diff --git a/docker-compose.rocm.yml b/docker-compose.rocm.yml index 63a562e2..17680bbe 100644 --- a/docker-compose.rocm.yml +++ b/docker-compose.rocm.yml @@ -34,3 +34,15 @@ services: # Tune the ROCm memory allocator - PYTORCH_HIP_ALLOC_CONF=garbage_collection_threshold:0.8,max_split_size_mb:512 + + # Redirect MIOpen kernel cache to a writable, persistent directory. + # Without this, MIOpen may fail to write its cache and throw + # miopenStatusUnknownError on fresh containers. + - MIOPEN_USER_DB_PATH=/app/data/cache/miopen_db + - MIOPEN_CUSTOM_CACHE_DIR=/app/data/cache/miopen_cache + + # Use fast heuristics for kernel selection instead of exhaustive + # benchmarking. On RDNA4, exhaustive mode tries kernels that fail to + # allocate workspace memory (ptr: 0 size: 0), causing system stuttering + # on every generation even when the cache is present. + - MIOPEN_FIND_MODE=FAST