viable/strict/1787807055: Move expandable_segments() out of the header (#194872)
- PyTorch 近 90 天出现 126 次
- PyTorch 近 90 天第 125 次发版
- 上一次：同一天稍早 · trunk/a5f76a0e8ca67c5ef3c095061187dd7c318e7197: [cuDNN] Add guards for cuDNN SDPA decode (#194927)
发生了什么
Summary CUDAAllocatorConfig::expandable_segments() branches on PYTORCH_C10_DRIVER_API_SUPPORTED , which c10/cuda/CMakeLists.txt defines PRIVATE to c10_cuda . Being inline, the function recompiles in every translation unit that includes the header without that macro, so consumers see an unconditional false while the library itself sees the real answer. CUDACachingAllocatorReserveDeviceTest hits exactly this — it enab…
摘要按规则整理自下方来源原文