[GPU] to use original dims for fc_onednn_impl #30087
Merged
Conversation
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
e-ddykim
force-pushed
the
gpu_onednn_fc_fix
branch
from
956fba1 to
f9a5278
Compare
yeonbok
reviewed
Apr 15, 2025
| << orig_impl_param.typed_desc<fully_connected>()->id | ||
| << " " << input_shape | ||
| << " + " << fuse_op_input_shape << std::endl; | ||
| can_apply_fake_alignment = false; |
There was a problem hiding this comment.
NOTE: I think we can add a reshape for this case in the future
There was a problem hiding this comment.
Oh, I'll try it in the future as you reviewed. Thank you!
yeonbok
approved these changes
Apr 15, 2025
ahnyoung-paul
approved these changes
Apr 15, 2025
isanghao
reviewed
Apr 16, 2025
| @@ -66,14 +66,14 @@ format::type get_preferred_format(fully_connected_node const& node, const kernel | |||
| } | |||
|
|
|||
| if (input_layout.data_type == data_types::f32 && | |||
| (input_layout.format == format::bfyx || input_layout.format == format::bfzyx) && | |||
| (input_layout.format == format::bfyx || input_layout.format == format::bfzyx || input_layout.format == format::bfwzyx) && | |||
There was a problem hiding this comment.
nit: You can use one_of(input_layout.format, format::bfyx, format::bfzyx, format::bfwzyx).
There was a problem hiding this comment.
I updated as you reviewed. Thank you.
src/plugins/intel_gpu/tests/unit/test_cases/fully_connected_gpu_test.cpp
Show resolved
Hide resolved
| group_size = ifm / ngroups; | ||
| if (!is_four_bit_weight) { | ||
| // 8-bit quantized weight | ||
| attr->set_scales(DNNL_ARG_WEIGHTS, PER_OC, dnnl::memory::dims{}, ds_data_type); | ||
| attr->set_scales(DNNL_ARG_WEIGHTS, (PER_OC << shift_size), dnnl::memory::dims{}, ds_data_type); |
There was a problem hiding this comment.
nit: what about introducing a local variable per_oc = PER_OC << shift_size? That will reduce duplicated expression.
There was a problem hiding this comment.
I updated as you reviewed. Thank you.
src/plugins/intel_gpu/src/graph/impls/onednn/fully_connected_onednn.cpp
Outdated
Show resolved
Hide resolved
e-ddykim
force-pushed
the
gpu_onednn_fc_fix
branch
2 times, most recently
from
a1e95d1 to
4d8f7dd
Compare
e-ddykim
force-pushed
the
gpu_onednn_fc_fix
branch
from
4d8f7dd to
4b0097b
Compare
isanghao
approved these changes
Apr 18, 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.
Details:
Tickets: