[ROCm][CI] Expand trunk.yml coverage for ROCm (pytorch#168162)

We are expanding the test coverage on pre-submit (PR-based) trunk.yml runs for ROCm to the full list of unit tests.

Consequently, we are swapping the labels (CSPs) for the rocm-mi300.yml and periodic-rocm-mi300.yml workflows to balance capacity concerns.

We will be disabling the shadow workflow trunk-rocm-mi300.yml as it is not required due to this PR anymore.

Fixes pytorch#166108

Pull Request resolved: pytorch#168162
Approved by: https://github.com/jeffdaily

Author: jithunnair-amd

.github/workflows/periodic-rocm-mi300.yml

@@ -60,9 +60,9 @@ jobs:
       docker-image-name: ci-image:pytorch-linux-jammy-rocm-n-py3
       test-matrix: |
         { include: [
-          { config: "distributed", shard: 1, num_shards: 3, runner: "linux.rocm.gpu.gfx942.4", owners: ["module:rocm", "oncall:distributed"] },
-          { config: "distributed", shard: 2, num_shards: 3, runner: "linux.rocm.gpu.gfx942.4", owners: ["module:rocm", "oncall:distributed"] },
-          { config: "distributed", shard: 3, num_shards: 3, runner: "linux.rocm.gpu.gfx942.4", owners: ["module:rocm", "oncall:distributed"] },
+          { config: "distributed", shard: 1, num_shards: 3, runner: "linux.rocm.gpu.gfx942.4.b", owners: ["module:rocm", "oncall:distributed"] },
+          { config: "distributed", shard: 2, num_shards: 3, runner: "linux.rocm.gpu.gfx942.4.b", owners: ["module:rocm", "oncall:distributed"] },
+          { config: "distributed", shard: 3, num_shards: 3, runner: "linux.rocm.gpu.gfx942.4.b", owners: ["module:rocm", "oncall:distributed"] },
         ]}
     secrets: inherit
 

.github/workflows/rocm-mi300.yml

@@ -48,12 +48,12 @@ jobs:
       docker-image-name: ci-image:pytorch-linux-noble-rocm-n-py3
       test-matrix: |
         { include: [
-          { config: "default", shard: 1, num_shards: 6, runner: "linux.rocm.gpu.gfx942.1" },
-          { config: "default", shard: 2, num_shards: 6, runner: "linux.rocm.gpu.gfx942.1" },
-          { config: "default", shard: 3, num_shards: 6, runner: "linux.rocm.gpu.gfx942.1" },
-          { config: "default", shard: 4, num_shards: 6, runner: "linux.rocm.gpu.gfx942.1" },
-          { config: "default", shard: 5, num_shards: 6, runner: "linux.rocm.gpu.gfx942.1" },
-          { config: "default", shard: 6, num_shards: 6, runner: "linux.rocm.gpu.gfx942.1" },
+          { config: "default", shard: 1, num_shards: 6, runner: "linux.rocm.gpu.gfx942.1.b" },
+          { config: "default", shard: 2, num_shards: 6, runner: "linux.rocm.gpu.gfx942.1.b" },
+          { config: "default", shard: 3, num_shards: 6, runner: "linux.rocm.gpu.gfx942.1.b" },
+          { config: "default", shard: 4, num_shards: 6, runner: "linux.rocm.gpu.gfx942.1.b" },
+          { config: "default", shard: 5, num_shards: 6, runner: "linux.rocm.gpu.gfx942.1.b" },
+          { config: "default", shard: 6, num_shards: 6, runner: "linux.rocm.gpu.gfx942.1.b" },
         ]}
     secrets: inherit
 

.github/workflows/trunk.yml

@@ -203,9 +203,15 @@ jobs:
       sync-tag: rocm-build
       test-matrix: |
         { include: [
-          { config: "default", shard: 1, num_shards: 2, runner: "linux.rocm.gpu.gfx942.1" },
-          { config: "default", shard: 2, num_shards: 2, runner: "linux.rocm.gpu.gfx942.1" },
-          { config: "distributed", shard: 1, num_shards: 1, runner: "linux.rocm.gpu.gfx942.4" },
+          { config: "default", shard: 1, num_shards: 6, runner: "linux.rocm.gpu.gfx942.1" },
+          { config: "default", shard: 2, num_shards: 6, runner: "linux.rocm.gpu.gfx942.1" },
+          { config: "default", shard: 3, num_shards: 6, runner: "linux.rocm.gpu.gfx942.1" },
+          { config: "default", shard: 4, num_shards: 6, runner: "linux.rocm.gpu.gfx942.1" },
+          { config: "default", shard: 5, num_shards: 6, runner: "linux.rocm.gpu.gfx942.1" },
+          { config: "default", shard: 6, num_shards: 6, runner: "linux.rocm.gpu.gfx942.1" },
+          { config: "distributed", shard: 1, num_shards: 3, runner: "linux.rocm.gpu.gfx942.4" },
+          { config: "distributed", shard: 2, num_shards: 3, runner: "linux.rocm.gpu.gfx942.4" },
+          { config: "distributed", shard: 3, num_shards: 3, runner: "linux.rocm.gpu.gfx942.4" },
         ]}
     secrets: inherit
 
@@ -223,7 +229,6 @@ jobs:
       build-environment: linux-jammy-rocm-py3.10
       docker-image: ${{ needs.linux-jammy-rocm-py3_10-build.outputs.docker-image }}
       test-matrix: ${{ needs.linux-jammy-rocm-py3_10-build.outputs.test-matrix }}
-      tests-to-include: "test_nn test_torch test_cuda test_ops test_unary_ufuncs test_binary_ufuncs test_autograd inductor/test_torchinductor distributed/test_c10d_common distributed/test_c10d_nccl"
     secrets: inherit
 
   inductor-build: