forked from llm-d/llm-d-benchmark
-
Notifications
You must be signed in to change notification settings - Fork 0
Expand file tree
/
Copy pathbr_v0_2_1_example.yaml
More file actions
979 lines (979 loc) · 25.5 KB
/
Copy pathbr_v0_2_1_example.yaml
File metadata and controls
979 lines (979 loc) · 25.5 KB
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
results:
component_health:
- component_label: vllm-svc-0
failed_replicas: 1
replica_health:
- healthy: true
logs: https://logs.example.com/vllm-svc-0-pod-1
replica_id: vllm-svc-0-pod-1
restarts: 1
- healthy: false
logs: /path/to/logs/vllm.log
replica_id: vllm-svc-0-pod-2
restarts: 2
- healthy: true
logs: s3://path/to/logs/vllm.log
replica_id: vllm-svc-0-pod-3
restarts: 0
total_restarts: 3
observability:
components:
- aggregate:
gpu_cache_usage:
max: 84.0
mean: 38.2
min: 10.1
p50: 37.5
p90: 58.7
p99: 78.3
units: percent
gpu_memory_usage:
max: 73.5
mean: 62.4
min: 58.1
p50: 62.0
p90: 66.8
p99: 71.2
units: GiB
gpu_utilization:
max: 96.8
mean: 76.0
min: 45.2
p50: 78.9
p90: 89.3
p99: 94.1
units: percent
kv_cache_usage:
max: 92.1
mean: 42.5
min: 12.3
p50: 41.8
p75: 52.1
p90: 63.4
p95: 71.2
p99: 85.6
units: percent
running_requests:
max: 62.0
mean: 24.6
min: 1.0
p50: 23.0
p90: 42.0
p99: 56.0
units: count
waiting_requests:
max: 25.0
mean: 3.2
min: 0.0
p50: 2.0
p90: 8.0
p99: 18.0
units: count
component_label: vllm-svc-0
time_series:
gpu_utilization:
series:
- mean: 73.2
p90: 85.4
ts: '2025-11-05T18:05:00Z'
- mean: 78.9
p90: 91.2
ts: '2025-11-05T18:10:00Z'
units: percent
kv_cache_usage:
series:
- mean: 35.2
p90: 48.1
p99: 62.5
ts: '2025-11-05T18:05:00Z'
- mean: 52.8
p90: 71.3
p99: 85.6
ts: '2025-11-05T18:10:00Z'
units: percent
- aggregate:
cpu_memory_usage:
max: 5.8
mean: 2.8
min: 1.2
p50: 2.6
p90: 4.1
p99: 5.3
units: GiB
cpu_utilization:
max: 42.0
mean: 18.4
min: 5.2
p50: 17.1
p90: 28.9
p99: 35.6
units: percent
running_requests:
max: 40.0
mean: 12.3
min: 0.0
p50: 11.0
p90: 22.0
p99: 34.0
units: count
waiting_requests:
max: 12.0
mean: 1.5
min: 0.0
p50: 1.0
p90: 4.0
p99: 9.0
units: count
component_label: epp-0
drop_rate:
max: 0.18
mean: 0.02
min: 0.0
p50: 0.0
p90: 0.05
p99: 0.12
units: percent
epp_dispatch_latency:
statistics:
graph_path: metrics/graphs/epp_dispatch_latency.png
mean: 0.0012
p50: 0.001
p99: 0.0035
stddev: 0.0008
units: seconds
epp_endpoint_scores:
components:
- component_id: qwen3-0p6b-decode-pod-1
statistics:
graph_path: metrics/graphs/epp_endpoint_scores.png
mean: 0.82
p50: 0.85
p99: 0.95
stddev: 0.08
units: score
epp_pool_avg_kv_cache_utilization:
components:
- component_id: epp-0
pod: qwen3-0p6b-epp-pod-1
role: epp
statistics:
graph_path: metrics/graphs/epp_pool_avg_kv_cache_utilization.png
mean: 40.3
p50: 39.8
p99: 82.1
stddev: 14.8
units: percent
epp_pool_avg_queue_size:
components:
- component_id: epp-0
pod: qwen3-0p6b-epp-pod-1
role: epp
statistics:
graph_path: metrics/graphs/epp_pool_avg_queue_size.png
mean: 4.7
p50: 3.0
p99: 22.0
stddev: 5.8
units: count
epp_pool_avg_running_requests:
components:
- component_id: epp-0
pod: qwen3-0p6b-epp-pod-1
role: epp
statistics:
graph_path: metrics/graphs/epp_pool_avg_running_requests.png
mean: 36.9
p50: 35.0
p99: 68.0
stddev: 16.2
units: count
epp_pool_ready_pods:
components:
- component_id: epp-0
pod: qwen3-0p6b-epp-pod-1
role: epp
statistics:
graph_path: metrics/graphs/epp_pool_ready_pods.png
mean: 2.4
p50: 2.0
p99: 3.0
stddev: 0.5
units: count
epp_request_distribution:
components:
- component_id: qwen3-0p6b-decode-pod-1
statistics:
count: 245
graph_path: metrics/graphs/epp_request_distribution.png
units: count
- component_id: qwen3-0p6b-decode-pod-2
statistics:
count: 255
graph_path: metrics/graphs/epp_request_distribution.png
units: count
pod_startup_times:
aggregate:
max: 95.0
mean: 84.0
min: 72.0
p50: 85.0
p90: 93.0
p99: 94.8
units: s
collected_at: '2025-11-05T18:17:00Z'
graph_path: metrics/graphs/pod_startup_times.png
pods:
- creation_timestamp: '2025-11-05T17:55:00Z'
model: Qwen/Qwen3-0.6B
name: qwen3-0p6b-decode-pod-1
node: gpu-node-01
ready_timestamp: '2025-11-05T17:56:12Z'
role: decode
startup_seconds: 72.0
- creation_timestamp: '2025-11-05T17:55:00Z'
model: Qwen/Qwen3-0.6B
name: qwen3-0p6b-decode-pod-2
node: gpu-node-02
ready_timestamp: '2025-11-05T17:56:25Z'
role: decode
startup_seconds: 85.0
- creation_timestamp: '2025-11-05T18:08:00Z'
model: Qwen/Qwen3-0.6B
name: qwen3-0p6b-decode-pod-3
node: gpu-node-03
ready_timestamp: '2025-11-05T18:09:35Z'
role: decode
startup_seconds: 95.0
replica_status:
aggregate_ready_replicas:
max: 3
mean: 2.4
min: 2
p50: 2
p90: 3
p99: 3
units: count
controllers:
- available_replicas: 3
desired_replicas: 3
kind: Deployment
model: Qwen/Qwen3-0.6B
name: qwen3-0p6b-decode
ready_replicas: 3
role: decode
updated_replicas: 3
- available_replicas: 0
desired_replicas: 0
kind: Deployment
model: Qwen/Qwen3-0.6B
name: qwen3-0p6b-prefill
ready_replicas: 0
role: prefill
updated_replicas: 0
graph_path: metrics/graphs/replica_status.png
namespace: benchmark-ns
time_series:
- controllers:
- available_replicas: 2
desired_replicas: 2
kind: Deployment
model: Qwen/Qwen3-0.6B
name: qwen3-0p6b-decode
ready_replicas: 2
role: decode
updated_replicas: 2
namespace: benchmark-ns
timestamp: '2025-11-05T18:00:40Z'
- controllers:
- available_replicas: 2
desired_replicas: 3
kind: Deployment
model: Qwen/Qwen3-0.6B
name: qwen3-0p6b-decode
ready_replicas: 2
role: decode
updated_replicas: 2
namespace: benchmark-ns
timestamp: '2025-11-05T18:08:10Z'
- controllers:
- available_replicas: 3
desired_replicas: 3
kind: Deployment
model: Qwen/Qwen3-0.6B
name: qwen3-0p6b-decode
ready_replicas: 3
role: decode
updated_replicas: 3
namespace: benchmark-ns
timestamp: '2025-11-05T18:09:40Z'
timestamp: '2025-11-05T18:17:00Z'
vllm_external_prefix_cache_hit_rate:
components:
- component_id: decode-engine
pod: qwen3-0p6b-decode-pod-1
role: decode
statistics:
graph_path: metrics/graphs/vllm_external_prefix_cache_hit_rate.png
mean: 42.9
p50: 44.2
p99: 68.5
stddev: 14.3
units: percent
vllm_external_prefix_cache_hits_total:
components:
- component_id: decode-engine
pod: qwen3-0p6b-decode-pod-1
role: decode
statistics:
graph_path: metrics/graphs/vllm_external_prefix_cache_hits_total.png
mean: 8200.0
p50: 8100.0
p99: 9500.0
stddev: 620.0
units: tokens
vllm_external_prefix_cache_queries_total:
components:
- component_id: decode-engine
pod: qwen3-0p6b-decode-pod-1
role: decode
statistics:
graph_path: metrics/graphs/vllm_external_prefix_cache_queries_total.png
mean: 19100.0
p50: 19000.0
p99: 21500.0
stddev: 1200.0
units: tokens
vllm_kv_cache_usage_perc:
aggregated:
mean: 40.3
p50: 39.8
p99: 82.1
stddev: 14.8
units: percent
components:
- component_id: decode-engine
pod: qwen3-0p6b-decode-pod-1
role: decode
statistics:
graph_path: metrics/graphs/vllm_kv_cache_usage_perc.png
mean: 42.5
p50: 41.8
p99: 85.6
stddev: 15.2
units: percent
vllm_nixl_bytes_transferred_count:
components:
- component_id: decode-engine
pod: qwen3-0p6b-decode-pod-1
role: decode
statistics:
graph_path: metrics/graphs/vllm_nixl_bytes_transferred_count.png
mean: 450.0
p50: 440.0
p99: 520.0
stddev: 35.0
units: count
vllm_nixl_bytes_transferred_sum:
components:
- component_id: decode-engine
pod: qwen3-0p6b-decode-pod-1
role: decode
statistics:
graph_path: metrics/graphs/vllm_nixl_bytes_transferred_sum.png
mean: 1250000000.0
p50: 1220000000.0
p99: 1480000000.0
stddev: 85000000.0
units: bytes
vllm_nixl_xfer_time_seconds_count:
components:
- component_id: decode-engine
pod: qwen3-0p6b-decode-pod-1
role: decode
statistics:
graph_path: metrics/graphs/vllm_nixl_xfer_time_seconds_count.png
mean: 450.0
p50: 440.0
p99: 520.0
stddev: 35.0
units: count
vllm_nixl_xfer_time_seconds_sum:
components:
- component_id: decode-engine
pod: qwen3-0p6b-decode-pod-1
role: decode
statistics:
graph_path: metrics/graphs/vllm_nixl_xfer_time_seconds_sum.png
mean: 0.0045
p50: 0.0042
p99: 0.0098
stddev: 0.0015
units: seconds
vllm_num_preemptions_total:
components:
- component_id: decode-engine
pod: qwen3-0p6b-decode-pod-1
role: decode
statistics:
graph_path: metrics/graphs/vllm_num_preemptions_total.png
mean: 0.5
p50: 0.0
p99: 3.0
stddev: 0.8
units: count
vllm_num_requests_running:
aggregated:
mean: 23.8
p50: 22.0
p99: 54.0
stddev: 11.5
units: count
components:
- component_id: decode-engine
pod: qwen3-0p6b-decode-pod-1
role: decode
statistics:
graph_path: metrics/graphs/vllm_num_requests_running.png
mean: 24.6
p50: 23.0
p99: 56.0
stddev: 12.1
units: count
vllm_num_requests_waiting:
components:
- component_id: decode-engine
pod: qwen3-0p6b-decode-pod-1
role: decode
statistics:
graph_path: metrics/graphs/vllm_num_requests_waiting.png
mean: 3.2
p50: 2.0
p99: 18.0
stddev: 4.5
units: count
vllm_prefix_cache_hit_rate:
components:
- component_id: decode-engine
pod: qwen3-0p6b-decode-pod-1
role: decode
statistics:
graph_path: metrics/graphs/vllm_prefix_cache_hit_rate.png
mean: 65.3
p50: 68.1
p99: 92.4
stddev: 18.7
units: percent
vllm_prefix_cache_hits_total:
components:
- component_id: decode-engine
pod: qwen3-0p6b-decode-pod-1
role: decode
statistics:
graph_path: metrics/graphs/vllm_prefix_cache_hits_total.png
mean: 12500.0
p50: 12400.0
p99: 14200.0
stddev: 850.0
units: tokens
vllm_prefix_cache_queries_total:
components:
- component_id: decode-engine
pod: qwen3-0p6b-decode-pod-1
role: decode
statistics:
graph_path: metrics/graphs/vllm_prefix_cache_queries_total.png
mean: 19100.0
p50: 19000.0
p99: 21500.0
stddev: 1200.0
units: tokens
profiling:
anything_goes: 5
request_performance:
aggregate:
latency:
inter_token_latency:
mean: 0.0368578234128654
units: s/token
normalized_time_per_output_token:
mean: 0.0368578234128654
units: s/token
request_latency:
mean: 0.0368578234128654
units: s
time_per_output_token:
mean: 0.0368578234128654
units: s/token
time_to_first_token:
max: 0.0574655020609498
mean: 0.0368578234128654
min: 0.027276142966002226
p0p1: 0.027322400792501866
p1: 0.02785794825293124
p10: 0.031235878029838203
p25: 0.033357357955537736
p5: 0.030159439309500158
p50: 0.03622930939309299
p75: 0.03993474820163101
p90: 0.04262634576298297
p95: 0.045246614259667695
p99: 0.05029489721637218
p99p9: 0.057014757215045425
units: s
requests:
failures: 0
input_length:
max: 2328.0
mean: 2262.448
min: 2223.0
p0p1: 2223.998
p1: 2227.99
p10: 2240.0
p25: 2252.0
p5: 2234.95
p50: 2262.0
p75: 2270.0
p90: 2286.0
p95: 2294.0
p99: 2322.0
p99p9: 2326.503
units: count
multimodal:
audio:
count:
mean: 1.0
units: count
duration:
mean: 10.0
units: s
image:
aspect_ratio:
mean: 1.78
units: ratio
filesize:
mean: 40000.0
units: bytes
count:
max: 4.0
mean: 2.0
units: count
pixels:
mean: 2073600.0
units: pixels
video:
filesize:
mean: 800000.0
units: bytes
count:
mean: 1.0
units: count
frames:
mean: 16.0
units: count
output_length:
max: 2328.0
mean: 2262.448
min: 2223.0
p0p1: 2223.998
p1: 2227.99
p10: 2240.0
p25: 2252.0
p5: 2234.95
p50: 2262.0
p75: 2270.0
p90: 2286.0
p95: 2294.0
p99: 2322.0
p99p9: 2326.503
units: count
request_size:
max: 120000.0
mean: 51234.0
p50: 48000.0
units: bytes
total: 500
throughput:
audio_rate:
mean: 2.0
units: audios/s
image_rate:
mean: 24.0
units: images/s
input_token_rate:
max: 2328.0
mean: 2262.448
min: 2223.0
p0p1: 2223.998
p1: 2227.99
p10: 2240.0
p25: 2252.0
p5: 2234.95
p50: 2262.0
p75: 2270.0
p90: 2286.0
p95: 2294.0
p99: 2322.0
p99p9: 2326.503
units: tokens/s
output_token_rate:
max: 2328.0
mean: 2262.448
min: 2223.0
p0p1: 2223.998
p1: 2227.99
p10: 2240.0
p25: 2252.0
p5: 2234.95
p50: 2262.0
p75: 2270.0
p90: 2286.0
p95: 2294.0
p99: 2322.0
p99p9: 2326.503
units: tokens/s
request_rate:
max: 2328.0
mean: 2262.448
min: 2223.0
p0p1: 2223.998
p1: 2227.99
p10: 2240.0
p25: 2252.0
p5: 2234.95
p50: 2262.0
p75: 2270.0
p90: 2286.0
p95: 2294.0
p99: 2322.0
p99p9: 2326.503
units: queries/s
total_token_rate:
max: 2328.0
mean: 2262.448
min: 2223.0
p0p1: 2223.998
p1: 2227.99
p10: 2240.0
p25: 2252.0
p5: 2234.95
p50: 2262.0
p75: 2270.0
p90: 2286.0
p95: 2294.0
p99: 2322.0
p99p9: 2326.503
units: tokens/s
video_rate:
mean: 3.0
units: videos/s
time_series:
latency:
time_to_first_token:
series:
- mean: 0.0368578234128654
p90: 0.04262634576298297
p95: 0.045246614259667695
ts: '2025-11-05T18:01:00Z'
- mean: 0.0368578234128654
p90: 0.04262634576298297
p95: 0.045246614259667695
ts: '2025-11-05T18:02:00Z'
units: s
throughput:
input_token_rate:
series:
- p90: 0.04262634576298297
p95: 0.045246614259667695
ts: '2025-11-05T18:01:00Z'
value: 204.0394
- p90: 0.04262634576298297
p95: 0.045246614259667695
ts: '2025-11-05T18:02:00Z'
value: 204.0394
units: tokens/s
output_token_rate:
series:
- p90: 0.04262634576298297
p95: 0.045246614259667695
ts: '2025-11-05T18:01:00Z'
value: 204.0394
- p90: 0.04262634576298297
p95: 0.045246614259667695
ts: '2025-11-05T18:02:00Z'
value: 204.0394
units: tokens/s
request_rate:
series:
- p90: 0.04262634576298297
p95: 0.045246614259667695
ts: '2025-11-05T18:01:00Z'
value: 204.0394
- p90: 0.04262634576298297
p95: 0.045246614259667695
ts: '2025-11-05T18:02:00Z'
value: 204.0394
units: queries/s
total_token_rate:
series:
- p90: 0.04262634576298297
p95: 0.045246614259667695
ts: '2025-11-05T18:01:00Z'
value: 204.0394
- p90: 0.04262634576298297
p95: 0.045246614259667695
ts: '2025-11-05T18:02:00Z'
value: 204.0394
units: tokens/s
units: tokens/s
session_performance:
sessions:
events_cancelled_per_session:
max: 3.0
mean: 0.18
min: 0.0
p50: 0.0
p99: 2.0
units: count
events_per_session:
max: 20.0
mean: 11.96
min: 1.0
p50: 12.0
p99: 20.0
units: count
failed: 2
input_tokens_per_session:
max: 46900.0
mean: 25612.4
min: 2148.0
p50: 25400.0
p99: 43020.0
units: count
output_tokens_per_session:
max: 22000.0
mean: 11800.2
min: 980.0
p50: 11700.0
p99: 20100.0
units: count
session_duration:
max: 80.1
mean: 48.3
min: 12.1
p50: 47.9
p90: 61.2
p99: 74.5
units: s
session_rate:
mean: 2.24
units: queries/s
succeeded: 110
total: 112
total_events: 1340
total_events_cancelled: 20
total_events_completed: 1320
run:
cid: 09825952f60004aa59bb5b2a2eefa6d1
description: Baseline latency test for Qwen3-0.6B on inference-scheduling stack
eid: d9c5a5ddce6c2f885a9f8e6a6a6db0fb
keywords:
- baseline
- latency
- qwen3
pid: 052d5f654ea08edf5caa2c0293fa7fb3
time:
duration: PT49.97206788184121S
end: '2025-11-05T18:17:03Z'
start: '2025-11-05T18:00:42Z'
uid: 38b1f169ca178b756f7483523b17de61
user: namasluk
scenario:
load:
metadata:
cfg_id: a4e18f265cc33786a42b8a3f7ac2edcb
description: Optional description of workload
schema_version: 0.0.1
native:
config:
api:
headers: null
streaming: true
type: completion
data:
input_distribution: null
output_distribution: null
path: null
shared_prefix:
num_groups: 32
num_prompts_per_group: 8
output_len: 1000
question_len: 100
system_prompt_len: 2048
type: shared_prefix
load:
interval: 1.0
num_workers: 112
stages:
- duration: 50
rate: 2.0
- duration: 50
rate: 5.0
- duration: 50
rate: 8.0
- duration: 50
rate: 10.0
- duration: 50
rate: 12.0
- duration: 50
rate: 15.0
- duration: 50
rate: 20.0
sweep: null
type: constant
worker_max_concurrency: 100
worker_max_tcp_connections: 2500
metrics: null
report:
prometheus:
per_stage: false
summary: true
request_lifecycle:
per_request: true
per_stage: true
summary: true
server:
api_key: null
base_url: http://infra-nam-release-inference-gateway.namasluk.svc.cluster.local:80/qwen-qwen3-0-6b
ignore_eos: true
model_name: Qwen/Qwen3-0.6B
type: vllm
storage:
google_cloud_storage: null
local_storage:
path: /requests/inference-perf_1759339259-cache_tracking-run_100_1000_llm-d-0p6b-base
report_file_prefix: null
simple_storage_service: null
tokenizer:
pretrained_model_name_or_path: Qwen/Qwen3-0.6B
token: null
trust_remote_code: null
standardized:
concurrency: .inf
input_seq_len:
distribution: fixed
value: 2148
multi_turn:
enabled: true
max_turns:
distribution: fixed
value: 20
output_seq_len:
distribution: gaussian
std_dev: 100.0
value: 1000
parallelism: 1
prefix:
num_groups: 32
num_prefixes: 8
num_users_per_group: 10
prefix_len:
distribution: fixed
value: 1000
rate_qps: 10.0
source: sampled
stage: 3
tool: inference-perf
tool_version: 0.3.0
stack:
- metadata:
cfg_id: cc73fc6b51a1d3b8128f312d70476d7c
description: Optional description of this component
label: vllm-svc-0
schema_version: 0.0.1
native:
args:
--block-size: 1024
--disable-uvicorn-access-log: null
--kv-transfer-config:
kv_connector: NixlConnector
kv_role: kv_both
--max-model-len: 1024
--no-enable-log-requests: null
--tensor-parallel-size: 8
envars:
NCCL_IB_HCA: mlx5_1
UCX_NET_DEVICES: mlx5_1:1
UCX_SOCKADDR_TLS_PRIORITY: tcp
UCX_TLS: rc,sm,cuda_ipc,cuda_copy,tcp
VLLM_ALLOW_LONG_MAX_MODEL_LEN: '1'
VLLM_LOGGING_LEVEL: DEBUG
VLLM_NIXL_SIDE_CHANNEL_HOST: 10.39.39.3
VLLM_NIXL_SIDE_CHANNEL_PORT: '5557'
standardized:
accelerator:
count: 8
model: NVIDIA-H100-80GB-HBM3
parallelism:
dp: 1
dp_local: 1
ep: 1
pp: 1
tp: 8
workers: 1
kind: inference_engine
model:
name: Qwen/Qwen3-0.6B
replicas: 2
role: decode
tool: llm-d
tool_version: ghcr.io/llm-d/llm-d-cuda:0.3.1
- metadata:
cfg_id: 47000e70c655e88198e0dc4e57d41d5f
description: This is a router, but no standardized component for this exists
label: epp-0
schema_version: 0.0.1
native:
config:
apiVersion: inference.networking.x-k8s.io/v1alpha1
kind: EndpointPickerConfig
plugins:
- type: single-profile-handler
- type: decode-filter
- parameters:
blockSize: 16
indexerConfig:
kvBlockIndexConfig:
enableMetrics: true
metricsLoggingInterval: 60000000000
tokenProcessorConfig:
blockSize: 64
hashSeed: '42'
lruCapacityPerServer: 31250
maxPrefixBlocksToMatch: 256
mode: cache_tracking
type: prefix-cache-scorer
- type: kv-cache-scorer
- type: queue-scorer
- type: max-score-picker
schedulingProfiles:
- name: default
plugins:
- pluginRef: decode-filter
- pluginRef: prefix-cache-scorer
weight: 2.0
- pluginRef: kv-cache-scorer
weight: 1.0
- pluginRef: queue-scorer
weight: 1.0
- pluginRef: max-score-picker
standardized:
another_thing:
- a: 5
- a: 1
- b: 3
kind: generic
kind_draft: router
some_config_param: 93
tool: llm-d-inference-scheduler
tool_version: ghcr.io/llm-d/llm-d-inference-scheduler:0.3.2
version: 0.2.1