-
When specify custom JDK path with asadmin file, asadmin doesn't work.
For example. My %JAVA_HOME% is set to **JDK 15**. For my best expirience, I have some environment variables, such as %JAVA8_HOME%…
-
Description
===========
This is related to https://github.com/arrayfire/arrayfire/issues/2938
Error Log OpenCL
---------
```
/Users/turgutabdullayev/Documents/arrayfire/src/backend/ope…
-
Hi:
Thank you for the model and for open sourcing the code. I am looking forward to using the same.
I am trying to reproduce the results of models in recipes/models/sota/2019/librispeech but I …
-
Description
===========
Hi. I cloned the repo again. But now I get the following. ~Related to https://github.com/arrayfire/arrayfire/issues/2951~
Error Log
---------
```
[ 8%] Buildin…
-
Hi,
Building `afopencl` target on x64 MSVS 2017 is failing at master d2df83d (5 June 2020) with a bunch of errors relating to "declaration is not allowed in 'constexpr' function body". The culprit …
-
I use cuda10.2 Arrayfire3.6.1 it recommends cuda 9.2. But i wanna try cuda 10.2.
So when i run
$ wav2letter/build/Train train --flagsfile wav2letter/tutorials/1-librispeech_clean/train.cfg
it ou…
-
Hello
This is 'just' a compilation error with todays master when building AF in a build folder not called 'build' (eg 'release', or 'debug'):
...
[ 23%] Built target af_cuda_static_cuda_library
In…
-
I use the following driver for my card with mesa-opencl-icd loader and radeon driver.
```
0d:00.0 Display controller: Advanced Micro Devices, Inc. [AMD/ATI] Sun XT [Radeon HD 8670A/8670M/8690M / R5 …
-
### `Conv1D` logic is implemented differently in fbgemm vs LibTorch/PyTorch/TensorFlow/Keras
TL;DR: FbGemm implementation of `Conv1D` keeps only one weight matrix overall vs one weight matrix per g…
-
Hi,
I am configuring arrayfire with OpenCL provided by Cuda, i.e.
`-DOPENCL_LIBRARIES:FILEPATH=/cvmfs/soft.computecanada.ca/easybuild/software/2017/avx2/Compiler/intel2016.4/cuda/9.0.176/lib64/libO…