Add OpProf campaign: protocols, results, patches, run evidence (P0-P6)
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>
This commit is contained in:
@@ -0,0 +1,74 @@
|
||||
{
|
||||
"anchor": 0.034252608017,
|
||||
"cell": "tp4_mns16",
|
||||
"early_stop_reason": "slo_pass_rate_unrecoverable",
|
||||
"early_stopped": true,
|
||||
"exact_output_count": 581,
|
||||
"feasible": false,
|
||||
"interval": {
|
||||
"elapsed_s": 63.947365009,
|
||||
"end_mono_ns": 201732794742162,
|
||||
"end_wall_ns": 1783870159921169450,
|
||||
"start_mono_ns": 201668847377153,
|
||||
"start_wall_ns": 1783870095973805093
|
||||
},
|
||||
"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": "anchor",
|
||||
"mns": 16,
|
||||
"observed_count": 600,
|
||||
"pass_rate": 0.8466666666666667,
|
||||
"schema": 1,
|
||||
"selection": {
|
||||
"arrival_order_sha256": "50f1b34c344c451c6d7501290d0b53f147d0c0b6a4bc078b303d5c33576ebe2d",
|
||||
"arrival_s": {
|
||||
"distinct_n": 599,
|
||||
"finite_n": 600,
|
||||
"max": 59.90990000000002,
|
||||
"min": 0.13000000000001818,
|
||||
"missing_n": 0,
|
||||
"n": 600
|
||||
},
|
||||
"count": 600,
|
||||
"long_gt4096": 187,
|
||||
"offered_req_s": 10.0,
|
||||
"offered_req_s_per_gpu": 2.5,
|
||||
"raw_input_tokens": {
|
||||
"distinct_n": 546,
|
||||
"finite_n": 600,
|
||||
"max": 8157.0,
|
||||
"min": 69.0,
|
||||
"missing_n": 0,
|
||||
"n": 600
|
||||
},
|
||||
"raw_length_order_sha256": "08eff8eecd74dc4646aa45945bd752432ee5eb795006755527dd32d7178c1140",
|
||||
"request_id_order_sha256": "21c1f3b7d972b19c952a894863a70d1e85bd0e24a7e7b6b0fcc088360d0e01bd"
|
||||
},
|
||||
"slo_pass_count": 508,
|
||||
"study_sha256": "76e9d642ba9570917cfeeea848ed92f27904e87760eedabd48805b84e82eca26",
|
||||
"tp": 4,
|
||||
"tpot_ms": {
|
||||
"distinct_n": 581,
|
||||
"finite_n": 581,
|
||||
"max": 17.87731141735631,
|
||||
"min": 4.250030724219904,
|
||||
"missing_n": 19,
|
||||
"n": 600
|
||||
},
|
||||
"ttft_ms": {
|
||||
"distinct_n": 581,
|
||||
"finite_n": 581,
|
||||
"max": 4941.07082701521,
|
||||
"min": 36.84570998302661,
|
||||
"missing_n": 19,
|
||||
"n": 600
|
||||
}
|
||||
}
|
||||
Reference in New Issue
Block a user