Skip to content
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

Jump to bottom

ftmDataFormat for direct convolution is wrong in convolution_transform_filter_fp16 #3

Closed
jerryjinjin opened this issue Dec 6, 2019 · 1 comment
Closed

ftmDataFormat for direct convolution is wrong in convolution_transform_filter_fp16 #3

jerryjinjin opened this issue Dec 6, 2019 · 1 comment

Comments

@jerryjinjin
Copy link

ftmDataFormat for CONVOLUTION_ALGORITHM_DIRECT should be DF_NCHWN16.
https://github.com/huawei-noah/bolt/blob/master/tensor_computing/src/cpu/arm/fp16/convolution_transform_fp16.h#L194
@jianfeifeng
Copy link
Collaborator

Thanks for your reminder, that was the bug when refactoring our code. We have fixed it, and you can refer to the latest code tensor_computing/src/cpu/arm/fp16/convolution_transform_fp16.h . By the way, we recommend you to use the implement of gemm to reach high performance computing instead of using the direct implement method. Thank you again!

Best Regards!
The Bolt Team.

jianfeifeng pushed a commit that referenced this issue Apr 13, 2021
@yunfanxiao
Merge pull request #3 from yunfanxiao/patch-4
7f4ede2 
Update bolt.cpp
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

No branches or pull requests
2 participants
@jerryjinjin @jianfeifeng