rank_zero_deprecation( Launching Web UI with arguments: --use-zluda --medvram-sdxl --update-check --skip-ort ZLUDA device failed to pass basic operation test: index=None, device_name=AMD Radeon RX 5700 XT [ZLUDA] CUDA error: CUBLAS_STATUS_NOT_SUPPORTED when calling `cublasSgemm( handle,...
Additional Information I've installed python-cupy-rocm in Arch linux. I am overriding the version of the graphics card with HSA_OVERRIDE_GFX_VERSION=9.0.0 because it seems that the integrated GPU on laptops are not officially supported.
GGML_CUDA_FORCE_CUBLAS: no ggml_cuda_init: found 1 ROCm devices: Device 0: AMD Radeon Graphics, compute capability 9.0, VMM: no llm_load_tensors: ggml ctx size = 0.24 MiB llm_load_tensors: offloading 28 repeating layers to GPU llm_load_tensors: offloading non-repeating layers to GPU...
hipBLAS exports an interface that does not require the client to change, regardless of the chosen backend. Currently, hipBLAS supports rocBLAS and cuBLAS as backends.AcronymExpansion BLAS Basic Linear Algebra Subprograms ROCm Radeon Open ECosystem HIP Heterogeneous-Compute Interface for Portability...