[Tensorize][TOPI] Add AMX Tensorizing for int8 batch matmul#13745
Merged
masahi merged 6 commits intoapache:mainfrom Jan 11, 2023
Merged
[Tensorize][TOPI] Add AMX Tensorizing for int8 batch matmul#13745masahi merged 6 commits intoapache:mainfrom
masahi merged 6 commits intoapache:mainfrom
Conversation
Collaborator
|
Thanks for contributing to TVM! Please refer to the contributing guidelines https://tvm.apache.org/docs/contribute/ for useful information and tips. Please request code reviews from Reviewers by @-ing them in a comment.
Generated by tvm-bot |
Qianshui-Jiang
changed the title
Qianshui-Jiang
changed the title
Contributor
Author
|
@tvm-bot rerun |
masahi
reviewed
Jan 10, 2023
| if "batch_matmul_int8" in op.tag: | ||
| layout_trans = op.input_tensors[1] | ||
| batch_matmul_vnni_schedule(cfg, s, op.output(0), outs[0], layout_trans) | ||
| print(layout_trans) |
| lib = relay.build(mod, target=target) | ||
|
|
||
| asm = lib.lib.get_source("asm") | ||
| assert "vpdpbusd" in asm |
Contributor
Author
There was a problem hiding this comment.
@masahi just for local test and forget to change it back, now already fixed
masahi
approved these changes
Jan 11, 2023
fzi-peccia
pushed a commit
to fzi-peccia/tvm
that referenced
this pull request
Mar 27, 2023
…3745) * amx int8 tensorized x86 bmm * remove the unused amx schedule * fix lint * fix lint * remove unused import * fix Instr. assert in testcase.
This file contains hidden or 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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
This PR is to tensorize X86 int8 batch matmul based on last AMX PR: #13642
unified batch matmul VNNI and AMX to
batch_matmul_int8, then chose different SIMD(VNNI or AMX) of the target to finish tensorizing.(basicly the same way to x86 dense_int8)