gahow/aituner
0
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Add OpProf campaign: protocols, results, patches, run evidence (P0-P6)

Browse Source 
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:
Gahow Wang
2026-07-13 11:06:10 +08:00
parent 607e88da3c
commit d5b276180d
412 changed files with 125056 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

1
runs/opprof-phase6/phase6/cells/tp4_mns16/anchor-0.033182214016.log Normal file
View File

@@ -0,0 +1 @@
{"cell": "tp4_mns16", "anchor": 0.033182214016, "kind": "anchor", "pass_rate": 0.6086956521739131, "feasible": false}

74
runs/opprof-phase6/phase6/cells/tp4_mns16/anchor-0.033182214016/result.json Normal file
View File

@@ -0,0 +1,74 @@
{
"anchor": 0.033182214016,
"cell": "tp4_mns16",
"early_stop_reason": "slo_pass_rate_unrecoverable",
"early_stopped": true,
"exact_output_count": 413,
"feasible": false,
"interval": {
"elapsed_s": 49.259978895,
"end_mono_ns": 201650764016357,
"end_wall_ns": 1783870077890443949,
"start_mono_ns": 201601504037462,
"start_wall_ns": 1783870028630464875
},
"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": 575,
"pass_rate": 0.6086956521739131,
"schema": 1,
"selection": {
"arrival_order_sha256": "86ee457cd91595c76454d54ebc69890836ef603ceb96deaf9c7cf05e91de9407",
"arrival_s": {
"distinct_n": 574,
"finite_n": 575,
"max": 59.90990000000002,
"min": 0.13000000000001818,
"missing_n": 0,
"n": 575
},
"count": 575,
"long_gt4096": 178,
"offered_req_s": 9.583333333333334,
"offered_req_s_per_gpu": 2.3958333333333335,
"raw_input_tokens": {
"distinct_n": 524,
"finite_n": 575,
"max": 8157.0,
"min": 69.0,
"missing_n": 0,
"n": 575
},
"raw_length_order_sha256": "8354300b1680841ab43ef8164307d7c1f28bc877994b8d1e7584aff030677560",
"request_id_order_sha256": "e6b624a0212807df3733c54b4a2160fba24f19ca4776b14f34918d83428bba7f"
},
"slo_pass_count": 350,
"study_sha256": "76e9d642ba9570917cfeeea848ed92f27904e87760eedabd48805b84e82eca26",
"tp": 4,
"tpot_ms": {
"distinct_n": 413,
"finite_n": 413,
"max": 16.932683684963983,
"min": 4.637226771506459,
"missing_n": 162,
"n": 575
},
"ttft_ms": {
"distinct_n": 413,
"finite_n": 413,
"max": 4166.508020018227,
"min": 38.04149298230186,
"missing_n": 162,
"n": 575
}
}

1
runs/opprof-phase6/phase6/cells/tp4_mns16/anchor-0.033717411016.log Normal file
View File

@@ -0,0 +1 @@
{"cell": "tp4_mns16", "anchor": 0.033717411016, "kind": "anchor", "pass_rate": 0.03583617747440273, "feasible": false}

74
runs/opprof-phase6/phase6/cells/tp4_mns16/anchor-0.033717411016/result.json Normal file
View File

@@ -0,0 +1,74 @@
{
"anchor": 0.033717411016,
"cell": "tp4_mns16",
"early_stop_reason": "slo_pass_rate_unrecoverable",
"early_stopped": true,
"exact_output_count": 199,
"feasible": false,
"interval": {
"elapsed_s": 41.949754618,
"end_mono_ns": 201595581507061,
"end_wall_ns": 1783870022707934728,
"start_mono_ns": 201553631752443,
"start_wall_ns": 1783869980758180054
},
"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": 586,
"pass_rate": 0.03583617747440273,
"schema": 1,
"selection": {
"arrival_order_sha256": "fe44bf6b0efd6e6ab4d19676e34ace608827e419dbb5c19c8595992f0339c9b1",
"arrival_s": {
"distinct_n": 585,
"finite_n": 586,
"max": 59.90990000000002,
"min": 0.13000000000001818,
"missing_n": 0,
"n": 586
},
"count": 586,
"long_gt4096": 184,
"offered_req_s": 9.766666666666667,
"offered_req_s_per_gpu": 2.441666666666667,
"raw_input_tokens": {
"distinct_n": 534,
"finite_n": 586,
"max": 8157.0,
"min": 69.0,
"missing_n": 0,
"n": 586
},
"raw_length_order_sha256": "e2d040068b279e99073f6965d1136eb0ca6e8534c883f331b4cb85c5fc7a4650",
"request_id_order_sha256": "472a874ecc0e844b5b36a715bbc309b63e73074d39327cd734869f76721d71db"
},
"slo_pass_count": 21,
"study_sha256": "76e9d642ba9570917cfeeea848ed92f27904e87760eedabd48805b84e82eca26",
"tp": 4,
"tpot_ms": {
"distinct_n": 199,
"finite_n": 199,
"max": 67.64735456709163,
"min": 4.931260244077526,
"missing_n": 387,
"n": 586
},
"ttft_ms": {
"distinct_n": 199,
"finite_n": 199,
"max": 16151.948870014166,
"min": 39.053958986187354,
"missing_n": 387,
"n": 586
}
}

1
runs/opprof-phase6/phase6/cells/tp4_mns16/anchor-0.034252608017.log Normal file
View File

@@ -0,0 +1 @@
{"cell": "tp4_mns16", "anchor": 0.034252608017, "kind": "anchor", "pass_rate": 0.8466666666666667, "feasible": false}

74
runs/opprof-phase6/phase6/cells/tp4_mns16/anchor-0.034252608017/result.json Normal file
View File

@@ -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
}
}

22
runs/opprof-phase6/phase6/cells/tp4_mns16/cell-valid.json Normal file
View File

@@ -0,0 +1,22 @@
{
"accounting_mode": "graceful-footer",
"cell": "tp4_mns16",
"invariants": {
"anchor_intervals_present": true,
"compile_capture_pre_ready": true,
"encoded_balanced": true,
"footer_sidecar_agrees": true,
"last_step_matches": true,
"layer1_contiguous": true,
"layer1_zero_drops": true,
"one_footer_last": true,
"one_ready_marker": true,
"sidecar_final": true,
"warmup_exact_16": true,
"warmup_long": true,
"written_matches_records": true
},
"layer1_records": 26137,
"post_ready_capture_events": [],
"stream": "/home/admin/cpfs/wjh/opprof-phase6-dash0-20260712/runs/phase6/cells/tp4_mns16/opprof/opprof-v1-dp0-pid2670831-1783869959685052499.jsonl"
}

6
runs/opprof-phase6/phase6/cells/tp4_mns16/commands.log Normal file
View File

@@ -0,0 +1,6 @@
SERVER taskset -c 80-99,100-119,120-139,140-159 /tmp/wjh-opprof-phase2-dash0-20260711/.venv/bin/vllm serve /home/admin/cpfs/wjh/models/Qwen/Qwen3-30B-A3B --host 127.0.0.1 --port 8521 --served-model-name qwen3-30b-a3b-community --max-num-batched-tokens 8192 --max-num-seqs 16 --tensor-parallel-size 4 --shutdown-timeout 120
CLIENT taskset -c 80-99,100-119,120-139,140-159 /tmp/wjh-opprof-phase2-dash0-20260711/.venv/bin/python /home/admin/cpfs/wjh/opprof-phase6-dash0-20260712/scripts/opprof_phase6_client.py warmup --study /home/admin/cpfs/wjh/opprof-phase6-dash0-20260712/provenance/study-tp4.json --cell tp4_mns16 --anchor 0.033717411016 --tp 4 --mns 16 --base-url http://127.0.0.1:8521 --result-dir /home/admin/cpfs/wjh/opprof-phase6-dash0-20260712/runs/phase6/cells/tp4_mns16/warmup
CLIENT taskset -c 80-99,100-119,120-139,140-159 /tmp/wjh-opprof-phase2-dash0-20260711/.venv/bin/python /home/admin/cpfs/wjh/opprof-phase6-dash0-20260712/scripts/opprof_phase6_client.py run-anchor --study /home/admin/cpfs/wjh/opprof-phase6-dash0-20260712/provenance/study-tp4.json --cell tp4_mns16 --anchor 0.033717411016 --tp 4 --mns 16 --base-url http://127.0.0.1:8521 --result-dir /home/admin/cpfs/wjh/opprof-phase6-dash0-20260712/runs/phase6/cells/tp4_mns16/anchor-0.033717411016
CLIENT taskset -c 80-99,100-119,120-139,140-159 /tmp/wjh-opprof-phase2-dash0-20260711/.venv/bin/python /home/admin/cpfs/wjh/opprof-phase6-dash0-20260712/scripts/opprof_phase6_client.py run-anchor --study /home/admin/cpfs/wjh/opprof-phase6-dash0-20260712/provenance/study-tp4.json --cell tp4_mns16 --anchor 0.033182214016 --tp 4 --mns 16 --base-url http://127.0.0.1:8521 --result-dir /home/admin/cpfs/wjh/opprof-phase6-dash0-20260712/runs/phase6/cells/tp4_mns16/anchor-0.033182214016
CLIENT taskset -c 80-99,100-119,120-139,140-159 /tmp/wjh-opprof-phase2-dash0-20260711/.venv/bin/python /home/admin/cpfs/wjh/opprof-phase6-dash0-20260712/scripts/opprof_phase6_client.py run-anchor --study /home/admin/cpfs/wjh/opprof-phase6-dash0-20260712/provenance/study-tp4.json --cell tp4_mns16 --anchor 0.034252608017 --tp 4 --mns 16 --base-url http://127.0.0.1:8521 --result-dir /home/admin/cpfs/wjh/opprof-phase6-dash0-20260712/runs/phase6/cells/tp4_mns16/anchor-0.034252608017
CLIENT taskset -c 80-99,100-119,120-139,140-159 /tmp/wjh-opprof-phase2-dash0-20260711/.venv/bin/python /home/admin/cpfs/wjh/opprof-phase6-dash0-20260712/scripts/opprof_phase6_client.py run-anchor --study /home/admin/cpfs/wjh/opprof-phase6-dash0-20260712/provenance/study-tp4.json --cell tp4_mns16 --anchor 0.033717411016 --tp 4 --mns 16 --base-url http://127.0.0.1:8521 --result-dir /home/admin/cpfs/wjh/opprof-phase6-dash0-20260712/runs/phase6/cells/tp4_mns16/confirm-1-anchor-0.033717411016

1
runs/opprof-phase6/phase6/cells/tp4_mns16/confirm-1-anchor-0.033717411016.log Normal file
View File

@@ -0,0 +1 @@
{"cell": "tp4_mns16", "anchor": 0.033717411016, "kind": "anchor", "pass_rate": 1.0, "feasible": true}

74
runs/opprof-phase6/phase6/cells/tp4_mns16/confirm-1-anchor-0.033717411016/result.json Normal file
View File

@@ -0,0 +1,74 @@
{
"anchor": 0.033717411016,
"cell": "tp4_mns16",
"early_stop_reason": "",
"early_stopped": false,
"exact_output_count": 586,
"feasible": true,
"interval": {
"elapsed_s": 61.364052595,
"end_mono_ns": 201800328727441,
"end_wall_ns": 1783870227455155582,
"start_mono_ns": 201738964674846,
"start_wall_ns": 1783870166091102450
},
"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": 586,
"pass_rate": 1.0,
"schema": 1,
"selection": {
"arrival_order_sha256": "fe44bf6b0efd6e6ab4d19676e34ace608827e419dbb5c19c8595992f0339c9b1",
"arrival_s": {
"distinct_n": 585,
"finite_n": 586,
"max": 59.90990000000002,
"min": 0.13000000000001818,
"missing_n": 0,
"n": 586
},
"count": 586,
"long_gt4096": 184,
"offered_req_s": 9.766666666666667,
"offered_req_s_per_gpu": 2.441666666666667,
"raw_input_tokens": {
"distinct_n": 534,
"finite_n": 586,
"max": 8157.0,
"min": 69.0,
"missing_n": 0,
"n": 586
},
"raw_length_order_sha256": "e2d040068b279e99073f6965d1136eb0ca6e8534c883f331b4cb85c5fc7a4650",
"request_id_order_sha256": "472a874ecc0e844b5b36a715bbc309b63e73074d39327cd734869f76721d71db"
},
"slo_pass_count": 586,
"study_sha256": "76e9d642ba9570917cfeeea848ed92f27904e87760eedabd48805b84e82eca26",
"tp": 4,
"tpot_ms": {
"distinct_n": 586,
"finite_n": 586,
"max": 14.420212055125647,
"min": 4.217363204660378,
"missing_n": 0,
"n": 586
},
"ttft_ms": {
"distinct_n": 586,
"finite_n": 586,
"max": 819.4046519929543,
"min": 36.395939998328686,
"missing_n": 0,
"n": 586
}
}

1
runs/opprof-phase6/phase6/cells/tp4_mns16/opprof/opprof-v1-dp0-pid2670831-1783869959685052499.jsonl.footer.json Normal file
View File

@@ -0,0 +1 @@
{"schema":1,"record_type":"footer_checkpoint","stream":"opprof-v1-dp0-pid2670831-1783869959685052499.jsonl","encoded_records":26137,"written_records":26137,"dropped_records":0,"last_step_index":26136,"checkpoint_wall_ns":1783870239366621112,"flush_interval_seconds":1.0,"final":true}

2003
runs/opprof-phase6/phase6/cells/tp4_mns16/server.log Normal file
View File

File diff suppressed because it is too large Load Diff

1
runs/opprof-phase6/phase6/cells/tp4_mns16/warmup.log Normal file
View File

@@ -0,0 +1 @@
{"cell": "tp4_mns16", "anchor": 0.033717411016, "kind": "warmup", "pass_rate": 1.0, "feasible": true}

74
runs/opprof-phase6/phase6/cells/tp4_mns16/warmup/result.json Normal file
View File

@@ -0,0 +1,74 @@
{
"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
}
}