Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

[VTA][Relay] Relay Compilation + AutoTVM compatible operator libraries for VTA #3135

Merged
merged 126 commits into from
Jun 28, 2019

Conversation

tmoreau89
Copy link
Contributor

@tmoreau89 tmoreau89 commented May 3, 2019

EDIT: this PR has grown to include front-end support with Relay of VTA. Support for quantization, graph packing, and topi operator support are now grouped in this rather sizeable PR. In terms of operator coverage, group_conv2d and conv2d_transpose will be added in later PRs.


Currently a WIP PR. The goal here is to bring AutoTVM-compatible TOPI operators to VTA so we can leverage the autotuner to quickly search good schedules, instead of using the old scheduler that was ad-hoc and bug-prone.

Added the conv2d operator, but will be adding group_conv2d (for mobilenets variants), conv2d_transpose (for DCGAN), and dense for MLPs/RNNs.

Credit for most of the AutoTVM support goes to @merrymercy; relay bitpacking to @jroesch and @MarisaKirisame who helped build the first prototype for Relay to VTA compilation.


@tmoreau89 tmoreau89 force-pushed the dev branch 3 times, most recently from f0a5472 to 804a55f Compare May 10, 2019 20:48
@tmoreau89 tmoreau89 changed the title [VTA] AutoTVM compatible TOPI operator libraries for VTA [VTA][TOPI] AutoTVM compatible operator libraries for VTA May 10, 2019
@tmoreau89 tmoreau89 force-pushed the dev branch 4 times, most recently from aa10d33 to a02050d Compare May 16, 2019 18:36
@tmoreau89 tmoreau89 changed the title [VTA][TOPI] AutoTVM compatible operator libraries for VTA [VTA][Relay] Relay Compilation + AutoTVM compatible operator libraries for VTA May 17, 2019
@tmoreau89
Copy link
Contributor Author

Extending the scope of this PR: in addition to AutoTVM support of VTA TOPI operators, we are working on making Relay to VTA compilation work on off-the-shelf Gluon models.

@tmoreau89 tmoreau89 force-pushed the dev branch 5 times, most recently from 7e4bc5a to 4a30921 Compare May 24, 2019 01:55
@tmoreau89 tmoreau89 force-pushed the dev branch 3 times, most recently from 4a92032 to 12ae5ae Compare May 29, 2019 23:57
@tmoreau89 tmoreau89 force-pushed the dev branch 6 times, most recently from 0d0334f to 03875c0 Compare June 10, 2019 21:35
@tmoreau89
Copy link
Contributor Author

@tqchen @vinx13 @merrymercy please let me know if there are other changes that you'd like to request

Copy link
Member

@vinx13 vinx13 left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

some minor

python/tvm/relay/quantize/quantize.py Outdated Show resolved Hide resolved
@tmoreau89
Copy link
Contributor Author

@tqchen let me know if I've addressed the changes you requested

@tqchen
Copy link
Member

tqchen commented Jun 28, 2019

Thanks, @merrymercy @vinx13 @ZihengJiang @tmoreau89 , this PR is now merged

wweic pushed a commit to wweic/tvm that referenced this pull request Jun 28, 2019
@tmoreau89 tmoreau89 deleted the dev branch June 28, 2019 23:35
wweic pushed a commit to neo-ai/tvm that referenced this pull request Jun 30, 2019
vinx13 added a commit to vinx13/tvm that referenced this pull request Jul 8, 2019
vinx13 added a commit to vinx13/tvm that referenced this pull request Jul 11, 2019
tqchen pushed a commit that referenced this pull request Jul 12, 2019
* [Relay][Quantization] Fix issue introduced in #3135

* Recover StopFusion

* Fix fmultiref

* Fix lint
wweic pushed a commit to wweic/tvm that referenced this pull request Aug 9, 2019
* [Relay][Quantization] Fix issue introduced in apache#3135

* Recover StopFusion

* Fix fmultiref

* Fix lint
wweic pushed a commit to neo-ai/tvm that referenced this pull request Sep 6, 2019
* [Relay][Quantization] Fix issue introduced in apache#3135

* Recover StopFusion

* Fix fmultiref

* Fix lint
tqchen pushed a commit to tqchen/tvm that referenced this pull request Mar 29, 2020
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Development

Successfully merging this pull request may close these issues.

7 participants