benchmark

5e0d57f8 - Add CUDA 13.2 CI coverage for inductor, H100, B200, and dtensor (#190948)

Add CUDA 13.2 CI coverage for inductor, H100, B200, and dtensor (#190948) Summary: This is the follow-up to #190641 which added CUDA 13.2 coverage for core trunk/pull/periodic workflows. That PR intentionally left the specialized workflows for a follow-up. This PR adds CUDA 13.2 jobs alongside the existing CUDA 13.0 jobs (not replacing them) in: - inductor.yml, inductor-unittest.yml: inductor-build-cuda132 / inductor-test-cuda132 pairs alongside the existing 13.0 jobs - trunk.yml: inductor-build-cuda132 alongside inductor-build - dtensor.yml: dtensor-build-cuda132 / dtensor-test-cuda132 - test-h100.yml, h100-distributed.yml, h100-symm-mem.yml, h100-cutlass-backend.yml: cuda13.2 build+test pairs alongside 13.0 - b200-distributed.yml, b200-symm-mem.yml: cuda13.2 build+test pairs alongside 13.0 Two new Docker images are added to support the inductor and B200 workflows: - pytorch-linux-jammy-cuda13.2-cudnn9-py3-gcc11-inductor-benchmarks - pytorch-linux-jammy-cuda13.2-cudnn9-py3.12-gcc11 X-link: https://github.com/pytorch/pytorch/pull/190948 Approved by: https://github.com/nWEIdia, https://github.com/atalman Reviewed By: huydhn Differential Revision: D113919116 fbshipit-source-id: 326a1f8fa58d0c563d1a11875e5393fad778391b Co-authored-by: Claude Sonnet 4.6 (1M context) <noreply@anthropic.com> Co-authored-by: Andrey Talman <atalman@fb.com>