-
Hi, apologies if this is mentioned anywhere in the docs but I couldn't find the answers to some questions I had about hipSYCL:
- AMD has notoriously poor support for ROCm on their consumer GPUs. Wh…
-
**Describe the bug**
I'm conducting experiments with opt-1.3B, and gpt-neo2.7b on wikitext2, with the official example from huggingface and deepspeed.
What I observed is that the accuracy and ppl ar…
-
Now that Apple [deprecated OpenCL support in `osx` 10.12]() and [NVIDIA will no longer provide CUDA support for `osx` after CUDA 10.2](https://docs.nvidia.com/cuda/cuda-toolkit-release-notes/index.htm…
-
# Issue
It appears that my primary adapter got changed to my secondary card which has no display cable nor drivers loaded (scroll down to https://github.com/NixOS/nixpkgs/issues/228586#issuecomment-1…
-
### Checklist
- [X] 1. I have searched related issues but cannot get the expected help.
- [X] 2. The bug has not been fixed in the latest version.
### Describe the bug
lmdeploy error when using pyt…
-
I can see that the generated Triton code results in inaccurate numerics for both supported backends - i.e., `cuda` and `hip`.
The original, `03-matrix-multiplication.py` compares numerics using `to…
-
build env : ubuntu 22.04 + cmake 3.29.3 (distro cmake version was not enough for build system)
Process :
```
git clone https://github.com/lamikr/rocm_sdk_builder.git
cd rocm_sdk_builder
git ch…
-
### 🐛 Describe the bug
On nightly build, when using torch.compile on a user defined triton kernel embedded in a custom autograd.Function, pytorch reports that `grad_output` was mutated even though …
-
**Summary**
CUDA's kernel launch mechanism requires each kernel's device stub function to have a unique address. When targeting Windows, the linker defaults to performing identical COMDAT folding (…
-
Since the document says that Antares supports ROCm for Windows, I wonder whether this kind of thing exists in the world? The ROCm has been officially declared that there's no plans for Windows!