QNN_DATATYPE_FLOAT_32_converted_x_QNN_DATATYPE_FLOAT_16", "type": "QNN", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 135879}, {"name": "rms_norm_/layers.0/Mul_1_output_0", "type": "rms", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 10234}, {"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": "node_Slice_399", "type": "node", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 252656}, {"name": "node_Slice_402", "type": "node", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 59233}, {"name": "node_Slice_405", "type": "node", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 62582}, {"name": "/layers.0/Reshape", "type": "/layers.0/Reshape", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 495}, {"name": "/layers.0/Reshape_1", "type": "/layers.0/Reshape", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 249}, {"name": "/layers.0/Reshape_2", "type": "/layers.0/Reshape", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 267}, {"name": "/layers.0/g_proj/MatMul", "type": "/layers.0/g", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "/layers.0/g_proj/MatMul_post_reshape", "type": "/layers.0/g", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "/layers.0/Sigmoid", "type": "/layers.0/Sigmoid", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 7496}, {"name": "/layers.0/Reshape_3", "type": "/layers.0/Reshape", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 6813}, {"name": "node_Slice_1001", "type": "node", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 5076}, {"name": "node_Slice_1004", "type": "node", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "QNN_DATATYPE_FLOAT_32_converted_cos_sw_QNN_DATATYPE_FLOAT_16", "type": "QNN", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 23170}, {"name": "/layers.0/Mul_2", "type": "/layers.0/Mul", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 1040}, {"name": "/layers.0/Neg", "type": "/layers.0/Neg", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 1180}, {"name": "/layers.0/Concat", "type": "/layers.0/Concat", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "QNN_DATATYPE_FLOAT_32_converted_sin_sw_QNN_DATATYPE_FLOAT_16", "type": "QNN", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 340}, {"name": "/layers.0/Mul_3", "type": "/layers.0/Mul", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 963}, {"name": "/layers.0/Add_1", "type": "/layers.0/Add", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 17162}, {"name": "node_Slice_2566", "type": "node", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "node_Slice_2569", "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": 3367}, {"name": "/layers.0/Neg_1", "type": "/layers.0/Neg", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 18517}, {"name": "/layers.0/Concat_1", "type": "/layers.0/Concat", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "/layers.0/Mul_5", "type": "/layers.0/Mul", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 344}, {"name": "/layers.0/Add_2", "type": "/layers.0/Add", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 5014}, {"name": "/layers.0/Reshape_4", "type": "/layers.0/Reshape", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 225}, {"name": "QNN_DATATYPE_FLOAT_32_converted_k_cache_0_QNN_DATATYPE_FLOAT_16", "type": "QNN", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 116401}, {"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": 456203}, {"name": "/layers.0/Mul_6", "type": "/layers.0/Mul", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 9395}, {"name": "/layers.0/Mul_7", "type": "/layers.0/Mul", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 11870}, {"name": "/layers.0/ReduceSum", "type": "/layers.0/ReduceSum", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 12926}, {"name": "/layers.0/Mul_8", "type": "/layers.0/Mul", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 4368}, {"name": "/layers.0/Concat_2", "type": "/layers.0/Concat", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "QNN_DATATYPE_FLOAT_32_converted_mask_sw_QNN_DATATYPE_FLOAT_16", "type": "QNN", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 1209}, {"name": "/layers.0/Add_3", "type": "/layers.0/Add", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 96928}, {"name": "/layers.0/Softmax", "type": "/layers.0/Softmax", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 22165}, {"name": "node_Slice_1202", "type": "node", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "node_Slice_1205", "type": "node", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 9717}, {"name": "QNN_DATATYPE_FLOAT_32_converted_v_cache_0_QNN_DATATYPE_FLOAT_16", "type": "QNN", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 105429}, {"name": "/layers.0/MatMul_1", "type": "/layers.0/MatMul", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 36403}, {"name": "/layers.0/Mul_9", "type": "/layers.0/Mul", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 3890}, {"name": "/layers.0/Add_4", "type": "/layers.0/Add", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 5853}, {"name": "/layers.0/Reshape_5", "type": "/layers.0/Reshape", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 469}, {"name": "/layers.0/Mul_10", "type": "/layers.0/Mul", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 3245}, {"name": "/layers.0/out_proj/MatMul", "type": "/layers.0/out", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 152121}, {"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_5", "type": "/layers.0/Add", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 70561}, {"name": "rms_norm_/layers.0/Mul_12_output_0", "type": "rms", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 4470}, {"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": "node_Gelu_0", "type": "node", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 804262}, {"name": "/layers.0/up_proj/MatMul", "type": "/layers.0/up", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 788118}, {"name": "/layers.0/up_proj/MatMul_post_reshape", "type": "/layers.0/up", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "/layers.0/Mul_15", "type": "/layers.0/Mul", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 11671}, {"name": "/layers.0/down_proj/MatMul", "type": "/layers.0/down", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 791431}, {"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_8", "type": "/layers.0/Add", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 6921}, {"name": "rms_norm_/layers.1/Mul_1_output_0", "type": "rms", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 3944}, {"name": "/layers.1/q_k_v_proj/MatMul", "type": "/layers.1/q", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "/layers.1/q_k_v_proj/MatMul_post_reshape", "type": "/layers.1/q", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "node_Slice_409", "type": "node", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 251811}, {"name": "node_Slice_412", "type": "node", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 60914}, {"name": "node_Slice_415", "type": "node", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 65435}, {"name": "/layers.1/Reshape", "type": "/layers.1/Reshape", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 526}, {"name": "/layers.1/Reshape_1", "type": "/layers.1/Reshape", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 243}, {"name": "/layers.1/Reshape_2", "type": "/layers.1/Reshape", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 245}, {"name": "/layers.1/g_proj/MatMul", "type": "/layers.1/g", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "/layers.1/g_proj/MatMul_post_reshape", "type": "/layers.1/g", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "/layers.1/Sigmoid", "type": "/layers.1/Sigmoid", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 4023}, {"name": "/layers.1/Reshape_3", "type": "/layers.1/Reshape", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 851}, {"name": "node_Slice_2573", "type": "node", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "node_Slice_2576", "type": "node", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "/layers.1/Mul_2", "type": "/layers.1/Mul", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 1048}, {"name": "/layers.1/Neg", "type": "/layers.1/Neg", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 1204}, {"name": "/layers.1/Concat", "type": "/layers.1/Concat", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "/layers.1/Mul_3", "type": "/layers.1/Mul", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 984}, {"name": "/layers.1/Add_1", "type": "/layers.1/Add", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 5316}, {"name": "node_Slice_2580", "type": "node", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "node_Slice_2583", "type": "node", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "/layers.1/Mul_4", "type": "/layers.1/Mul", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 398}, {"name": "/layers.1/Neg_1", "type": "/layers.1/Neg", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 1112}, {"name": "/layers.1/Concat_1", "type": "/layers.1/Concat", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "/layers.1/Mul_5", "type": "/layers.1/Mul", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 363}, {"name": "/layers.1/Add_2", "type": "/layers.1/Add", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 941}, {"name": "/layers.1/Reshape_4", "type": "/layers.1/Reshape", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 211}, {"name": "QNN_DATATYPE_FLOAT_32_converted_k_cache_1_QNN_DATATYPE_FLOAT_16", "type": "QNN", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 107571}, {"name": "/layers.1/Transpose", "type": "/layers.1/Transpose", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "/layers.1/MatMul", "type": "/layers.1/MatMul", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 248512}, {"name": "/layers.1/Mul_6", "type": "/layers.1/Mul", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 6852}, {"name": "/layers.1/Mul_7", "type": "/layers.1/Mul", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 4133}, {"name": "/layers.1/ReduceSum", "type": "/layers.1/ReduceSum", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 2850}, {"name": "/layers.1/Mul_8", "type": "/layers.1/Mul", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 1773}, {"name": "/layers.1/Concat_2", "type": "/layers.1/Concat", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "/layers.1/Add_3", "type": "/layers.1/Add", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 12370}, {"name": "/layers.1/Softmax", "type": "/layers.1/Softmax", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 5507}, {"name": "node_Slice_1036", "type": "node", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "node_Slice_1039", "type": "node", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 4464}, {"name": "QNN_DATATYPE_FLOAT_32_converted_v_cache_1_QNN_DATATYPE_FLOAT_16", "type": "QNN", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 105004}, {"name": "/layers.1/MatMul_1", "type": "/layers.1/MatMul", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 42048}, {"name": "/layers.1/Mul_9", "type": "/layers.1/Mul", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 4523}, {"name": "/layers.1/Add_4", "type": "/layers.1/Add", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 4362}, {"name": "/layers.1/Reshape_5", "type": "/layers.1/Reshape", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 437}, {"name": "/layers.1/Mul_10", "type": "/layers.1/Mul", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 912}, {"name": "/layers.1/out_proj/MatMul", "type": "/layers.1/out", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 169538}, {"name": "/layers.1/out_proj/MatMul_post_reshape", "type": "/layers.1/out", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "/layers.1/Add_5", "type": "/layers.1/Add", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 6735}, {"name": "rms_norm_/layers.1/Mul_12_output_0", "type": "rms", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 4174}, {"name": "/layers.1/gate_proj/MatMul", "type": "/layers.1/gate", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "/layers.1/gate_proj/MatMul_post_reshape", "type": "/layers.1/gate", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "node_Gelu_1", "type": "node", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 784833}, {"name": "/layers.1/up_proj/MatMul", "type": "/layers.1/up", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 779331}, {"name": "/layers.1/up_proj/MatMul_post_reshape", "type": "/layers.1/up", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "/layers.1/Mul_15", "type": "/layers.1/Mul", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 11480}, {"name": "/layers.1/down_proj/MatMul", "type": "/layers.1/down", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 807496}, {"name": "/layers.1/down_proj/MatMul_post_reshape", "type": "/layers.1/down", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "/layers.1/Add_8", "type": "/layers.1/Add", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 6734}, {"name": "rms_norm_/layers.2/Mul_1_output_0", "type": "rms", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 5632}, {"name": "/layers.2/q_k_v_proj/MatMul", "type": "/layers.2/q", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "/layers.2/q_k_v_proj/MatMul_post_reshape", "type": "/layers.2/q", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "node_Slice_1325", "type": "node", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 250735}, {"name": "node_Slice_1328", "type": "node", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 61151}, {"name": "node_Slice_1331", "type": "node", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 66637}, {"name": "/layers.2/Reshape", "type": "/layers.2/Reshape", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 436}, {"name": "/layers.2/Reshape_1", "type": "/layers.2/Reshape", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 219}, {"name": "/layers.2/Reshape_2", "type": "/layers.2/Reshape", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 215}, {"name": "/layers.2/g_proj/MatMul", "type": "/layers.2/g", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "/layers.2/g_proj/MatMul_post_reshape", "type": "/layers.2/g", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "/layers.2/Sigmoid", "type": "/layers.2/Sigmoid", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 4063}, {"name": "/layers.2/Reshape_3", "type": "/layers.2/Reshape", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 795}, {"name": "node_Slice_357", "type": "node", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "node_Slice_360", "type": "node", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "/layers.2/Mul_2", "type": "/layers.2/Mul", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 1009}, {"name": "/layers.2/Neg", "type": "/layers.2/Neg", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 1089}, {"name": "/layers.2/Concat", "type": "/layers.2/Concat", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "/layers.2/Mul_3", "type": "/layers.2/Mul", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 972}, {"name": "/layers.2/Add_1", "type": "/layers.2/Add", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 5694}, {"name": "node_Slice_364", "type": "node", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "node_Slice_367", "type": "node", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "/layers.2/Mul_4", "type": "/layers.2/Mul", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 384}, {"name": "/layers.2/Neg_1", "type": "/layers.2/Neg", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 1117}, {"name": "/layers.2/Concat_1", "type": "/layers.2/Concat", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "/layers.2/Mul_5", "type": "/layers.2/Mul", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 393}, {"name": "/layers.2/Add_2", "type": "/layers.2/Add", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 878}, {"name": "/layers.2/Reshape_4", "type": "/layers.2/Reshape", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 262}, {"name": "QNN_DATATYPE_FLOAT_32_converted_k_cache_2_QNN_DATATYPE_FLOAT_16", "type": "QNN", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 92625}, {"name": "/layers.2/Transpose", "type": "/layers.2/Transpose", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "/layers.2/MatMul", "type": "/layers.2/MatMul", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 243893}, {"name": "/layers.2/Mul_6", "type": "/layers.2/Mul", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 11304}, {"name": "/layers.2/Mul_7", "type": "/layers.2/Mul", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 4767}, {"name": "/layers.2/ReduceSum", "type": "/layers.2/ReduceSum", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 4344}, {"name": "/layers.2/Mul_8", "type": "/layers.2/Mul", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 1296}, {"name": "/layers.2/Concat_2", "type": "/layers.2/Concat", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "/layers.2/Add_3", "type": "/layers.2/Add", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 11414}, {"name": "/layers.2/Softmax", "type": "/layers.2/Softmax", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 5029}, {"name": "node_Slice_371", "type": "node", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "node_Slice_374", "type": "node", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 5563}, {"name": "QNN_DATATYPE_FLOAT_32_converted_v_cache_2_QNN_DATATYPE_FLOAT_16", "type": "QNN", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 117744}, {"name": "/layers.2/MatMul_1", "type": "/layers.2/MatMul", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 35549}, {"name": "/layers.2/Mul_9", "type": "/layers.2/Mul", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 3757}, {"name": "/layers.2/Add_4", "type": "/layers.2/Add", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 3972}, {"name": "/layers.2/Reshape_5", "type": "/layers.2/Reshape", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 444}, {"name": "/layers.2/Mul_10", "type": "/layers.2/Mul", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 1387}, {"name": "/layers.2/out_proj/MatMul", "type": "/layers.2/out", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 211582}, {"name": "/layers.2/out_proj/MatMul_post_reshape", "type": "/layers.2/out", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "/layers.2/Add_5", "type": "/layers.2/Add", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 7735}, {"name": "rms_norm_/layers.2/Mul_12_output_0", "type": "rms", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 3612}, {"name": "/layers.2/gate_proj/MatMul", "type": "/layers.2/gate", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "/layers.2/gate_proj/MatMul_post_reshape", "type": "/layers.2/gate", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "node_Gelu_2", "type": "node", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 858318}, {"name": "/layers.2/up_proj/MatMul", "type": "/layers.2/up", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 796230}, {"name": "/layers.2/up_proj/MatMul_post_reshape", "type": "/layers.2/up", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "/layers.2/Mul_15", "type": "/layers.2/Mul", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 13558}, {"name": "/layers.2/down_proj/MatMul", "type": "/layers.2/down", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 780322}, {"name": "/layers.2/down_proj/MatMul_post_reshape", "type": "/layers.2/down", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "/layers.2/Add_8", "type": "/layers.2/Add", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 6983}, {"name": "rms_norm_/layers.3/Mul_1_output_0", "type": "rms", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 3544}, {"name": "/layers.3/q_k_v_proj/MatMul", "type": "/layers.3/q", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "/layers.3/q_k_v_proj/MatMul_post_reshape", "type": "/layers.3/q", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "node_Slice_1335", "type": "node", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 252410}, {"name": "node_Slice_1338", "type": "node", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 63173}, {"name": "node_Slice_1341", "type": "node", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 59786}, {"name": "/layers.3/Reshape", "type": "/layers.3/Reshape", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 514}, {"name": "/layers.3/Reshape_1", "type": "/layers.3/Reshape", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 274}, {"name": "/layers.3/Reshape_2", "type": "/layers.3/Reshape", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 214}, {"name": "/layers.3/g_proj/MatMul", "type": "/layers.3/g", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "/layers.3/g_proj/MatMul_post_reshape", "type": "/layers.3/g", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "/layers.3/Sigmoid", "type": "/layers.3/Sigmoid", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 4866}, {"name": "/layers.3/Reshape_3", "type": "/layers.3/Reshape", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 1917}, {"name": "node_Slice_378", "type": "node", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "node_Slice_381", "type": "node", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 763}, {"name": "fused_/layers.3/Slice_3_/layers.3/Slice_5", "type": "fused", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "fused_/layers.3/Slice_3_/layers.3/Slice_6", "type": "fused", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "QNN_DATATYPE_FLOAT_32_converted_cos_full_QNN_DATATYPE_FLOAT_16", "type": "QNN", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 403}, {"name": "/layers.3/Mul_2", "type": "/layers.3/Mul", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 360}, {"name": "/layers.3/Neg", "type": "/layers.3/Neg", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 439}, {"name": "/layers.3/Concat", "type": "/layers.3/Concat", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "QNN_DATATYPE_FLOAT_32_converted_sin_full_QNN_DATATYPE_FLOAT_16", "type": "QNN", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 366}, {"name": "/layers.3/Mul_3", "type": "/layers.3/Mul", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 300}, {"name": "/layers.3/Add_1", "type": "/layers.3/Add", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 878}, {"name": "/layers.3/Concat_1", "type": "/layers.3/Concat", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 2381}, {"name": "node_Slice_385", "type": "node", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "node_Slice_388", "type": "node", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "fused_/layers.3/Slice_7_/layers.3/Slice_9", "type": "fused", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "fused_/layers.3/Slice_7_/layers.3/Slice_10", "type": "fused", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "/layers.3/Mul_4", "type": "/layers.3/Mul", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 225}, {"name": "/layers.3/Neg_1", "type": "/layers.3/Neg", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 597}, {"name": "/layers.3/Concat_2", "type": "/layers.3/Concat", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "/layers.3/Mul_5", "type": "/layers.3/Mul", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 201}, {"name": "/layers.3/Add_2", "type": "/layers.3/Add", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 217}, {"name": "/layers.3/Concat_3", "type": "/layers.3/Concat", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "/layers.3/Reshape_4", "type": "/layers.3/Reshape", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 208}, {"name": "QNN_DATATYPE_FLOAT_32_converted_k_cache_3_QNN_DATATYPE_FLOAT_16", "type": "QNN", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 138545}, {"name": "/layers.3/Transpose", "type": "/layers.3/Transpose", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "/layers.3/MatMul", "type": "/layers.3/MatMul", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 450217}, {"name": "/layers.3/Mul_6", "type": "/layers.3/Mul", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 14017}, {"name": "/layers.3/Mul_7", "type": "/layers.3/Mul", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 3134}, {"name": "/layers.3/ReduceSum", "type": "/layers.3/ReduceSum", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 1648}, {"name": "/layers.3/Mul_8", "type": "/layers.3/Mul", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 534}, {"name": "/layers.3/Concat_4", "type": "/layers.3/Concat", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "QNN_DATATYPE_FLOAT_32_converted_mask_full_QNN_DATATYPE_FLOAT_16", "type": "QNN", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 6297}, {"name": "/layers.3/Add_3", "type": "/layers.3/Add", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 20477}, {"name": "/layers.3/Softmax", "type": "/layers.3/Softmax", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 6507}, {"name": "node_Slice_392", "type": "node", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "node_Slice_395", "type": "node", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "QNN_DATATYPE_FLOAT_32_converted_v_cache_3_QNN_DATATYPE_FLOAT_16", "type": "QNN", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 170588}, {"name": "/layers.3/MatMul_1", "type": "/layers.3/MatMul", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 170262}, {"name": "/layers.3/Mul_9", "type": "/layers.3/Mul", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 3684}, {"name": "/layers.3/Add_4", "type": "/layers.3/Add", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 7606}, {"name": "/layers.3/Reshape_5", "type": "/layers.3/Reshape", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 441}, {"name": "/layers.3/Mul_10", "type": "/layers.3/Mul", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 924}, {"name": "/layers.3/out_proj/MatMul", "type": "/layers.3/out", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 193264}, {"name": "/layers.3/out_proj/MatMul_post_reshape", "type": "/layers.3/out", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "/layers.3/Add_5", "type": "/layers.3/Add", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 6918}, {"name": "rms_norm_/layers.3/Mul_12_output_0", "type": "rms", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 3566}, {"name": "/layers.3/gate_proj/MatMul", "type": "/layers.3/gate", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "/layers.3/gate_proj/MatMul_post_reshape", "type": "/layers.3/gate", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "node_Gelu_3", "type": "node", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 792873}, {"name": "/layers.3/up_proj/MatMul", "type": "/layers.3/up", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 780398}, {"name": "/layers.3/up_proj/MatMul_post_reshape", "type": "/layers.3/up", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "/layers.3/Mul_15", "type": "/layers.3/Mul", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 16344}, {"name": "/layers.3/down_proj/MatMul", "type": "/layers.3/down", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 789082}, {"name": "/layers.3/down_proj/MatMul_post_reshape", "type": "/layers.3/down", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "/layers.3/Add_8", "type": "/layers.3/Add", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 6894}, {"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": "/layers.1/Add_2_identity", "type": "/layers.1/Add", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "/layers.1/Reshape_2_identity", "type": "/layers.1/Reshape", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "/layers.2/Add_2_identity", "type": "/layers.2/Add", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "/layers.2/Reshape_2_identity", "type": "/layers.2/Reshape", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "/layers.3/Concat_3_identity", "type": "/layers.3/Concat", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "/layers.3/Reshape_2_identity", "type": "/layers.3/Reshape", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "QNN_DATATYPE_FLOAT_16_converted_output_0_QNN_DATATYPE_FLOAT_32", "type": "QNN", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "QNN_DATATYPE_FLOAT_16_converted_output_1_QNN_DATATYPE_FLOAT_32", "type": "QNN", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "QNN_DATATYPE_FLOAT_16_converted_output_2_QNN_DATATYPE_FLOAT_32", "type": "QNN", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "QNN_DATATYPE_FLOAT_16_converted_output_3_QNN_DATATYPE_FLOAT_32", "type": "QNN", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "QNN_DATATYPE_FLOAT_16_converted_output_4_QNN_DATATYPE_FLOAT_32", "type": "QNN", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "QNN_DATATYPE_FLOAT_16_converted_output_5_QNN_DATATYPE_FLOAT_32", "type": "QNN", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "QNN_DATATYPE_FLOAT_16_converted_output_6_QNN_DATATYPE_FLOAT_32", "type": "QNN", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "QNN_DATATYPE_FLOAT_16_converted_output_7_QNN_DATATYPE_FLOAT_32", "type": "QNN", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "QNN_DATATYPE_FLOAT_16_converted_output_8_QNN_DATATYPE_FLOAT_32", "type": "QNN", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 0}, {"name": "Output", "type": "Output", "compute_unit": "NPU", "execution_time": 0, "execution_cycles": 48049}]}
Reference in new issue View Git Blame Copy Permalink