-
### 🐛 Describe the bug
As discussed in https://github.com/pytorch/pytorch/pull/139942, current test of mkldnn pattern matcher checks the total number and nodes of patterns matched which is fragile to…
-
### Suggestion
Extend Spack's compiler spec syntax by a reserved value that accommodates packages that _do not need a compiler_. For concreteness (ahem), let me suggest:
```
%none
```
* Spa…
-
Right now it looks like MKL and MKL headers are not fully linked together with the versions that are allowed. That means I can imagine there being scenarios where a package requests both but has a com…
-
I have used **BLAS** and **LAPACK** in **Intel-MKL** to build **SuiteSparse** and **OpenMVG** for windows. Most 3rd-libs are built from Vcpkg.
I thinck it will be very helpful to implement the featur…
-
### Describe your issue.
I'm trying to use the shgo function from scipy.optimize to minimize a fairly complicated function of two variables. When computing the cost function, several intermediate res…
-
# Problem
Transfer model from pytorch to onnx by torch.onnx.dynamo_export failed.
These failed occured when array slicing inside model and set torch.onnx.ExportOptions(dynamic_shapes=True), it can run…
-
Using SKL caused a MKL Error for me, had to uninstall MKL to use it, would be great to fix this for the future.
-
Intel MKL FATAL ERROR: Cannot load libmkl_intel_thread.dylib.
It's here: /opt/intel/mkl/lib
But, not working for some reason...
-
Hello,
I was planning to make some benchmark with different BLAS implementation on CPU so I built several Pytorch package with BLAS=Eigen/OpenBLAS/ATLAS/MKL.
Sadly something unexpected occurred: A…
-
Most people use the LAPACK and BLAS part of MKL and don't need the full set of libraries.
Uncompressed it takes 741MB for me and the lapack and blas part is 347 MB (`libmkl_avx*.so libmkl_core.so l…