add srun and sbatch configs

richardhuo-nv · richardhuo-nv · commit 48c05c0395ed · 2026-06-24T15:47:28.000-07:00
diff --git a/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3-gb300-fp8/1k1k/1p1d-dep2-dep4-1k1k.yaml b/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3-gb300-fp8/1k1k/1p1d-dep2-dep4-1k1k.yaml
@@ -18,6 +18,13 @@ health_check:
   max_attempts: 720
   interval_seconds: 10
 
+sbatch_directives:
+  mem: "0"
+  cpus-per-task: "72"
+
+srun_options:
+  mem: "0"
+
 resources:
   gpu_type: "gb300"
   gpus_per_node: 4
diff --git a/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3-gb300-fp8/1k1k/1p2d-dep2-tep8-1k1k.yaml b/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3-gb300-fp8/1k1k/1p2d-dep2-tep8-1k1k.yaml
@@ -18,6 +18,13 @@ health_check:
   max_attempts: 720
   interval_seconds: 10
 
+sbatch_directives:
+  mem: "0"
+  cpus-per-task: "72"
+
+srun_options:
+  mem: "0"
+
 resources:
   gpu_type: "gb300"
   gpus_per_node: 4
diff --git a/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3-gb300-fp8/1k1k/2p2d-dep2-tep8-1k1k.yaml b/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3-gb300-fp8/1k1k/2p2d-dep2-tep8-1k1k.yaml
@@ -18,6 +18,13 @@ health_check:
   max_attempts: 720
   interval_seconds: 10
 
+sbatch_directives:
+  mem: "0"
+  cpus-per-task: "72"
+
+srun_options:
+  mem: "0"
+
 resources:
   gpu_type: "gb300"
   gpus_per_node: 4
diff --git a/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3-gb300-fp8/1k1k/2p3d-dep2-dep4-1k1k.yaml b/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3-gb300-fp8/1k1k/2p3d-dep2-dep4-1k1k.yaml
@@ -18,6 +18,13 @@ health_check:
   max_attempts: 720
   interval_seconds: 10
 
+sbatch_directives:
+  mem: "0"
+  cpus-per-task: "72"
+
+srun_options:
+  mem: "0"
+
 resources:
   gpu_type: "gb300"
   gpus_per_node: 4
diff --git a/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3-gb300-fp8/1k1k/2p4d-dep2-dep4-1k1k.yaml b/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3-gb300-fp8/1k1k/2p4d-dep2-dep4-1k1k.yaml
@@ -18,6 +18,13 @@ health_check:
   max_attempts: 720
   interval_seconds: 10
 
+sbatch_directives:
+  mem: "0"
+  cpus-per-task: "72"
+
+srun_options:
+  mem: "0"
+
 resources:
   gpu_type: "gb300"
   gpus_per_node: 4
diff --git a/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3-gb300-fp8/1k1k/4p2d-dep2-dep8-1k1k.yaml b/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3-gb300-fp8/1k1k/4p2d-dep2-dep8-1k1k.yaml
@@ -18,6 +18,13 @@ health_check:
   max_attempts: 720
   interval_seconds: 10
 
+sbatch_directives:
+  mem: "0"
+  cpus-per-task: "72"
+
+srun_options:
+  mem: "0"
+
 resources:
   gpu_type: "gb300"
   gpus_per_node: 4
diff --git a/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3-gb300-fp8/8k1k/1p1d-dep2-dep8-8k1k.yaml b/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3-gb300-fp8/8k1k/1p1d-dep2-dep8-8k1k.yaml
@@ -18,6 +18,13 @@ health_check:
   max_attempts: 720
   interval_seconds: 10
 
+sbatch_directives:
+  mem: "0"
+  cpus-per-task: "72"
+
+srun_options:
+  mem: "0"
+
 resources:
   gpu_type: "gb300"
   gpus_per_node: 4
diff --git a/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3-gb300-fp8/8k1k/1p1d-dep2-tep8-8k1k.yaml b/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3-gb300-fp8/8k1k/1p1d-dep2-tep8-8k1k.yaml
@@ -18,6 +18,13 @@ health_check:
   max_attempts: 720
   interval_seconds: 10
 
+sbatch_directives:
+  mem: "0"
+  cpus-per-task: "72"
+
+srun_options:
+  mem: "0"
+
 resources:
   gpu_type: "gb300"
   gpus_per_node: 4
diff --git a/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3-gb300-fp8/8k1k/1p2d-dep2-tep8-8k1k.yaml b/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3-gb300-fp8/8k1k/1p2d-dep2-tep8-8k1k.yaml
@@ -18,6 +18,13 @@ health_check:
   max_attempts: 720
   interval_seconds: 10
 
+sbatch_directives:
+  mem: "0"
+  cpus-per-task: "72"
+
+srun_options:
+  mem: "0"
+
 resources:
   gpu_type: "gb300"
   gpus_per_node: 4
diff --git a/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3-gb300-fp8/8k1k/2p1d-dep2-dep8-8k1k.yaml b/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3-gb300-fp8/8k1k/2p1d-dep2-dep8-8k1k.yaml
@@ -18,6 +18,13 @@ health_check:
   max_attempts: 720
   interval_seconds: 10
 
+sbatch_directives:
+  mem: "0"
+  cpus-per-task: "72"
+
+srun_options:
+  mem: "0"
+
 resources:
   gpu_type: "gb300"
   gpus_per_node: 4
diff --git a/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3-gb300-fp8/8k1k/2p2d-dep2-tep8-8k1k.yaml b/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3-gb300-fp8/8k1k/2p2d-dep2-tep8-8k1k.yaml
@@ -18,6 +18,13 @@ health_check:
   max_attempts: 720
   interval_seconds: 10
 
+sbatch_directives:
+  mem: "0"
+  cpus-per-task: "72"
+
+srun_options:
+  mem: "0"
+
 resources:
   gpu_type: "gb300"
   gpus_per_node: 4
diff --git a/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3-gb300-fp8/8k1k/2p4d-dep2-tep4-8k1k.yaml b/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3-gb300-fp8/8k1k/2p4d-dep2-tep4-8k1k.yaml
@@ -18,6 +18,13 @@ health_check:
   max_attempts: 720
   interval_seconds: 10
 
+sbatch_directives:
+  mem: "0"
+  cpus-per-task: "72"
+
+srun_options:
+  mem: "0"
+
 resources:
   gpu_type: "gb300"
   gpus_per_node: 4
diff --git a/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3-gb300-fp8/8k1k/3p1d-dep2-dep8-8k1k.yaml b/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3-gb300-fp8/8k1k/3p1d-dep2-dep8-8k1k.yaml
@@ -18,6 +18,13 @@ health_check:
   max_attempts: 720
   interval_seconds: 10
 
+sbatch_directives:
+  mem: "0"
+  cpus-per-task: "72"
+
+srun_options:
+  mem: "0"
+
 resources:
   gpu_type: "gb300"
   gpus_per_node: 4
diff --git a/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3-gb300-fp8/8k1k/3p2d-dep2-dep8-8k1k.yaml b/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3-gb300-fp8/8k1k/3p2d-dep2-dep8-8k1k.yaml
@@ -18,6 +18,13 @@ health_check:
   max_attempts: 720
   interval_seconds: 10
 
+sbatch_directives:
+  mem: "0"
+  cpus-per-task: "72"
+
+srun_options:
+  mem: "0"
+
 resources:
   gpu_type: "gb300"
   gpus_per_node: 4
diff --git a/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3-gb300-fp8/8k1k/6p1d-dep2-dep8-8k1k.yaml b/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3-gb300-fp8/8k1k/6p1d-dep2-dep8-8k1k.yaml
@@ -18,6 +18,13 @@ health_check:
   max_attempts: 720
   interval_seconds: 10
 
+sbatch_directives:
+  mem: "0"
+  cpus-per-task: "72"
+
+srun_options:
+  mem: "0"
+
 resources:
   gpu_type: "gb300"
   gpus_per_node: 4
