-
Hi, I am trying to build this library on Stampede2, which is essentially a large-scale cluster of Intel KNL nodes.
I follow the provided steps and run into the following error:
```
login2.stampede2…
-
Hello,
I found out that the following (OpenMP) code generate a sub-optimal code:
```lang-cpp
float compute(const float* v, const uint8_t* m, int size) {
float vsum = 0.0f;
int vcount …
-
AdaptiveCpp would be more usable if it was packaged in linux distributions.
I am considering it for Fedora because there is distro support for at least the OpenMP and HIP backends.
The big change …
-
```
Provide an OpenMP backend like Thrust's
```
Original issue reported on code.google.com by `wnbell` on 6 Aug 2010 at 9:56
-
```
Provide an OpenMP backend like Thrust's
```
Original issue reported on code.google.com by `wnbell` on 6 Aug 2010 at 9:56
-
```
Provide an OpenMP backend like Thrust's
```
Original issue reported on code.google.com by `wnbell` on 6 Aug 2010 at 9:56
-
### Describe the issue
Similarly to #428, I tried `torch.linalg.eigh` on a Max Series GPU using the Intel Devcloud and packages from the intel conda channel, the performance on XPU is not much bett…
-
I get this error when trying to install py3nj:
'''
pip install --no-cache-dir py3nj
Collecting py3nj
Downloading py3nj-0.2.1.tar.gz (49 kB)
━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 50.0/50.0 kB 6.…
-
We should list what classifier the user has loaded (x86, avx2, avx512)
```
o.n.l.f.Nd4jBackend - Loaded [CpuBackend] backend
o.n.n.NativeOpsHolder - Number of threads used for linear algebra: 6
o.…
-
first of all. thanks for your cool work.
I run the darknet_video.py . and it give me inference time about 14s I change darknet_video.py to print time not fps. one frame. Is this normal?
*…