Gahow Wang
d5b276180d
Workload-conditioned operator profiling on patched vLLM 0.24.0 + Qwen3-30B-A3B/H20. H1b PASS (irregular patterns carry +23-45pp R64 raggedness, 8-45% token-efficiency loss vs rectangular controls); mechanism decomposition kills the padding narrative and finds the arrival-uniformization artifact (-12.9%); cross-version churn surface shows TP2/MNS64 -29.4% across vLLM 0.20->0.24 while the argmax held. Raw Layer-1 JSONL streams (507 MB) stay on disk, git-ignored; footer sidecars and metrics are tracked. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
75 lines
1.9 KiB
JSON
75 lines
1.9 KiB
JSON
{
|
|
"anchor": 0.033717411016,
|
|
"cell": "tp4_mns16",
|
|
"early_stop_reason": "",
|
|
"early_stopped": false,
|
|
"exact_output_count": 16,
|
|
"feasible": true,
|
|
"interval": {
|
|
"elapsed_s": 3.923353377,
|
|
"end_mono_ns": 201544201950878,
|
|
"end_wall_ns": 1783869971328377792,
|
|
"start_mono_ns": 201540278597501,
|
|
"start_wall_ns": 1783869967405025053
|
|
},
|
|
"invariants": {
|
|
"arrival_nondecreasing": true,
|
|
"exact_output_or_failed": true,
|
|
"outcomes_cover_selected": true,
|
|
"raw_lengths_present": true,
|
|
"selected_nonempty": true,
|
|
"warmup_16": true,
|
|
"warmup_exact_16": true,
|
|
"warmup_long": true
|
|
},
|
|
"kind": "warmup",
|
|
"mns": 16,
|
|
"observed_count": 16,
|
|
"pass_rate": 1.0,
|
|
"schema": 1,
|
|
"selection": {
|
|
"arrival_order_sha256": "90b5197b9bfe4a9d6dfa881ec683b528ba46dd7bdc3a42fe2c91f6aefc6c8308",
|
|
"arrival_s": {
|
|
"distinct_n": 16,
|
|
"finite_n": 16,
|
|
"max": 1.7244999999999893,
|
|
"min": 0.0,
|
|
"missing_n": 0,
|
|
"n": 16
|
|
},
|
|
"count": 16,
|
|
"long_gt4096": 2,
|
|
"offered_req_s": 0.26666666666666666,
|
|
"offered_req_s_per_gpu": 0.06666666666666667,
|
|
"raw_input_tokens": {
|
|
"distinct_n": 16,
|
|
"finite_n": 16,
|
|
"max": 8150.0,
|
|
"min": 371.0,
|
|
"missing_n": 0,
|
|
"n": 16
|
|
},
|
|
"raw_length_order_sha256": "d416a666bb3b2d9a0f1ffe0c51cc4e41249123c02ecb4eadb6fbb049422d7e3d",
|
|
"request_id_order_sha256": "01b7744965c81da433fd52073d45f36d4b5bc750ca0bbd5ab0db8d8eac316fa3"
|
|
},
|
|
"slo_pass_count": 16,
|
|
"study_sha256": "76e9d642ba9570917cfeeea848ed92f27904e87760eedabd48805b84e82eca26",
|
|
"tp": 4,
|
|
"tpot_ms": {
|
|
"distinct_n": 16,
|
|
"finite_n": 16,
|
|
"max": 21.061303456796068,
|
|
"min": 6.4699021653880875,
|
|
"missing_n": 0,
|
|
"n": 16
|
|
},
|
|
"ttft_ms": {
|
|
"distinct_n": 16,
|
|
"finite_n": 16,
|
|
"max": 1454.2098789825104,
|
|
"min": 132.21911297296174,
|
|
"missing_n": 0,
|
|
"n": 16
|
|
}
|
|
}
|