lattice / quda

QUDA is a library for performing calculations in lattice QCD on GPUs.
https://lattice.github.io/quda
Other
289 stars 97 forks source link

Hotfix to fix compilation on Summit #1337

Closed weinbe2 closed 1 year ago

weinbe2 commented 1 year ago

This narrow hotfix PR addresses an issue with compilation that only seemed to creep up on Summit, where the cmake internal variable CXX_OPT caused compiler errors (seemingly unrelated to the specialization of CXX_OPT for Power9). The issue appeared with https://github.com/lattice/quda/commit/738e4be0f0ce56a7ee33c6cf8af2d70f8a53fa3b (which needs the subsequent commit, https://github.com/lattice/quda/commit/63182d6bfe58ed1a96399f86ebdce13bbd176792 , to compile).

I confirmed that this fixes the compile-time issue on Summit, and "does no harm" on x86 as verified by looking at the outputs from a --verbose compile of QUDA through nvcc.

I also fixed a few harmless typos in CMakeLists.txt.

weinbe2 commented 1 year ago

~Do not merge yet...~ This is good to go, the issue I hit was because I merged this into a different branch.

weinbe2 commented 1 year ago

I confirmed that all other build types (DEVEL, DEBUG, HOSTDEBUG, STRICT, SANITIZE) compile without an issue. As a note for future reference, all of these tests were with CUDA 11.0, which is the current default on Summit.