-
**ISIS version(s) affected**: all
No LSB modules are available.
Distributor ID: Ubuntu
Description: Ubuntu 22.04.5 LTS
Release: 22.04
Codename: jammy
**Description**
There…
-
Consider implementing the Liger Kernels which has shown to yield large memory savings.
- RoPE: 3X speedup with ~3X peak memory reduction.
- SwiGLU: 1.5X peak memory reduction
- Cross Entropy: >4X…
-
### Required prerequisites
- [x] Search the [issue tracker](https://github.com/NVIDIA/cuda-quantum/issues) to check if your feature has already been mentioned or rejected in other issues.
### Descri…
-
Refractor the code to change all CUDA kernels into Kernel Abstraction kernels, rerun the tests and benchmarks .
-
# Feature Request
**Describe the Feature Request**
Hello! I understand the overall computation would not be cost-efficient, but I was wondering if there are any plans to expand the functionality t…
-
### 🐛 Describe the bug
It is found that the vectorized kernels are not performing well. For instance, copy_() is just utilizing only 40-50% of the theoretical bandwidth.
### Versions
Collecting en…
-
-
**Describe the bug**
I cant extract many kernels of velocity tendencies because of the following error:
I use the following function to cut out kernels, where I call the following function for every…
-
### Description
We are currently using `gather` (take_unchecked) kernels from `polars-arrow`. We should add and use optimized gather kernels to `polars-compute` instead.
-
> there is an issue that I hadn't anticipated with removing two of the extended mesh PSyKAl-lite kernels.
> Both of those kernels loop over halo cells, and also use stencils. Given the kernel metad…