-
Notifications
You must be signed in to change notification settings - Fork 520
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
cmake: introduce USE_SYSTEM_* flags #1859
base: main
Are you sure you want to change the base?
Conversation
✅ Deploy Preview for pytorch-fbgemm-docs canceled.
|
@q10 I see that you've worked with the CMake in this repo before, would you have the bandwidth to review this? |
0fd8f32
to
a78f826
Compare
Yes I can. Let me run it through the OSS CI system before reviewing this more thoroughly on Monday. |
Much appreciated! And any insight on changes I'd need to make for the GPU variant would be awesome :) |
a78f826
to
465f924
Compare
@q10 has imported this pull request. If you are a Meta employee, you can view this diff on Phabricator. |
b08bac7
to
2a0fb67
Compare
@q10 has imported this pull request. If you are a Meta employee, you can view this diff on Phabricator. |
As a followup, any thoughts on offering an I know that Although, I'm not very proficient in writing or interpreting CMake, so that may be working as intended or a best practice. |
I think it's a good idea to offer an I'm looking at the CI builds, and it appears that the CMake changes made in this PR have broken all the CMake-based builds of FBGEMM (see here)... |
ff1ed83
to
93cdf9b
Compare
@q10 would you mind triggering CI again to see if it's fixed? |
Looks to be broken still :( |
9f635c8
to
eea5f2a
Compare
6375da8
to
87b615c
Compare
87b615c
to
8dc96b9
Compare
Alrighty, @q10 would you mind running the CI again? |
Similar to the work done in pytorch/pytorch#37137, this adds the following CMake options:
USE_SYSTEM_LIBS
USE_SYSTEM_ASMJIT
USE_SYSTEM_CPUINFO
USE_SYSTEM_GOOGLETEST
This is particularly useful in the context of Nix, where we can build these libraries once and then re-use them elsewhere to avoid rebuilding vendors dependencies.
I'm unsure of how to best enable this for
fbgemm_gpu
.What's the relationship between FBGEMM and
libtorch
? Doeslibtorch
requirefbgemm
,fbgemm_gpu
, or both?