-
Reviewing the grafana logs for the UAT testing, and performing various test runs to confirm, there seems to be some unusual variation in workload distribution to MPI workers in regards to their CPU us…
-
While our runtime executors run reactions to completion, there is an opportunity (in the presence of an operating system) to leverage the underlying thread scheduler to achieve preemption, assuming th…
-
# Sighting report
When enabling QoS on a bdev, we simply pick the thread of first available bdev channel as the QoS thread. And the QoS ability for the bdev relies on that thread. However, such ran…
-
**Describe the bug**
The application lags when dragging the window around from the title bar.
The application appears to be doing non-trivial work while it's being dragged. With a high polling rat…
-
Without the compiler option: https://github.com/mrakgr/Spiral-s-ML-Library/blob/c5d8a529b210f84dc955a017aeff455c2d27affd/game/leduc/fast_compile.py
With --Ofast-compile=max: https://github.com/mrakgr…
-
If I set LoRA training to CPU in Kohya, will it proceed?
-
### What is the issue?
I use Proxmox VE for virtualization. If I install ollama in a Linux VM it works fine. If I install Ollama in a LXC (Host Kernel 6.8.4-3) it don't works with CPU.
#####
olla…
-
### Describe the issue
Hi,
I've noticed that a significant chunk of time is spent on locks inside `onnxruntime`. Specifically, inside `BFCArena::AllocateRawInternal`
https://github.com/microsoft/…
-
### 🐛 Describe the bug
As per test output should be just doubled. But compile mode it gets 4 times.
### Error logs
_No response_
### Minified repro
```
import torch
@torch.compile(bac…
-
@xqgex Thanks for the code & comments! A great complement to the original paper.
However my server can compile the program but always encounter either SIGSEGV or SIGABRT. Any possible fixes to that?
…