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 2020/12/23 07:31:57 UTC
[GitHub] [tvm] masahi opened a new pull request #7157: [TOPI] GPU sort IR refactor to enable sort by keys
masahi opened a new pull request #7157:
URL: https://github.com/apache/tvm/pull/7157
----------------------------------------------------------------
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
[GitHub] [tvm] mbrookhart commented on pull request #7157: [TOPI] GPU sort IR refactor to enable sort by keys
Posted by GitBox <gi...@apache.org>.
mbrookhart commented on pull request #7157:
URL: https://github.com/apache/tvm/pull/7157#issuecomment-750491074
Yep, this matches the scaling I would expect with that algorithm. I'm kind of surprised that sequential scatter isn't worse than it is. I'm looking for ways to improve the mergesort, but I haven't found a good idea that we can implement in TIR yet, what I've seen requries a `while` loop.
----------------------------------------------------------------
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
[GitHub] [tvm] mbrookhart commented on pull request #7157: [TOPI] GPU sort IR refactor to enable sort by keys
Posted by GitBox <gi...@apache.org>.
mbrookhart commented on pull request #7157:
URL: https://github.com/apache/tvm/pull/7157#issuecomment-750466642
I definitely didn't promise to be faster than sort with #7099, just faster that what we had in TIR :)
----------------------------------------------------------------
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
[GitHub] [tvm] masahi edited a comment on pull request #7157: [TOPI] GPU sort IR refactor to enable sort by keys
Posted by GitBox <gi...@apache.org>.
masahi edited a comment on pull request #7157:
URL: https://github.com/apache/tvm/pull/7157#issuecomment-750473094
Ok here it is. I think TIR numbers for 5000 and 10000 agree with the ones you had in https://github.com/apache/tvm/pull/7099
| size | current (sequential) | via thrust radix sort | via TIR merge sort
| ---- | ---- | ---- | ---- |
| 5000 | 0.169332 | 0.168660 | 5.319211
| 10000 | 0.301014 | 0.172381 | 10.534689
| 25000 | 0.730493 | 0.169007| 22.173880
| 50000 | 1.457996 | 0.315582| 44.466231
| 100000 | 2.943622 | 0.395291| 101.347441
| 500000 | 24.262521 | 1.254097| 481.400628
----------------------------------------------------------------
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
[GitHub] [tvm] zhiics commented on pull request #7157: [TOPI] GPU sort IR refactor to enable sort by keys
Posted by GitBox <gi...@apache.org>.
zhiics commented on pull request #7157:
URL: https://github.com/apache/tvm/pull/7157#issuecomment-750498301
Thanks @masahi @mbrookhart @Laurawly
----------------------------------------------------------------
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
[GitHub] [tvm] mbrookhart edited a comment on pull request #7157: [TOPI] GPU sort IR refactor to enable sort by keys
Posted by GitBox <gi...@apache.org>.
mbrookhart edited a comment on pull request #7157:
URL: https://github.com/apache/tvm/pull/7157#issuecomment-750466642
I definitely didn't promise to be faster than thrust with #7099, just faster that what we had in TIR :)
----------------------------------------------------------------
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
[GitHub] [tvm] masahi commented on pull request #7157: [TOPI] GPU sort IR refactor to enable sort by keys
Posted by GitBox <gi...@apache.org>.
masahi commented on pull request #7157:
URL: https://github.com/apache/tvm/pull/7157#issuecomment-750473094
Ok here it is. I think TIR numbers for 5000 and 10000 agree with the ones you had in https://github.com/apache/tvm/pull/7099
| size | current (sequential) | via thrust radix sort | via TIR sort
| ---- | ---- | ---- | ---- |
| 5000 | 0.169332 | 0.168660 | 5.319211
| 10000 | 0.301014 | 0.172381 | 10.534689
| 25000 | 0.730493 | 0.169007| 22.173880
| 50000 | 1.457996 | 0.315582| 44.466231
| 100000 | 2.943622 | 0.395291| 101.347441
| 500000 | 24.262521 | 1.254097| 481.400628
----------------------------------------------------------------
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
[GitHub] [tvm] zhiics merged pull request #7157: [TOPI] GPU sort IR refactor to enable sort by keys
Posted by GitBox <gi...@apache.org>.
zhiics merged pull request #7157:
URL: https://github.com/apache/tvm/pull/7157
----------------------------------------------------------------
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