Add smmla/ummla support in quantized Conv2d #6802
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
|
cc: @anijain2305 , @FrozenGene , @mbaret , @u99127 |
mbaret
requested changes
Oct 30, 2020
giuseros
force-pushed
the
conv2d_mmla_support
branch
from
9c8a6fe
to
b8c8c91
Compare
|
Thanks for the review @mbaret ! The only caveat now is that we would need support for LLVM-11 in the CI to run the integration test (llvm-8 does not have i8mm support). I am commenting out the test for now, and I will enable it back once the LLVM version in the CI will be updated |
mbaret
approved these changes
Nov 4, 2020
added 4 commits
November 6, 2020 13:26
This introduces support for `smmla`/`ummla` instructions in TVM: - Added `is_mmla_available` function in `arm_utils.py` - Added the tiling node + tensorization schedule in `conv2d_gemm.py` - Added the intrinsic support in `tensor_intrin.py` - Added the test-case in `test_topi_conv2d_int8.py` Change-Id: Iff48c77f16fe1e64ecb733da965a879651ce635f
giuseros
force-pushed
the
conv2d_mmla_support
branch
from
32f03d4
to
fe99830
Compare
|
Hi @anijain2305 , @FrozenGene , Any chance you could review this? Thanks in advance! |
FrozenGene
approved these changes
Nov 6, 2020
|
About llvm11, i think we could update it for our ci |
|
Hi @FrozenGene , |
yes |
trevor-m
pushed a commit
to trevor-m/tvm
that referenced
this pull request
Dec 2, 2020
* Add smmla/ummla support in quantized Conv2d This introduces support for `smmla`/`ummla` instructions in TVM: - Added `is_mmla_available` function in `arm_utils.py` - Added the tiling node + tensorization schedule in `conv2d_gemm.py` - Added the intrinsic support in `tensor_intrin.py` - Added the test-case in `test_topi_conv2d_int8.py` Change-Id: Iff48c77f16fe1e64ecb733da965a879651ce635f * Address review comments and test failures * Fix linting * Rebasing
trevor-m
pushed a commit
to trevor-m/tvm
that referenced
this pull request
Dec 4, 2020
* Add smmla/ummla support in quantized Conv2d This introduces support for `smmla`/`ummla` instructions in TVM: - Added `is_mmla_available` function in `arm_utils.py` - Added the tiling node + tensorization schedule in `conv2d_gemm.py` - Added the intrinsic support in `tensor_intrin.py` - Added the test-case in `test_topi_conv2d_int8.py` Change-Id: Iff48c77f16fe1e64ecb733da965a879651ce635f * Address review comments and test failures * Fix linting * Rebasing
trevor-m
pushed a commit
to neo-ai/tvm
that referenced
this pull request
Dec 4, 2020
* Add smmla/ummla support in quantized Conv2d This introduces support for `smmla`/`ummla` instructions in TVM: - Added `is_mmla_available` function in `arm_utils.py` - Added the tiling node + tensorization schedule in `conv2d_gemm.py` - Added the intrinsic support in `tensor_intrin.py` - Added the test-case in `test_topi_conv2d_int8.py` Change-Id: Iff48c77f16fe1e64ecb733da965a879651ce635f * Address review comments and test failures * Fix linting * Rebasing
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
High level description of the submission
This introduces support for
smmla/ummlainstructions in TVM:is_mmla_availablefunction inarm_utils.pyconv2d_gemm.pytensor_intrin.pytest_topi_conv2d_int8.pyRFC
This PR is based on the following RFC: https://discuss.tvm.apache.org/t/rfc-improve-quantized-convolution-through-mmla-instructions/8336
Change-Id: Iff48c77f16fe1e64ecb733da965a879651ce635f