Add CUDA 13.2 support to CI and release workflows (#5610)

Summary:
Pull Request resolved: #5610

X-link: https://github.com/facebookresearch/FBGEMM/pull/2563

Add CUDA 13.2.0 to the FBGEMM GPU CI and release build matrices. Default publish version remains 13.0.2.

Changes:
- generate_ci_matrix.py: Add 13.2.0 to cuda_versions() for default and genai targets
- fbgemm_gpu_release_cuda.yml: Add 13.2.0 to release options
- fbgemm_gpu_release_genai.yml: Add 13.2.0 to genai release options and build/test matrices

Differential Revision: D100229467

Author: gchalump

.github/scripts/generate_ci_matrix.py

@@ -304,10 +304,10 @@ def cuda_versions(self) -> List[str]:
             # FBGEMM HSTU is expensive, so conserve CI resources
             return ["12.8.1"]
         elif self.target == TARGET_GENAI:
-            return ["12.6.3", "12.8.1", "12.9.1", "13.0.2"]
+            return ["12.6.3", "12.8.1", "12.9.1", "13.0.2", "13.2.0"]
         else:
             # GenAI is unable to support 11.8.0 anymore as of https://github.com/pytorch/FBGEMM/pull/4138
-            return ["12.6.3", "12.8.1", "12.9.1", "13.0.2"]
+            return ["12.6.3", "12.8.1", "12.9.1", "13.0.2", "13.2.0"]
 
     def rocm_versions(self) -> List[str]:
         if GitRepo.ref() == REFS_MAIN and GitRepo.event_name() == EVENT_NAME_PUSH:

.github/workflows/fbgemm_gpu_release_cuda.yml

@@ -34,7 +34,7 @@ on:
         description: CUDA Version to Use for Building Artifact
         type: choice
         required: false
-        options: [ "12.6.3", "12.8.1", "12.9.1", "13.0.2" ]
+        options: [ "12.6.3", "12.8.1", "12.9.1", "13.0.2", "13.2.0" ]
         default: "13.0.2"
       publish-to-pypi:
         description: Publish Artifact to PyPI

.github/workflows/fbgemm_gpu_release_genai.yml

@@ -34,7 +34,7 @@ on:
         description: CUDA Version to Use for Building Artifact
         type: choice
         required: false
-        options: [ "12.6.3", "12.8.1", "12.9.1", "13.0.2" ]
+        options: [ "12.6.3", "12.8.1", "12.9.1", "13.0.2", "13.2.0" ]
         default: "13.0.2"
       publish-to-pypi:
         description: Publish Artifact to PyPI
@@ -72,7 +72,7 @@ jobs:
           { arch: x86, instance: "linux.12xlarge.memory" },
         ]
         python-version: [ "3.10", "3.11", "3.12", "3.13", "3.14" ]
-        cuda-version: [ "12.6.3", "12.8.1", "13.0.2" ]
+        cuda-version: [ "12.6.3", "12.8.1", "13.0.2", "13.2.0" ]
 
     steps:
     - name: Setup Build Container
@@ -146,7 +146,7 @@ jobs:
           { arch: x86, instance: "linux.g5.4xlarge.nvidia.gpu" },
         ]
         python-version: [ "3.10", "3.11", "3.12", "3.13", "3.14" ]
-        cuda-version: [ "12.6.3", "12.8.1", "13.0.2" ]
+        cuda-version: [ "12.6.3", "12.8.1", "13.0.2", "13.2.0" ]
     needs: build_artifact
 
     steps: