Skip to content

Commit 6544533

Browse files
amdfaaamdfaa
authored
[ROCm CI] Migrate to MI325 Capacity (#60)
* [ROCm CI] Migrate to MI325 Capacity * Update generate_vllm_benchmark_matrix.py
1 parent 52d3fe9 commit 6544533

2 files changed

Lines changed: 21 additions & 21 deletions

File tree

‎.github/scripts/generate_vllm_benchmark_matrix.py‎

Lines changed: 7 additions & 7 deletions
Original file line numberDiff line numberDiff line change
@@ -16,24 +16,24 @@
1616
1: [
1717
"linux.aws.a100",
1818
"linux.aws.h100",
19-
"linux.rocm.gpu.mi300.2", # No single ROCm GPU?
19+
"linux.rocm.gpu.gfx942.2", # No single ROCm GPU?
2020
"linux.24xl.spr-metal",
2121
],
2222
# NB: There is no 2xH100 runner at the momement, so let's use the next one
2323
# in the list here which is 4xH100
2424
2: [
2525
"linux.aws.h100.4",
26-
"linux.rocm.gpu.mi300.2",
26+
"linux.rocm.gpu.gfx942.2",
2727
],
2828
4: [
2929
"linux.aws.h100.4",
30-
"linux.rocm.gpu.mi300.4",
30+
"linux.rocm.gpu.gfx942.4",
3131
# TODO (huydhn): Enable this when Intel's runners are ready
3232
# "intel-cpu-emr",
3333
],
3434
8: [
3535
"linux.aws.h100.8",
36-
"linux.rocm.gpu.mi300.8",
36+
"linux.rocm.gpu.gfx942.8",
3737
],
3838
}
3939

@@ -43,9 +43,9 @@
4343
"linux.aws.h100": "cuda",
4444
"linux.aws.h100.4": "cuda",
4545
"linux.aws.h100.8": "cuda",
46-
"linux.rocm.gpu.mi300.2": "rocm",
47-
"linux.rocm.gpu.mi300.4": "rocm",
48-
"linux.rocm.gpu.mi300.8": "rocm",
46+
"linux.rocm.gpu.gfx942.2": "rocm",
47+
"linux.rocm.gpu.gfx942.4": "rocm",
48+
"linux.rocm.gpu.gfx942.8": "rocm",
4949
"linux.24xl.spr-metal": "cpu",
5050
}
5151

‎.github/scripts/test_generate_vllm_benchmark_matrix.py‎

Lines changed: 14 additions & 14 deletions
Original file line numberDiff line numberDiff line change
@@ -46,23 +46,23 @@ def test_generate_benchmark_matrix():
4646
"models": "meta-llama/llama-4-maverick-17b-128e-instruct-fp8"
4747
},
4848
{
49-
"runner": "linux.rocm.gpu.mi300.2",
49+
"runner": "linux.rocm.gpu.gfx942.2",
5050
"models": "meta-llama/meta-llama-3.1-8b-instruct"
5151
},
5252
{
53-
"runner": "linux.rocm.gpu.mi300.4",
53+
"runner": "linux.rocm.gpu.gfx942.4",
5454
"models": "meta-llama/meta-llama-3.1-70b-instruct"
5555
},
5656
{
57-
"runner": "linux.rocm.gpu.mi300.2",
57+
"runner": "linux.rocm.gpu.gfx942.2",
5858
"models": "mistralai/mixtral-8x7b-instruct-v0.1"
5959
},
6060
{
61-
"runner": "linux.rocm.gpu.mi300.4",
61+
"runner": "linux.rocm.gpu.gfx942.4",
6262
"models": "meta-llama/llama-4-scout-17b-16e-instruct"
6363
},
6464
{
65-
"runner": "linux.rocm.gpu.mi300.8",
65+
"runner": "linux.rocm.gpu.gfx942.8",
6666
"models": "meta-llama/llama-4-maverick-17b-128e-instruct-fp8"
6767
}
6868
]
@@ -89,7 +89,7 @@ def test_generate_benchmark_matrix():
8989
"models": "meta-llama/meta-llama-3.1-8b-instruct"
9090
},
9191
{
92-
"runner": "linux.rocm.gpu.mi300.2",
92+
"runner": "linux.rocm.gpu.gfx942.2",
9393
"models": "meta-llama/meta-llama-3.1-8b-instruct"
9494
}
9595
]
@@ -123,11 +123,11 @@ def test_generate_benchmark_matrix():
123123
"models": "meta-llama/llama-4-maverick-17b-128e-instruct-fp8"
124124
},
125125
{
126-
"runner": "linux.rocm.gpu.mi300.2",
126+
"runner": "linux.rocm.gpu.gfx942.2",
127127
"models": "meta-llama/meta-llama-3.1-8b-instruct"
128128
},
129129
{
130-
"runner": "linux.rocm.gpu.mi300.8",
130+
"runner": "linux.rocm.gpu.gfx942.8",
131131
"models": "meta-llama/llama-4-maverick-17b-128e-instruct-fp8"
132132
}
133133
]
@@ -154,7 +154,7 @@ def test_generate_benchmark_matrix():
154154
"models": "meta-llama/meta-llama-3.1-8b-instruct"
155155
},
156156
{
157-
"runner": "linux.rocm.gpu.mi300.2",
157+
"runner": "linux.rocm.gpu.gfx942.2",
158158
"models": "meta-llama/meta-llama-3.1-8b-instruct"
159159
}
160160
]
@@ -181,7 +181,7 @@ def test_generate_benchmark_matrix():
181181
"models": "meta-llama/meta-llama-3.1-8b-instruct"
182182
},
183183
{
184-
"runner": "linux.rocm.gpu.mi300.2",
184+
"runner": "linux.rocm.gpu.gfx942.2",
185185
"models": "meta-llama/meta-llama-3.1-8b-instruct"
186186
}
187187
]
@@ -370,11 +370,11 @@ def test_generate_benchmark_matrix():
370370
"models": "meta-llama/meta-llama-3.1-8b-instruct"
371371
},
372372
{
373-
"runner": "linux.rocm.gpu.mi300.2",
373+
"runner": "linux.rocm.gpu.gfx942.2",
374374
"models": "meta-llama/meta-llama-3.1-8b-instruct"
375375
},
376376
{
377-
"runner": "linux.rocm.gpu.mi300.2",
377+
"runner": "linux.rocm.gpu.gfx942.2",
378378
"models": "mistralai/mixtral-8x7b-instruct-v0.1"
379379
}
380380
]
@@ -393,7 +393,7 @@ def test_generate_benchmark_matrix():
393393
{
394394
"include": [
395395
{
396-
"runner": "linux.rocm.gpu.mi300.2",
396+
"runner": "linux.rocm.gpu.gfx942.2",
397397
"models": "meta-llama/meta-llama-3.1-8b-instruct"
398398
}
399399
]
@@ -412,7 +412,7 @@ def test_generate_benchmark_matrix():
412412
{
413413
"include": [
414414
{
415-
"runner": "linux.rocm.gpu.mi300.2",
415+
"runner": "linux.rocm.gpu.gfx942.2",
416416
"models": "meta-llama/meta-llama-3.1-8b-instruct"
417417
}
418418
]

0 commit comments

Comments
 (0)