-
Notifications
You must be signed in to change notification settings - Fork 42
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
Vectorized decompression #38
Conversation
can't we specialize for GPU and keep the looping default? |
Yes, this is just prototyping. I wrote this on a plane and pushed it so we wouldn't lose what the correct translation is, because it's pretty gnarly. I may finish it up later or if @pkj-m has a GPU and wants to try pushing the GPU algorithm until |
6f30078
to
a3b145b
Compare
add a comment to the very non-obvious vectorized code
Co-Authored-By: Mathieu Besançon <mathieu.besancon@gmail.com>
Co-Authored-By: Mathieu Besançon <mathieu.besancon@gmail.com>
fa62624
to
d7ce4ef
Compare
struct ContractionColor <: ColoringAlgorithm end | ||
struct GreedyStar1Color <: ColoringAlgorithm end | ||
struct GreedyStar2Color <: ColoringAlgorithm end | ||
abstract type SparseDiffToolsColoringAlgorithm <: ArrayInterface.ColoringAlgorithm end |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
SparseDiffToolsColoringAlgorithm
not a fan of the package name plugged into types, C-like. Couldn't we not-export ColoringAlgorithm
, thus calling it with SparseDiffToolsColoringAlgorithm
. Other solution would be to remove the abstract type all together and use ArrayInterface.ColoringAlgorithm
directly
Co-Authored-By: Mathieu Besançon <mathieu.besancon@gmail.com>
It is slower but potentially useful for GPUs: