[GPU] Expand matmul decomp cases #2916
Open
+171
−117
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
github-actions
bot
added
the
platform:gpu-intel
kealan-barbieri
force-pushed
the
kealanba/gemm_fpmath_fix
branch
from
2f3e014 to
e3a9503
Compare
github-actions
bot
added
platform:cpu-x64
dzarukin
force-pushed
the
dzarukin/allow_int4_odd
branch
3 times, most recently
from
b34fab9 to
2e2c013
Compare
rjoursler
reviewed
Mar 19, 2025
dzarukin
force-pushed
the
dzarukin/allow_int4_odd
branch
4 times, most recently
from
f4d86c8 to
8e04316
Compare
kealan-barbieri
force-pushed
the
kealanba/gemm_fpmath_fix
branch
from
e3a9503 to
4375e7a
Compare
github-actions
bot
removed
platform:cpu-x64
kealan-barbieri
force-pushed
the
kealanba/gemm_fpmath_fix
branch
from
4375e7a to
6892eb5
Compare
dzarukin
force-pushed
the
dzarukin/allow_int4_odd
branch
from
8e04316 to
608baa3
Compare
kealan-barbieri
force-pushed
the
kealanba/gemm_fpmath_fix
branch
from
6892eb5 to
ebf1fa9
Compare
vpirogov
force-pushed
the
dzarukin/allow_int4_odd
branch
from
608baa3 to
f933c1e
Compare
|
make test |
kealan-barbieri
changed the title
kealan-barbieri
force-pushed
the
kealanba/gemm_fpmath_fix
branch
from
ebf1fa9 to
d7701d3
Compare
|
make test |
kealan-barbieri
changed the title
kealan-barbieri
force-pushed
the
kealanba/gemm_fpmath_fix
branch
from
d7701d3 to
7ea8b1d
Compare
|
make test |
kealan-barbieri
force-pushed
the
kealanba/gemm_fpmath_fix
branch
from
7ea8b1d to
3cce26c
Compare
|
make test |
hidefromkgb
approved these changes
Apr 2, 2025
syurkevi
approved these changes
Apr 2, 2025
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Outdated suggestions cannot be applied.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.
Description
Expands cases where inputs are decompressed to include integer activations as well as weights. Integer weights are still required but in the presence of fpmath setting activations will also be upconverted.
This means that cases intending to use integer accumulation must not supply eg
attr-fpmath=f16:trueas all such cases must be upconverted per https://jira.devtools.intel.com/browse/MFDNN-13380.Align behavior with doc for application of fpmath to integer inputs, if integer fpmath is true then upconversion must occur, this changes the strategy selection. Relabel strategies that perform upconversion to make them explicit. Update strategy selection to distinguish ordering between different "compound" types eg "[OH]" "[OB]".
Fixes # (github issue)
Checklist
General
make testandmake test_benchdnn_*) pass locally for each commit?