-
Notifications
You must be signed in to change notification settings - Fork 6.7k
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
[flashlight-cuda] New Port #14676
[flashlight-cuda] New Port #14676
Conversation
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.
Can you revert changes about nccl and depends on #14683?
ca5969a
to
622f802
Compare
@JackBoosY done — I'll rebase on top of the NCCL port once it's done though. |
622f802
to
257be38
Compare
@JackBoosY how can I get detailed logs of this failure in CI? |
@JackBoosY I have permission. How do I get to this page for future CI runs? |
@jacobkahn @strega-nil Maybe we need to write documentation to the contributors. |
@JackBoosY is there some way I can know what command CI ran when this error happened? I'm unable to reproduce it. |
@jacobkahn Maybe the environment of our CI machine is different from yours. You can only use the log content to solve the regression. |
@JackBoosY fixed the issue. Needed a default feature. |
Thanks @jacobkahn |
Adds a new port for flashlight, an ML library written in C++. I'm one of the project authors.
A few things to note:
find_package
natively with aFindNCCL.cmake
module.NCCL
andcuDNN
are very similar in their installations next to CUDA, so the NCCL port works exactly like the vcpkg cuDNN port.Supports
linux
for now. I've testedx64-linux
. The port has!(windows|osx)
, but I will test the port on macOS shortly. I don't expect Windows integration to be complete for some time, but I will update the port if/when it is.Yes.
cc @tlikhomanenko @vineelpratap @xuqiantong @avidov @andresy @syhw @padentomasello