ggml/hip: fix APU compatibility - soft error handling for hipMemAdviseSetCoarseGrain (#20536)

* ggml/hip: fix APU compatibility - soft error handling for hipMemAdviseSetCoarseGrain

On AMD APU/iGPU devices (unified memory architecture), hipMemAdviseSetCoarseGrain
returns hipErrorInvalidValue because the hint is not applicable to UMA systems.
The previous CUDA_CHECK() call treated this as a fatal error, causing crashes on
APU systems such as AMD Strix Halo (gfx1151).

Fix: treat hipMemAdviseSetCoarseGrain as an optional performance hint - call it
without error checking and clear any resulting error with hipGetLastError().

Also add pre-allocation debug logging (GGML_LOG_DEBUG) to help diagnose memory
issues on APU systems, and store totalGlobalMem in device info.

Context: AMD APUs on Windows are affected by a ROCm runtime bug that limits
hipMallocManaged to ~64GB regardless of available system RAM. A fix has been
submitted upstream: https://github.com/ROCm/rocm-systems/pull/4077

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

* ggml/hip: remove unrelated changes, keep only hipMemAdviseSetCoarseGrain fix

---------

Co-authored-by: moonshadow-25 <moonshadow-25@users.noreply.github.com>
Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com>
This commit is contained in:
MoonShadow
2026-03-16 00:23:58 +08:00
committed by GitHub
parent 559646472d
commit 8b7d340b6f

View File

@@ -124,7 +124,10 @@ static cudaError_t ggml_cuda_device_malloc(void ** ptr, size_t size, int device)
err = cudaMallocManaged(ptr, size);
#if defined(GGML_USE_HIP)
if (err == hipSuccess) {
CUDA_CHECK(cudaMemAdvise(*ptr, size, hipMemAdviseSetCoarseGrain, device));
// hipMemAdviseSetCoarseGrain is an optional performance hint;
// ignore errors (e.g. hipErrorInvalidValue on some APU/iGPU configs).
cudaMemAdvise(*ptr, size, hipMemAdviseSetCoarseGrain, device);
(void)hipGetLastError(); // clear any error
}
// fall back to cudaMalloc if not supported (e.g. on Windows)