-
### Problem Description
I have a Ubuntu 22.04 machine with two AMD MI100 GPUs installed. When trying to run a PyTorch training script, using DDP and `backend="nccl"` (which under the hood should us…
-
# Why asynchronous Rust doesn't work
In 2017, I said that “asynchronous Rust programming is a disaster and a mess”. In 2021a lot more of the Rust ecosystem has become asynchronous – such that it...
…
-
### Search before asking
- [X] I have searched the YOLOv8 [issues](https://github.com/ultralytics/ultralytics/issues) and found no similar bug report.
### YOLOv8 Component
_No response_
### Bug
…
-
As with [the Radxa CM3](https://github.com/geerlingguy/raspberry-pi-pcie-devices/issues/327), I would also like to test Pine64's [SOQuartz](https://wiki.pine64.org/wiki/SOQuartz) with some CM4 boards,…
-
微博内容精选
-
Hi
the official docker images of pytorch and tf docker are avialble only for gfx900(Vega10-type GPU - MI25, Vega56, Vega64), gfx906 (Vega20-type GPU - MI50, MI60) and gfx908 (MI100), gfx90a (MI200) …
-
Parent Issue: #9106
Current status:
- [x] BFLOAT16 -> UINT32 - Completed
- [x] BFLOAT16 -> UINT16 - Completed
- [x] UINT16 -> BFLOAT16 - Completed
- [x] INT32 -> BFLOAT16 - Completed
- [x] BFL…
-
### Search before asking
- [X] I have searched the YOLOv8 [issues](https://github.com/ultralytics/ultralytics/issues) and found no similar bug report.
### YOLOv8 Component
Train
### Bug
I am try…
Nuna7 updated
7 months ago
-
QubesOS is the most secure operating system available, by far. However, it unfortunately only runs on the x86 instruction set, which runs on unauditable and insecure firmware. The Power Architecture…
-
I observed that in the is_fp16_supported function in cuda_accelerator.py, the method to determine whether fp16 can be used is to obtain the computing power of CUDA and determine whether it is greater …