You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@tvm.apache.org by GitBox <gi...@apache.org> on 2020/04/19 02:01:13 UTC
[GitHub] [incubator-tvm] jwfromm opened a new pull request #5376:
[Relay][Strategy] Add cuda target check to dense tensorcore schedule.
jwfromm opened a new pull request #5376: [Relay][Strategy] Add cuda target check to dense tensorcore schedule.
URL: https://github.com/apache/incubator-tvm/pull/5376
Calling `nvcc.have_tensorcore` will cause errors when using a non-nvidia GPU. This one line PR adds a check that the current target is `cuda` before the `nvcc` call. The equivalent `conv2d` and `conv3d` strategies already have this check, but it looks like we forgot to add it to `dense`. Hopefully, this addresses the error found in Issue #5370.
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
With regards,
Apache Git Services
[GitHub] [incubator-tvm] JishinMaster commented on issue #5376:
[Relay][Strategy] Add cuda target check to dense tensorcore schedule.
Posted by GitBox <gi...@apache.org>.
JishinMaster commented on issue #5376: [Relay][Strategy] Add cuda target check to dense tensorcore schedule.
URL: https://github.com/apache/incubator-tvm/pull/5376#issuecomment-616048946
With your PR it worked for me with OpenCL and Vulkan backend.
It still fails with RoCM though, but it might be a different problem since I am using ROCM-3.3 and built TVM with LLVM 9.
Thanks!
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
With regards,
Apache Git Services
[GitHub] [incubator-tvm] jwfromm commented on issue #5376: [Relay][Strategy]
Add cuda target check to dense tensorcore schedule.
Posted by GitBox <gi...@apache.org>.
jwfromm commented on issue #5376: [Relay][Strategy] Add cuda target check to dense tensorcore schedule.
URL: https://github.com/apache/incubator-tvm/pull/5376#issuecomment-616006345
@Shawn-Inspur, @Hzfengsy , @JishinMaster, can you take a look at this?
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
With regards,
Apache Git Services
[GitHub] [incubator-tvm] tqchen commented on issue #5376: [Relay][Strategy]
Add cuda target check to dense tensorcore schedule.
Posted by GitBox <gi...@apache.org>.
tqchen commented on issue #5376: [Relay][Strategy] Add cuda target check to dense tensorcore schedule.
URL: https://github.com/apache/incubator-tvm/pull/5376#issuecomment-616200873
Thanks @jwfromm , @Hzfengsy can you double check if the conv2d case has been fixed, and raise another PR if it is not?
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
With regards,
Apache Git Services
[GitHub] [incubator-tvm] tqchen merged pull request #5376: [Relay][Strategy]
Add cuda target check to dense tensorcore schedule.
Posted by GitBox <gi...@apache.org>.
tqchen merged pull request #5376: [Relay][Strategy] Add cuda target check to dense tensorcore schedule.
URL: https://github.com/apache/incubator-tvm/pull/5376
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
With regards,
Apache Git Services
[GitHub] [incubator-tvm] jwfromm commented on issue #5376: [Relay][Strategy] Add cuda target check to dense tensorcore schedule.
Posted by GitBox <gi...@apache.org>.
jwfromm commented on issue #5376:
URL: https://github.com/apache/incubator-tvm/pull/5376#issuecomment-616240213
@Hzfengsy, `conv2d` is already covered. I think all calls to tensorcore schedules now check that `target==cuda`.
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org