Skip to content
This repository has been archived by the owner on Nov 17, 2023. It is now read-only.

Backport #16798, #16836 and #16838 to 1.6 #16874

Merged
merged 5 commits into from
Nov 22, 2019

Conversation

ptrendx
Copy link
Member

@ptrendx ptrendx commented Nov 20, 2019

Description

Backporting #16798, #16836 and #16838 to 1.6.x branch.

@haojin2 @DickJC123 FYI

ptrendx and others added 2 commits November 20, 2019 10:32
* Add unoptimized symbol to executor for sharing

* Copy the symbol in Reshape

* Added test for multiple reshapes
* mixed precison binary op backward

* reduce unix cpu runtime
@ptrendx ptrendx added the R1.6.0 label Nov 20, 2019
… fusion. (apache#16838)

* Rename USE_NVRTC -> ENABLE_CUDA_RTC to fix maven build.  Compile-guard fusion framework.

* Fix fusion-not-supported warning.

* Fix compile guards

* Fix cmake build so -DMXNET_ENABLE_CUDA_RTC=1 is passed to nvcc

* Minimize side-effects of prev change
@ptrendx ptrendx changed the title Backport #16798 and #16791 to 1.6 Backport #16798, #16791 and #16838 to 1.6 Nov 20, 2019
@ptrendx
Copy link
Member Author

ptrendx commented Nov 21, 2019

@haojin2 Windows build failed with fatal error C1002: compiler is out of heap space in pass 2 - did you do anything in the other PRs to reduce the memory usage during compilation?

… graph (apache#16836)

* Fix the attribute inference omitting nodes

* Add test

* Cleaning

* Fix lint

* Fix TransposeShape

* Fix WhileLoopType

* Changing a/b test for fusion to a/(b+1) to increase numerical stability
@ptrendx ptrendx changed the title Backport #16798, #16791 and #16838 to 1.6 Backport #16798, #16791, #16836 and #16838 to 1.6 Nov 21, 2019
@ptrendx ptrendx changed the title Backport #16798, #16791, #16836 and #16838 to 1.6 Backport #16798, #16836 and #16838 to 1.6 Nov 21, 2019
@ptrendx
Copy link
Member Author

ptrendx commented Nov 21, 2019

Due to problems with compilation on Windows I removed #16791 from this bulk of cherry-picks. @haojin2 Please make a separate PR to bring #16791 to 1.6.x branch and work on fixing those issues there.

@haojin2
Copy link
Contributor

haojin2 commented Nov 21, 2019

@ptrendx There's a separate PR #16827 that is needed to fix such issues. #16827 made a major refactor to the np_elemwise_binary_broadcast part and reduced the pressure on windows vs compilers.

@ptrendx
Copy link
Member Author

ptrendx commented Nov 22, 2019

Ok, so @haojin2 please make a PR to 1.6.x branch with both #16827 and #16791.

@ptrendx ptrendx merged commit e73c186 into apache:v1.6.x Nov 22, 2019
Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.
Labels
Projects
None yet
Development

Successfully merging this pull request may close these issues.

3 participants