Skip to content

[ET-VK][ez] Make q8ta_conv2d use 4C1W layout #17390

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
SS-JIA wants to merge 7 commits into
base: gh/SS-JIA/421/base
Choose a base branch
from
Open

[ET-VK][ez] Make q8ta_conv2d use 4C1W layout #17390

Changes from all commits
Commits
Show all changes
7 commits
Select commit Hold shift + click to select a range
f21bfb9
[ET-VK][ez] Make q8ta_conv2d use 4C1W layout
Feb 11, 2026
0578684
Update on "[ET-VK][ez] Make q8ta_conv2d use 4C1W layout"
Feb 11, 2026
03aec14
Update on "[ET-VK][ez] Make q8ta_conv2d use 4C1W layout"
Feb 11, 2026
85e5541
Update on "[ET-VK][ez] Make q8ta_conv2d use 4C1W layout"
Feb 12, 2026
c11b3d5
Update on "[ET-VK][ez] Make q8ta_conv2d use 4C1W layout"
Feb 12, 2026
c715909
Update on "[ET-VK][ez] Make q8ta_conv2d use 4C1W layout"
Feb 12, 2026
4e071b7
Update on "[ET-VK][ez] Make q8ta_conv2d use 4C1W layout"
Feb 13, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
8 changes: 7 additions & 1 deletion backends/vulkan/op_registry.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -778,6 +778,9 @@ def register_q8ta_conv_pw_op():
utils.NO_STORAGE, # groups (non tensor)
utils.NO_STORAGE, # original OC count (non tensor)
],
outputs_storage=[
utils.PACKED_INT8_CHANNELS_PACKED_BUFFER,
],
supports_resize=False,
supports_prepacking=True,
)
Expand All @@ -792,7 +795,7 @@ def register_q8ta_conv_pw_op():
def register_q8ta_conv2d_ops():
return OpFeatures(
inputs_storage=[
utils.PACKED_INT8_4W4C_BUFFER, # input
utils.PACKED_INT8_4C1W_BUFFER, # input
utils.NO_STORAGE, # input_scale (non tensor)
utils.NO_STORAGE, # input_zero_point (non tensor)
utils.NO_STORAGE, # weight (prepacked)
Expand All @@ -808,6 +811,9 @@ def register_q8ta_conv2d_ops():
utils.NO_STORAGE, # groups (non tensor)
utils.NO_STORAGE, # original OC count (non tensor)
],
outputs_storage=[
utils.PACKED_INT8_CHANNELS_PACKED_BUFFER,
],
supports_resize=False,
supports_prepacking=True,
)
Expand Down