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.
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
supports empty kernels in cuda::SeparableLinearFilters #3731
Open
chacha21
wants to merge
4
commits into
opencv:4.x
Choose a base branch
from
chacha21:cuda_separable_filter_single
base: 4.x
Could not load branches
Branch not found: {{ refName }}
Could not load tags
Nothing to show
Are you sure you want to change the base?
Some commits from the old base branch may be removed from the timeline,
and old review comments may become outdated.
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
When only 1D convolution is needed (row or column filter only), `cuda::LinearFilter` is slower than `cuda::SeparableLinearFilter` Using `cuda::SeparableLinearFilter` for 1D convolution can be tricked by using a `(1)` kernel for the ignored dimension By supporting empty kernels in `cuda::SeparableLinearFilter`, there is no need for that `(1)` kernel any more Additionaly, the inner `_buf ` used to store the intermediate convolutio result can be saved when a single convolution is needed In "legacy" usage (row+col kernels), there is no regerssion in `cuda::SeparableLinearFilter` performance As soon as an empty kernel is used, the performance is largely increased
The previous commit was incomplete and did not allow correct handling of non-CV_32F case To save cuda code instanciation, the rowFilter is always srcType->CV_32F and the colFilter is CV_32F->dstType Thus the intermediate buffer is always CV_32F This is a little tricky when only a single kernel is used (either row or column), because src or dst adaptation might be needed.
|
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.
#25408
When only 1D convolution is needed (row or column filter only),
cuda::LinearFilter
might be slower thancuda::SeparableLinearFilter
Using
cuda::SeparableLinearFilter
for 1D convolution can be done by using a(1)
kernel for the ignored dimension.By supporting empty kernels in
cuda::SeparableLinearFilter
, there is no need for that(1)
kernel any more.Additionaly, the inner
_buf
used to store the intermediate convolution result can be saved when a single convolution is needed.In "legacy" usage (row+col kernels), there is no regression in
cuda::SeparableLinearFilter
performance.As soon as an empty kernel is used, the performance is largely increased.
Devil in the details : the "in-place" processing is supported and might need intermediate buf, but still no regression.
Patch to opencv_extra has the same branch name.