Fix vmlal.s16 code generation for int8 x int8 -> int32 #2748
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
tqchen
requested changes
Mar 8, 2019
src/pass/lower_intrin.cc
Outdated
| @@ -50,7 +50,23 @@ class IntrinInjecter : public IRMutator { | |||
| // on ARM. | |||
| if (const Broadcast* bcast = e.as<Broadcast>()) { | |||
| if (const Cast* cast = bcast->value.as<Cast>()) { | |||
| if (cast->type.bits() == cast->value.type().bits() * 2) { | |||
| auto shouldSwap = [&]() { | |||
There was a problem hiding this comment.
nit: code style, local variable need to be should_swap?
FrozenGene
reviewed
Mar 8, 2019
| A[k, n].astype("int32") * B[k, n].astype("int32"), axis=[k]), name='C') | ||
| s = tvm.create_schedule(C.op) | ||
| s[C].vectorize(s[C].op.axis[0]) | ||
| print(tvm.lower(s, [A, B, C], simple_mode=True)) |
There was a problem hiding this comment.
maybe we should remove this print
ajtulloch
force-pushed
the
widening-int8-int32-arm-codegen
branch
from
98b54d4 to
17112b5
Compare
|
Thanks for the comments, I've updated the patch with the review comments. |
tqchen
approved these changes
Mar 9, 2019
FrozenGene
approved these changes
Mar 9, 2019
|
Thanks @ajtulloch @FrozenGene , this is now merged! |
|
Thanks for merging @tqchen. |
wweic
pushed a commit
to neo-ai/tvm
that referenced
this pull request
Mar 12, 2019
wweic
pushed a commit
to neo-ai/tvm
that referenced
this pull request
Mar 12, 2019
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Outdated suggestions cannot be applied.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.
The
IntrinInjecter::SwapBroadcastCastfunction was limited to cases where the result bitwidth was exactly 2x the input bitwidth. This fails for cases of relevance such as a vectorized int8 x int8 -> int32 GEMM. This helps improve the code-generation somewhat by outputtingVMLAL.S16instructions instead of aMOVL+ VMLA.S32`.