openvino
[CPU][ARM]Non i32 conv bias support
#33072
Merged
Overview
Commits
86
Changes
View On GitHub

[CPU][ARM]Non i32 conv bias support #33072

alvoron merged 86 commits into openvinotoolkit:master from alvoron:non_i32_conv_bias_support
alvoron
alvoron init
4e58c7a9
alvoron address comments
06a154da
alvoron fixed comments
feabefb3
alvoron fix clang-tidy and arm32 tests
3c43a840
alvoron fix clang-tidy
3ed4e83d
alvoron fix clang
ba07c2d2
alvoron added comment
08ddf4e9
alvoron init
b2b7d942
github-actions github-actions added category: CPU
alvoron subgraph test handles non-i32 bias case
1a5ca1d4
github-actions github-actions added category: LP transformations
alvoron register ConvertConvolutionBias separately
f8c7eb19
github-actions github-actions removed category: LP transformations
alvoron alvoron marked this pull request as ready for review 172 days ago
alvoron alvoron requested a review 172 days ago
alvoron alvoron requested a review 172 days ago
alvoron Register ConvertConvolutionBias on ARM only + cleanup
4e02b17d
alvoron alvoron assigned v-Golubev v-Golubev 172 days ago
v-Golubev
v-Golubev commented on 2025-12-09
alvoron addressed comments
ab71e7bb
alvoron fix comparison of different signedness
680df8a6
v-Golubev
v-Golubev commented on 2025-12-10
alvoron addressed comments
ab75009d
alvoron trigger AddTransformation for non-convolution bias
1a7826ab
alvoron fix tests
73ca02a0
v-Golubev
v-Golubev commented on 2025-12-12
alvoron addressed comments
b20fe80d
alvoron Merge branch 'master' into non_i32_conv_bias_support
3c756991
alvoron reverted legacy logic in ConvAndFQ
6690abb5
alvoron alvoron added platform: arm
alvoron fix clang and compilation issue
9247b9af
alvoron fix clang
a12b120b
alvoron fix clang
6a21f55c
v-Golubev
v-Golubev commented on 2025-12-22
alvoron fix comments
385c5e7e
alvoron fix AddTransformation callback condition
45f1be0e
alvoron fix comment
adb5f3d2
alvoron fix comment and potential crash
78cb4c9e
alvoron fix clang-tidy
ac4bba89
v-Golubev
v-Golubev approved these changes on 2025-12-23
alvoron check fq output precision in ConvertConvolutionBias transformation
e3d6525b
alvoron unit tests are aligned with the the latest transformation changes
30895011
alvoron Merge branch 'master' into non_i32_conv_bias_support
408669e2
alvoron updated quantizationParams after merge
0df317e0
alvoron force nhwc for int8 conv executor
2098d761
alvoron force fp16 post-op precision
af68b561
alvoron conv fallback to fp16 if int8 conv is not applicable
8b147bf8
alvoron support u8 activations in FC
837321d5
alvoron keep decomposed FQ if its output prec does not match Conv activation …
e226ecd0
alvoron arm callbacks for ConvolutionTransformation and FuseMultiplyToFakeQua…
6ac4e614
alvoron fix post-op fp16 force condition
6ce944ed
alvoron int8 acl executor accepts 4D activations and 4D weight only
cc2eff7c
alvoron rollback wei type to ncsp for dnnl acl executor
2f83ac63
alvoron fix clang-tidy
37ee5403
alvoron add FuseSubtractToFakeQuantizeTransformation callback and use matcher…
85c2d6d8
alvoron refactor AddTransformation callback
94f96b8e
alvoron fix clang-tidy
b8519e25
v-Golubev
v-Golubev commented on 2026-01-28
alvoron fix typo
14e903b4
alvoron fix comments
1b841124
alvoron fix supported condition in ACLConvolutionExecutor::supports
d1c899e1
alvoron fix clang
5362030b
alvoron fix clang
d22f5a2a
v-Golubev
v-Golubev commented on 2026-01-29
alvoron Apply suggestions from code review
7248bb2d
alvoron move callback utility methods to utils.cpp
073205dc
alvoron fix clang
f2587c49
alvoron fix clang
608cbbff
alvoron fix
3be0bea1
alvoron fix
13c6e910
alvoron fix
a85d6ca4
alvoron linker fix
ef45a983
alvoron fix
8b1c46d6
alvoron fix
6d78c174
alvoron fix
ffe26f4c
alvoron fix
0bb89532
alvoron fix
986fe102
v-Golubev
v-Golubev approved these changes on 2026-01-30
v-Golubev v-Golubev assigned EgorDuplensky EgorDuplensky 119 days ago
alvoron Update src/plugins/intel_cpu/src/transformations/utils.cpp
07d0bcf1
EgorDuplensky
EgorDuplensky commented on 2026-01-30
alvoron fix comments
6d7aaacc
alvoron fix clang-tidy
ca44b1a1
alvoron fix
99ce6732
EgorDuplensky
EgorDuplensky commented on 2026-02-02
EgorDuplensky
EgorDuplensky commented on 2026-02-02
alvoron applied comment
f5b0543e
alvoron use useF16Binary unconditionally
bab181c8
EgorDuplensky
EgorDuplensky approved these changes on 2026-02-02
alvoron alvoron added do not merge
alvoron
alvoron [experiment] swap Matmul and Add
565c40c1
alvoron [experiment] fold dq scales directly
be2fe37c
alvoron move dq scales folding on ngraph level
2476e7a5
alvoron fix ConvAndFQ tests
393c20f2
alvoron fix ConvertConvolutionBias TransformationTests
4eb5808d
alvoron revert unnessesary change
7411642f
alvoron fix risc-v build
89cea490
alvoron fix risc-v
9c873b88
v-Golubev
v-Golubev commented on 2026-02-10
alvoron addressed comments
2e93b124
alvoron fix clang-tidy
1561de29
alvoron fix clang-tidy
d717a202
alvoron temporary disabled ConvolutionTransformation
59b66392
v-Golubev
v-Golubev approved these changes on 2026-02-12
v-Golubev v-Golubev removed do not merge
v-Golubev v-Golubev added this to the 2026.1 milestone 106 days ago
alvoron fix clang-tidy
279e46b6
alvoron disable smoke_ConvAndFQ_CPU tests
18b960d2
alvoron Merge branch 'master' into non_i32_conv_bias_support
7f6d7eb8
alvoron alvoron enabled auto-merge 106 days ago
alvoron Add SKIP_IF_CURRENT_TEST_IS_DISABLED to int8 test
ae12a3f8
alvoron alvoron added category: LP transformations
alvoron fix Copyright year
a96d2921
github-actions github-actions removed category: LP transformations
alvoron Merge branch 'master' into non_i32_conv_bias_support
ed5f9c9a
alvoron alvoron merged b88f5e78 into master 105 days ago
alvoron alvoron deleted the non_i32_conv_bias_support branch 105 days ago

Login to write a write a comment.

Login via GitHub

Reviewers
Assignees
Labels
category: CPU platform: arm
Milestone
2026.1
Terms Of Use Privacy Policy Refund Policy Impressum