leejet / stable-diffusion.cpp

Stable Diffusion in pure C/C++
MIT License
2.9k stars 232 forks source link

tried rocm precompiled with amd gpu, it gives cuda error ? #292

Open patientx opened 2 weeks ago

patientx commented 2 weeks ago

RX 6600. Windows 10. Hip 5.7.1 installed. hipinfo correctly identifies my gpu. (gfx1032) Yet when I run sd.exe with correct example, from "sd-master-9c51d87-bin-win-rocm5.5-x64.zip" first it says :

' [DEBUG] stable-diffusion.cpp:149 - Using CUDA backend ggml_cuda_init: GGML_CUDA_FORCE_MMQ: no ggml_cuda_init: CUDA_USE_TENSOR_CORES: yes ggml_cuda_init: found 1 ROCm devices: Device 0: AMD Radeon RX 6600, compute capability 10.3, VMM: no '

and in the end

'ggml_cuda_compute_forward: GET_ROWS failed CUDA error: invalid device function current device: 0, in function ggml_cuda_compute_forward at D:/a/stable-diffusion.cpp/stable-diffusion.cpp/ggml/src/ggml-cuda.cu:2360 err GGML_ASSERT: D:/a/stable-diffusion.cpp/stable-diffusion.cpp/ggml/src/ggml-cuda.cu:100: !"CUDA error" '

Punos805 commented 15 hours ago

RX 6600 is Target gfx1032. gfx1032 is not on the list. https://github.com/ROCm/rocBLAS/blob/develop/CMakeLists.txt (line 110) If you upgrade to RX 7800 XT, wait until "sd-master-id-bin-win-rocm6.1-x64.zip" file come out. because 7800 XT dose not works on ROCm 5.5.

ROCm 5.5.1 https://rocm.docs.amd.com/en/docs-5.5.1/release/windows_support.html Name Architecture LLVM Target Runtime HIP SDK AMD Radeon™ RX 7900 XTX RDNA3 gfx1100 ✅ ✅ AMD Radeon™ RX 7900 XT RDNA3 gfx1100 ✅ ✅ AMD Radeon™ RX 7600 RDNA3 gfx1100 ✅ ✅ AMD Radeon™ RX 6950 XT RDNA2 gfx1030 ✅ ✅ AMD Radeon™ RX 6900 XT RDNA2 gfx1030 ✅ ✅ AMD Radeon™ RX 6800 XT RDNA2 gfx1030 ✅ ✅ AMD Radeon™ RX 6800 RDNA2 gfx1030 ✅ ✅ AMD Radeon™ RX 6750 RDNA2 gfx1032 ✅ ❌ AMD Radeon™ RX 6700 XT RDNA2 gfx1032 ✅ ❌ AMD Radeon™ RX 6700 RDNA2 gfx1032 ✅ ❌ AMD Radeon™ RX 6650 XT RDNA2 gfx1032 ✅ ❌ AMD Radeon™ RX 6600 XT RDNA2 gfx1032 ✅ ❌ AMD Radeon™ RX 6600 RDNA2 gfx1032 ✅ ❌

ROCm 6.1.0 https://rocm.docs.amd.com/projects/install-on-windows/en/docs-6.1.0/reference/system-requirements.html Name Architecture LLVM target Runtime HIP SDK AMD Radeon RX 7900 XTX RDNA3 gfx1100 ✅ ✅ AMD Radeon RX 7900 XT RDNA3 gfx1100 ✅ ✅ AMD Radeon RX 7800 XT RDNA3 gfx1101 ✅ ✅ AMD Radeon RX 7700 XT RDNA3 gfx1101 ✅ ✅ AMD Radeon RX 7600 RDNA3 gfx1102 ✅ ✅ AMD Radeon RX 6950 XT RDNA2 gfx1030 ✅ ✅ AMD Radeon RX 6900 XT RDNA2 gfx1030 ✅ ✅ AMD Radeon RX 6800 XT RDNA2 gfx1030 ✅ ✅ AMD Radeon RX 6800 RDNA2 gfx1030 ✅ ✅ AMD Radeon RX 6750 XT RDNA2 gfx1031 ✅ ❌ AMD Radeon RX 6700 XT RDNA2 gfx1031 ✅ ❌ AMD Radeon RX 6700 RDNA2 gfx1031 ✅ ❌ AMD Radeon RX 6650 XT RDNA2 gfx1032 ✅ ❌ AMD Radeon RX 6600 XT RDNA2 gfx1032 ✅ ❌ AMD Radeon RX 6600 RDNA2 gfx1032 ✅ ❌