-
No problem
Would like to be able to use AMD EOCm HIP support to use AMD GPUs like Nvidia GPUS for machine learning
[Gogle Search of AMD EOCm HIP](https://www.google.com/search?q=AMD+EOCm+HIP&oq=…
-
We already use ECR (AWS) for publishing our docker images. I would like to use the same repository for publishing Helm charts.
Right now, OCI support is experimental in Helm but works already quite w…
-
According to the [Readme](https://github.com/crackededed/Xtra/blob/master/README.md), the Android 4.1 and Android 5 apk's use separate download pages... but~ if there's no way to actually know which v…
-
When I compile oneDNN source files using make -j 20,I get error as following:
error: no matching function for call to 'get_native'
error: no matching function for call to 'get_native_mem'
I ask a…
-
CUDA is a nvidia only acceleration solution, provide a mean to support acceleration on AMD GPUs
For instance, consider migrating the CUDA code using https://github.com/ROCm-Developer-Tools/HIP. It …
-
@tboch - I see you gave an example of a FITS WCS for a given FITS HIPS tile:
https://hips.readthedocs.io/en/latest/drawing_algo.html#wcs-for-fits-tiles
Can you describe how it could be computed fo…
cdeil updated
5 years ago
-
Please look at the attached image. At HIP API activity record, we have executed hipMemcpyDtoD but in ACTIVITY_DOMAIN_HIP_OPS it shows a CopyHostToDevice (kind = 4596).
![Capture](https://user-images…
-
In PR https://github.com/pmodels/mpich/pull/6451, ipc read bcast and alltoall requires MPL_gpu_imemcpy to move the data. But this function is only implemented in mpl_gpu_ze.c (no implementation of thi…
-
When managing a large decentralized collaboration, one of the most common client problems is the need to force a specific GPU by ID - to avoid an iGPU, a hot card, etc.. It would be _very_ useful for …
-
Hi, I'm experiencing hangs with `hipFreeAsync` and was wondering what could potentially cause that.
From my perspective it looks like some kind of racing condition.
It consistently happens at the …