Daniel Tang
Daniel Tang
> GCVM_L2_PROTECTION_FAULT_STATUS ~~Duplicate~~ Potential duplicate of my #2596 They said it's caused by a bug in the CPU page table update code. You can either revert the commit I mentioned...
What does exist is `/opt/rocm-5.7.0/lib/rocblas/library/TensileLibrary_lazy_gfx1030.dat`. Have you tried `sudo env HSA_OVERRIDE_GFX_VERSION=10.3.0 ~/llama.cpp/build/bin/main`?
> Stable diffusion Try: ```bash sudo apt install wget git python3 python3-venv libgl1 libglib2.0-0 sudo apt install libtcmalloc4 libgthread-2_0-0 git clone https://github.com/AUTOMATIC1111/stable-diffusion-webui.git cd stable-diffusion-webui git checkout dev rm -rf venv...
The situation is worse on Linux 6.6-rc7: ``` Oct 23 07:17:04 daniel-desktop3 kernel: amdgpu 0000:2f:00.0: amdgpu: bo 00000000b0b9bc73 va 0x0800000000-0x0800000001 conflict with 0x0800000000-0x0800000200 Oct 23 07:17:04 daniel-desktop3 kernel: amdgpu: Failed...
I bisected this, resulting in https://lists.freedesktop.org/archives/amd-gfx/2023-October/100298.html
@xstraok We are still waiting for AMD to reply to my thread on that mailing list. I said in the commit message "Although [...] were working in v6.2 and ROCm...
Interesting that their bisect finished on a different commit. I haven't gotten around to applying it yet, but does their `git revert 96c211f1f9ef82183493f4ceed4e347b52849149` solution (maybe start with a clean kernel...
I found a workaround! I got things to work without crashing with `GRUB_CMDLINE_LINUX_DEFAULT="quiet splash amdgpu.runpm=0 amdgpu.vm_update_mode=3"` in `/etc/default/grub`. I updated to ROCM 6.0. Thank you @terryrankine for https://github.com/ROCm/ROCm/issues/2766#issuecomment-1867179386 . Btw,...
@pLuca70 `lspci -nn`? This should have been already fixed in tag 1.4 at 15521706464b78dc9af60495b648b9d536b4d085. Try updating to the latest os major version. I tried guessing your pci id to be...
Hi @GreyXor, more testers are always more help! What you can do immediately is delete the `&& fam < VEGAM` from the following code and report back whether it works...