A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 8, 01:15 UTC
[ROCm][CI] Removing all blocking labels from MI355 until stable infra (#34879)
Signed-off-by: Andreas Karatzas <akaratza@amd.com>
Andreas Karatzas committed
Feb 19, 2026 at 01:53 UTC
2df2bb27b0fd624d8abd0fda3f8c337f1e8c60fc
1 file changed
-103
.buildkite/test-amd.yaml
-103
@@ -1107,18 +1107,6 @@ steps:
1107
commands:
1108
- pytest -v -s models/quantization
1109
1110
-# This test is used only in PR development phase to test individual models and should never run on main
1111
-- label: Custom Models Test
1112
- mirror_hardwares: [amdexperimental, amdproduction]
1113
- agent_pool: mi325_1
1114
- # grade: Blocking
1115
- optional: true
1116
- commands:
1117
- - echo 'Testing custom models...'
1118
- # PR authors can temporarily add commands below to test individual models
1119
- # e.g. pytest -v -s models/encoder_decoder/vision_language/test_mllama.py
1120
- # *To avoid merge conflicts, remember to REMOVE (not just comment out) them before merging the PR*
1121
-
1110
- label: Transformers Nightly Models Test
1111
mirror_hardwares: [amdexperimental]
1112
agent_pool: mi325_1
@@ -1709,7 +1697,6 @@ steps:
1697
# in /vllm/tools/pre_commit/generate_nightly_torch_test.py
1698
mirror_hardwares: [amdexperimental, amdproduction, amdtentative]
1699
agent_pool: mi355_1
1712
- grade: Blocking
1700
soft_fail: true
1701
source_file_dependencies:
1702
- requirements/nightly_torch_test.txt
@@ -1720,7 +1707,6 @@ steps:
1707
timeout_in_minutes: 15
1708
mirror_hardwares: [amdexperimental, amdproduction, amdtentative]
1709
agent_pool: mi355_1
1723
- grade: Blocking
1710
source_file_dependencies:
1711
- vllm/
1712
- tests/multimodal
@@ -1733,7 +1719,6 @@ steps:
1719
timeout_in_minutes: 30
1720
mirror_hardwares: [amdexperimental, amdproduction, amdtentative]
1721
agent_pool: mi355_1
1736
- grade: Blocking
1722
source_file_dependencies:
1723
- vllm/
1724
- tests/test_inputs.py
@@ -1763,7 +1748,6 @@ steps:
1748
timeout_in_minutes: 20
1749
mirror_hardwares: [amdexperimental]
1750
agent_pool: mi355_1
1766
- # grade: Blocking
1751
source_file_dependencies:
1752
- tests/standalone_tests/python_only_compile.sh
1753
- setup.py
@@ -1774,7 +1758,6 @@ steps:
1758
timeout_in_minutes: 30
1759
mirror_hardwares: [amdexperimental, amdproduction]
1760
agent_pool: mi355_1
1777
- # grade: Blocking
1761
fast_check: true
1762
torch_nightly: true
1763
source_file_dependencies:
@@ -1791,7 +1774,6 @@ steps:
1774
- label: Entrypoints Unit Tests # 5min
1775
mirror_hardwares: [amdexperimental, amdproduction, amdtentative]
1776
agent_pool: mi355_1
1794
- grade: Blocking
1777
timeout_in_minutes: 10
1778
working_dir: "/vllm-workspace/tests"
1779
fast_check: true
@@ -1806,7 +1788,6 @@ steps:
1788
timeout_in_minutes: 40
1789
mirror_hardwares: [amdexperimental, amdproduction]
1790
agent_pool: mi355_1
1809
- # grade: Blocking
1791
working_dir: "/vllm-workspace/tests"
1792
fast_check: true
1793
torch_nightly: true
@@ -1824,7 +1805,6 @@ steps:
1805
timeout_in_minutes: 130
1806
mirror_hardwares: [amdexperimental]
1807
agent_pool: mi355_1
1827
- # grade: Blocking
1808
working_dir: "/vllm-workspace/tests"
1809
fast_check: true
1810
torch_nightly: true
@@ -1841,7 +1821,6 @@ steps:
1821
timeout_in_minutes: 50
1822
mirror_hardwares: [amdexperimental]
1823
agent_pool: mi355_1
1844
- # grade: Blocking
1824
working_dir: "/vllm-workspace/tests"
1825
fast_check: true
1826
torch_nightly: true
@@ -1860,7 +1839,6 @@ steps:
1839
timeout_in_minutes: 50
1840
mirror_hardwares: [amdexperimental]
1841
agent_pool: mi355_1
1863
- # grade: Blocking
1842
working_dir: "/vllm-workspace/tests"
1843
fast_check: true
1844
torch_nightly: true
@@ -1875,7 +1853,6 @@ steps:
1853
timeout_in_minutes: 50
1854
mirror_hardwares: [amdexperimental]
1855
agent_pool: mi355_1
1878
- # grade: Blocking
1856
working_dir: "/vllm-workspace/tests"
1857
fast_check: true
1858
torch_nightly: true
@@ -1890,7 +1867,6 @@ steps:
1867
timeout_in_minutes: 50
1868
mirror_hardwares: [amdexperimental]
1869
agent_pool: mi355_4
1893
- # grade: Blocking
1870
working_dir: "/vllm-workspace/tests"
1871
num_gpus: 4
1872
source_file_dependencies:
@@ -1952,7 +1928,6 @@ steps:
1928
timeout_in_minutes: 10
1929
mirror_hardwares: [amdexperimental]
1930
agent_pool: mi355_8
1955
- # grade: Blocking
1931
gpu: h100
1932
num_gpus: 8
1933
working_dir: "/vllm-workspace/tests"
@@ -1973,7 +1948,6 @@ steps:
1948
- label: EPLB Algorithm Test # 5min
1949
mirror_hardwares: [amdexperimental, amdproduction, amdtentative]
1950
agent_pool: mi355_1
1976
- grade: Blocking
1951
timeout_in_minutes: 15
1952
working_dir: "/vllm-workspace/tests"
1953
source_file_dependencies:
@@ -1985,7 +1959,6 @@ steps:
1959
- label: EPLB Execution Test # 10min
1960
mirror_hardwares: [amdexperimental, amdproduction]
1961
agent_pool: mi355_4
1988
- # grade: Blocking
1962
timeout_in_minutes: 20
1963
working_dir: "/vllm-workspace/tests"
1964
num_gpus: 4
@@ -2000,7 +1973,6 @@ steps:
1973
timeout_in_minutes: 20
1974
mirror_hardwares: [amdexperimental, amdproduction]
1975
agent_pool: mi355_2
2003
- # grade: Blocking
1976
num_gpus: 2
1977
source_file_dependencies:
1978
- vllm/
@@ -2020,7 +1992,6 @@ steps:
1992
timeout_in_minutes: 20
1993
mirror_hardwares: [amdexperimental, amdproduction, amdtentative]
1994
agent_pool: mi355_1
2023
- grade: Blocking
1995
source_file_dependencies:
1996
- vllm/
1997
- tests/test_regression
@@ -2033,7 +2004,6 @@ steps:
2004
timeout_in_minutes: 15
2005
mirror_hardwares: [amdexperimental, amdproduction]
2006
agent_pool: mi355_1
2036
- # grade: Blocking
2007
source_file_dependencies:
2008
- vllm/
2009
- tests/engine
@@ -2050,7 +2020,6 @@ steps:
2020
# The test uses 4 GPUs, but we schedule it on 8-GPU machines for stability.
2021
# See discussion here: https://github.com/vllm-project/vllm/pull/31040
2022
agent_pool: mi355_8
2053
- # grade: Blocking
2023
source_file_dependencies:
2024
- vllm/
2025
- tests/v1
@@ -2064,7 +2033,6 @@ steps:
2033
timeout_in_minutes: 50
2034
mirror_hardwares: [amdexperimental, amdproduction, amdtentative]
2035
agent_pool: mi355_1
2067
- grade: Blocking
2036
source_file_dependencies:
2037
- vllm/
2038
- tests/v1
@@ -2075,7 +2043,6 @@ steps:
2043
timeout_in_minutes: 60
2044
mirror_hardwares: [amdexperimental]
2045
agent_pool: mi355_1
2078
- # grade: Blocking
2046
source_file_dependencies:
2047
- vllm/
2048
- tests/v1
@@ -2103,7 +2070,6 @@ steps:
2070
- label: V1 Test attention (H100) # 10min
2071
mirror_hardwares: [amdexperimental]
2072
agent_pool: mi355_1
2106
- # grade: Blocking
2073
timeout_in_minutes: 30
2074
gpu: h100
2075
source_file_dependencies:
@@ -2143,7 +2109,6 @@ steps:
2109
- label: V1 Test others (CPU) # 5 mins
2110
mirror_hardwares: [amdexperimental, amdproduction, amdtentative]
2111
agent_pool: mi355_1
2146
- grade: Blocking
2112
source_file_dependencies:
2113
- vllm/
2114
- tests/v1
@@ -2161,7 +2126,6 @@ steps:
2126
timeout_in_minutes: 45
2127
mirror_hardwares: [amdexperimental, amdproduction]
2128
agent_pool: mi355_1
2164
- # grade: Blocking
2129
working_dir: "/vllm-workspace/examples"
2130
source_file_dependencies:
2131
- vllm/entrypoints
@@ -2196,7 +2160,6 @@ steps:
2160
timeout_in_minutes: 15
2161
mirror_hardwares: [amdexperimental, amdproduction]
2162
agent_pool: mi355_1
2199
- # grade: Blocking
2163
source_file_dependencies:
2164
- vllm/
2165
- tests/cuda
@@ -2208,7 +2171,6 @@ steps:
2171
timeout_in_minutes: 75
2172
mirror_hardwares: [amdexperimental]
2173
agent_pool: mi355_1
2211
- # grade: Blocking
2174
source_file_dependencies:
2175
- vllm/model_executor/layers
2176
- vllm/sampling_metadata.py
@@ -2221,7 +2183,6 @@ steps:
2183
timeout_in_minutes: 30
2184
mirror_hardwares: [amdexperimental]
2185
agent_pool: mi355_1
2224
- # grade: Blocking
2186
source_file_dependencies:
2187
- vllm/lora
2188
- tests/lora
@@ -2242,7 +2203,6 @@ steps:
2203
timeout_in_minutes: 30
2204
mirror_hardwares: [amdexperimental, amdproduction]
2205
agent_pool: mi355_1
2245
- # grade: Blocking
2206
torch_nightly: true
2207
source_file_dependencies:
2208
- vllm/
@@ -2259,7 +2219,6 @@ steps:
2219
timeout_in_minutes: 30
2220
mirror_hardwares: [amdexperimental, amdproduction]
2221
agent_pool: mi355_1
2262
- # grade: Blocking
2222
torch_nightly: true
2223
source_file_dependencies:
2224
- vllm/
@@ -2305,7 +2264,6 @@ steps:
2264
timeout_in_minutes: 75
2265
mirror_hardwares: [amdexperimental, amdproduction]
2266
agent_pool: mi355_1
2308
- # grade: Blocking
2267
source_file_dependencies:
2268
- csrc/
2269
- tests/kernels/core
@@ -2317,7 +2275,6 @@ steps:
2275
timeout_in_minutes: 35
2276
mirror_hardwares: [amdexperimental, amdproduction]
2277
agent_pool: mi355_1
2320
- # grade: Blocking
2278
source_file_dependencies:
2279
- csrc/attention/
2280
- vllm/v1/attention
@@ -2332,7 +2289,6 @@ steps:
2289
timeout_in_minutes: 90
2290
mirror_hardwares: [amdexperimental]
2291
agent_pool: mi355_1
2335
- # grade: Blocking
2292
source_file_dependencies:
2293
- csrc/quantization/
2294
- vllm/model_executor/layers/quantization
@@ -2345,7 +2301,6 @@ steps:
2301
timeout_in_minutes: 60
2302
mirror_hardwares: [amdexperimental, amdproduction]
2303
agent_pool: mi355_1
2348
- # grade: Blocking
2304
source_file_dependencies:
2305
- csrc/quantization/cutlass_w8a8/moe/
2306
- csrc/moe/
@@ -2362,7 +2317,6 @@ steps:
2317
timeout_in_minutes: 45
2318
mirror_hardwares: [amdexperimental, amdproduction]
2319
agent_pool: mi355_1
2365
- # grade: Blocking
2320
source_file_dependencies:
2321
- csrc/mamba/
2322
- tests/kernels/mamba
@@ -2406,7 +2360,6 @@ steps:
2360
torch_nightly: true
2361
mirror_hardwares: [amdexperimental, amdproduction]
2362
agent_pool: mi355_1
2409
- # grade: Blocking
2363
source_file_dependencies:
2364
- vllm/engine/arg_utils.py
2365
- vllm/config/model.py
@@ -2423,7 +2376,6 @@ steps:
2376
timeout_in_minutes: 20
2377
mirror_hardwares: [amdexperimental, amdproduction]
2378
agent_pool: mi355_1
2426
- # grade: Blocking
2379
working_dir: "/vllm-workspace/.buildkite"
2380
source_file_dependencies:
2381
- benchmarks/
@@ -2434,7 +2386,6 @@ steps:
2386
timeout_in_minutes: 20
2387
mirror_hardwares: [amdexperimental, amdproduction]
2388
agent_pool: mi355_1
2437
- # grade: Blocking
2389
source_file_dependencies:
2390
- vllm/
2391
- tests/benchmarks/
@@ -2445,7 +2396,6 @@ steps:
2396
timeout_in_minutes: 90
2397
mirror_hardwares: [amdexperimental, amdproduction]
2398
agent_pool: mi355_1
2448
- # grade: Blocking
2399
source_file_dependencies:
2400
- csrc/
2401
- vllm/model_executor/layers/quantization
@@ -2466,7 +2416,6 @@ steps:
2416
timeout_in_minutes: 75
2417
mirror_hardwares: [amdexperimental]
2418
agent_pool: mi355_1
2469
- # grade: Blocking
2419
source_file_dependencies:
2420
- csrc/
2421
- vllm/model_executor/layers/quantization
@@ -2478,7 +2427,6 @@ steps:
2427
timeout_in_minutes: 15
2428
mirror_hardwares: [amdexperimental, amdproduction]
2429
agent_pool: mi355_1
2481
- # grade: Blocking
2430
source_file_dependencies:
2431
- csrc/
2432
- vllm/entrypoints/openai/
@@ -2495,7 +2443,6 @@ steps:
2443
timeout_in_minutes: 45
2444
mirror_hardwares: [amdexperimental, amdproduction]
2445
agent_pool: mi355_1
2498
- # grade: Blocking
2446
torch_nightly: true
2447
source_file_dependencies:
2448
- vllm/
@@ -2508,7 +2455,6 @@ steps:
2455
timeout_in_minutes: 45
2456
mirror_hardwares: [amdexperimental, amdproduction]
2457
agent_pool: mi355_1
2511
- # grade: Blocking
2458
torch_nightly: true
2459
source_file_dependencies:
2460
- vllm/model_executor/models/
@@ -2528,7 +2474,6 @@ steps:
2474
timeout_in_minutes: 45
2475
mirror_hardwares: [amdexperimental]
2476
agent_pool: mi355_1
2531
- # grade: Blocking
2477
torch_nightly: true
2478
source_file_dependencies:
2479
- vllm/
@@ -2541,7 +2486,6 @@ steps:
2486
- label: Basic Models Test (Other CPU) # 5min
2487
mirror_hardwares: [amdexperimental, amdproduction]
2488
agent_pool: mi355_1
2544
- # grade: Blocking
2489
timeout_in_minutes: 10
2490
torch_nightly: true
2491
source_file_dependencies:
@@ -2556,7 +2500,6 @@ steps:
2500
timeout_in_minutes: 25
2501
mirror_hardwares: [amdexperimental, amdproduction]
2502
agent_pool: mi355_1
2559
- # grade: Blocking
2503
torch_nightly: true
2504
source_file_dependencies:
2505
- vllm/
@@ -2570,7 +2513,6 @@ steps:
2513
timeout_in_minutes: 45
2514
mirror_hardwares: [amdexperimental]
2515
agent_pool: mi355_1
2573
- # grade: Blocking
2516
torch_nightly: true
2517
source_file_dependencies:
2518
- vllm/model_executor/models/
@@ -2591,7 +2533,6 @@ steps:
2533
timeout_in_minutes: 75
2534
mirror_hardwares: [amdexperimental]
2535
agent_pool: mi355_1
2594
- # grade: Blocking
2536
torch_nightly: true
2537
source_file_dependencies:
2538
- vllm/
@@ -2612,7 +2553,6 @@ steps:
2553
timeout_in_minutes: 110
2554
mirror_hardwares: [amdexperimental]
2555
agent_pool: mi355_1
2615
- # grade: Blocking
2556
optional: true
2557
source_file_dependencies:
2558
- vllm/
@@ -2628,7 +2568,6 @@ steps:
2568
timeout_in_minutes: 110
2569
mirror_hardwares: [amdexperimental]
2570
agent_pool: mi355_1
2631
- # grade: Blocking
2571
optional: true
2572
source_file_dependencies:
2573
- vllm/
@@ -2640,7 +2579,6 @@ steps:
2579
timeout_in_minutes: 50
2580
mirror_hardwares: [amdexperimental]
2581
agent_pool: mi355_1
2643
- # grade: Blocking
2582
optional: true
2583
source_file_dependencies:
2584
- vllm/
@@ -2676,7 +2614,6 @@ steps:
2614
timeout_in_minutes: 60
2615
mirror_hardwares: [amdexperimental]
2616
agent_pool: mi355_1
2679
- # grade: Blocking
2617
source_file_dependencies:
2618
- vllm/
2619
- tests/models/multimodal
@@ -2688,7 +2625,6 @@ steps:
2625
timeout_in_minutes: 100
2626
mirror_hardwares: [amdexperimental]
2627
agent_pool: mi355_1
2691
- # grade: Blocking
2628
torch_nightly: true
2629
source_file_dependencies:
2630
- vllm/
@@ -2706,7 +2642,6 @@ steps:
2642
timeout_in_minutes: 10
2643
mirror_hardwares: [amdexperimental, amdproduction]
2644
agent_pool: mi355_1
2709
- # grade: Blocking
2645
working_dir: "/vllm-workspace/.buildkite/lm-eval-harness"
2646
source_file_dependencies:
2647
- vllm/multimodal/
@@ -2721,7 +2656,6 @@ steps:
2656
timeout_in_minutes: 120
2657
mirror_hardwares: [amdexperimental]
2658
agent_pool: mi355_1
2724
- # grade: Blocking
2659
optional: true
2660
source_file_dependencies:
2661
- vllm/
@@ -2736,7 +2670,6 @@ steps:
2670
timeout_in_minutes: 120
2671
mirror_hardwares: [amdexperimental]
2672
agent_pool: mi355_1
2739
- # grade: Blocking
2673
optional: true
2674
source_file_dependencies:
2675
- vllm/
@@ -2751,7 +2684,6 @@ steps:
2684
timeout_in_minutes: 150
2685
mirror_hardwares: [amdexperimental]
2686
agent_pool: mi355_1
2754
- # grade: Blocking
2687
optional: true
2688
source_file_dependencies:
2689
- vllm/
@@ -2766,29 +2698,15 @@ steps:
2698
timeout_in_minutes: 60
2699
mirror_hardwares: [amdexperimental, amdproduction]
2700
agent_pool: mi355_1
2769
- # grade: Blocking
2701
source_file_dependencies:
2702
- vllm/model_executor/layers/quantization
2703
- tests/models/quantization
2704
commands:
2705
- pytest -v -s models/quantization
2706
2776
-# This test is used only in PR development phase to test individual models and should never run on main
2777
-- label: Custom Models Test
2778
- mirror_hardwares: [amdexperimental, amdproduction]
2779
- agent_pool: mi355_1
2780
- # grade: Blocking
2781
- optional: true
2782
- commands:
2783
- - echo 'Testing custom models...'
2784
- # PR authors can temporarily add commands below to test individual models
2785
- # e.g. pytest -v -s models/encoder_decoder/vision_language/test_mllama.py
2786
- # *To avoid merge conflicts, remember to REMOVE (not just comment out) them before merging the PR*
2787
-
2707
- label: Transformers Nightly Models Test
2708
mirror_hardwares: [amdexperimental]
2709
agent_pool: mi355_1
2791
- # grade: Blocking
2710
working_dir: "/vllm-workspace/"
2711
optional: true
2712
commands:
@@ -2927,7 +2845,6 @@ steps:
2845
timeout_in_minutes: 20
2846
mirror_hardwares: [amdexperimental, amdproduction]
2847
agent_pool: mi355_2
2930
- # grade: Blocking
2848
working_dir: "/vllm-workspace/tests"
2849
num_gpus: 2
2850
source_file_dependencies:
@@ -2943,7 +2860,6 @@ steps:
2860
timeout_in_minutes: 30
2861
mirror_hardwares: [amdexperimental, amdmultinode]
2862
agent_pool: mi355_4
2946
- # grade: Blocking
2863
working_dir: "/vllm-workspace/tests"
2864
num_gpus: 2
2865
num_nodes: 2
@@ -2970,7 +2886,6 @@ steps:
2886
timeout_in_minutes: 90
2887
mirror_hardwares: [amdexperimental]
2888
agent_pool: mi355_2
2973
- # grade: Blocking
2889
working_dir: "/vllm-workspace/tests"
2890
num_gpus: 2
2891
source_file_dependencies:
@@ -3010,7 +2925,6 @@ steps:
2925
timeout_in_minutes: 50
2926
mirror_hardwares: [amdexperimental]
2927
agent_pool: mi355_2
3013
- # grade: Blocking
2928
working_dir: "/vllm-workspace/tests"
2929
num_gpus: 2
2930
source_file_dependencies:
@@ -3032,7 +2946,6 @@ steps:
2946
timeout_in_minutes: 60
2947
mirror_hardwares: [amdexperimental, amdproduction]
2948
agent_pool: mi355_2
3035
- # grade: Blocking
2949
working_dir: "/vllm-workspace/tests"
2950
num_gpus: 2
2951
source_file_dependencies:
@@ -3066,7 +2979,6 @@ steps:
2979
timeout_in_minutes: 60
2980
mirror_hardwares: [amdexperimental, amdproduction]
2981
agent_pool: mi355_4
3069
- # grade: Blocking
2982
working_dir: "/vllm-workspace/tests"
2983
num_gpus: 4
2984
source_file_dependencies:
@@ -3083,7 +2995,6 @@ steps:
2995
timeout_in_minutes: 30
2996
mirror_hardwares: [amdexperimental, amdproduction]
2997
agent_pool: mi355_4
3086
- # grade: Blocking
2998
num_gpus: 4
2999
source_file_dependencies:
3000
- vllm/lora
@@ -3108,7 +3019,6 @@ steps:
3019
timeout_in_minutes: 45
3020
mirror_hardwares: [amdexperimental, amdproduction]
3021
agent_pool: mi355_2
3111
- # grade: Blocking
3022
working_dir: "/vllm-workspace/tests"
3023
num_gpus: 2
3024
optional: true
@@ -3121,7 +3031,6 @@ steps:
3031
- label: Weight Loading Multiple GPU Test - Large Models # optional
3032
mirror_hardwares: [amdexperimental]
3033
agent_pool: mi355_2
3124
- # grade: Blocking
3034
working_dir: "/vllm-workspace/tests"
3035
num_gpus: 2
3036
optional: true
@@ -3134,7 +3043,6 @@ steps:
3043
- label: NixlConnector PD accuracy tests (Distributed) # 30min
3044
mirror_hardwares: [amdexperimental, amdproduction]
3045
agent_pool: mi355_4
3137
- # grade: Blocking
3046
timeout_in_minutes: 30
3047
working_dir: "/vllm-workspace/tests"
3048
num_gpus: 4
@@ -3148,7 +3056,6 @@ steps:
3056
- label: DP EP NixlConnector PD accuracy tests (Distributed) # 15min
3057
mirror_hardwares: [amdexperimental, amdproduction]
3058
agent_pool: mi355_4
3151
- # grade: Blocking
3059
timeout_in_minutes: 15
3060
working_dir: "/vllm-workspace/tests"
3061
num_gpus: 4
@@ -3165,7 +3072,6 @@ steps:
3072
- label: Distributed Tests (A100) # optional
3073
mirror_hardwares: [amdexperimental]
3074
agent_pool: mi355_4
3168
- # grade: Blocking
3075
gpu: a100
3076
optional: true
3077
num_gpus: 4
@@ -3188,7 +3094,6 @@ steps:
3094
optional: true
3095
mirror_hardwares: [amdexperimental]
3096
agent_pool: mi355_4
3191
- # grade: Blocking
3097
num_gpus: 4
3098
working_dir: "/vllm-workspace/.buildkite/lm-eval-harness"
3099
source_file_dependencies:
@@ -3204,7 +3109,6 @@ steps:
3109
optional: true
3110
mirror_hardwares: [amdexperimental]
3111
agent_pool: mi355_4
3207
- # grade: Blocking
3112
num_gpus: 4
3113
working_dir: "/vllm-workspace/.buildkite/lm-eval-harness"
3114
source_file_dependencies:
@@ -3219,7 +3123,6 @@ steps:
3123
- label: Distributed Tests (H200) # optional
3124
mirror_hardwares: [amdexperimental]
3125
agent_pool: mi355_2
3222
- # grade: Blocking
3126
gpu: h200
3127
optional: true
3128
working_dir: "/vllm-workspace/"
@@ -3254,7 +3157,6 @@ steps:
3157
timeout_in_minutes: 20
3158
mirror_hardwares: [amdexperimental, amdproduction]
3159
agent_pool: mi355_1
3257
- # grade: Blocking
3160
source_file_dependencies:
3161
- csrc/
3162
- vllm/model_executor/layers/quantization
@@ -3264,7 +3166,6 @@ steps:
3166
- label: LM Eval Large Models (4 Card)
3167
mirror_hardwares: [amdexperimental, amdproduction]
3168
agent_pool: mi355_4
3267
- # grade: Blocking
3169
gpu: a100
3170
optional: true
3171
num_gpus: 4
@@ -3304,7 +3205,6 @@ steps:
3205
- label: Prime-RL Integration Test # 15min
3206
mirror_hardwares: [amdexperimental]
3207
agent_pool: mi355_2
3307
- # grade: Blocking
3208
timeout_in_minutes: 30
3209
optional: true
3210
num_gpus: 2
@@ -3319,7 +3219,6 @@ steps:
3219
- label: DeepSeek V2-Lite Accuracy
3220
mirror_hardwares: [amdexperimental, amdproduction]
3221
agent_pool: mi355_4
3322
- # grade: Blocking
3222
timeout_in_minutes: 60
3223
gpu: h100
3224
optional: true
@@ -3331,7 +3230,6 @@ steps:
3230
- label: Qwen3-30B-A3B-FP8-block Accuracy (H100)
3231
mirror_hardwares: [amdexperimental, amdproduction]
3232
agent_pool: mi355_4
3334
- # grade: Blocking
3233
timeout_in_minutes: 60
3234
gpu: h100
3235
optional: true
@@ -3354,7 +3252,6 @@ steps:
3252
timeout_in_minutes: 60
3253
mirror_hardwares: [amdexperimental]
3254
agent_pool: mi355_4
3357
- # grade: Blocking
3255
optional: true
3256
num_gpus: 4
3257
working_dir: "/vllm-workspace"