fix the compilation for power pc

[eddy16112]

This PR is to fix the following error on power pc

In file included from /autofs/nccs-svm1_home1/wwu12/install/TBLIS/include/tblis/util/../external/marray/include/expression.hpp:4,
                 from /autofs/nccs-svm1_home1/wwu12/install/TBLIS/include/tblis/util/../external/marray/include/marray_base.hpp:509,
                 from /autofs/nccs-svm1_home1/wwu12/install/TBLIS/include/tblis/util/../external/marray/include/varray_base.hpp:4,
                 from /autofs/nccs-svm1_home1/wwu12/install/TBLIS/include/tblis/util/../external/marray/include/varray_view.hpp:4,
                 from /autofs/nccs-svm1_home1/wwu12/install/TBLIS/include/tblis/util/../external/marray/include/varray.hpp:4,
                 from /autofs/nccs-svm1_home1/wwu12/install/TBLIS/include/tblis/util/basic_types.h:38,
                 from /autofs/nccs-svm1_home1/wwu12/install/TBLIS/include/tblis/util/configs.h:4,
                 from /autofs/nccs-svm1_home1/wwu12/install/TBLIS/include/tblis/tblis.h:4,
                 from cunumeric/matrix/contract.cc:21:
/autofs/nccs-svm1_sw/summit/gcc/9.3.0-2/lib/gcc/powerpc64le-unknown-linux-gnu/9.3.0/include/x86intrin.h:32:2: error: #error "Please read comment above.  Use -DNO_WARN_X86_INTRINSICS to disable this error."

[magnatelee]

LGTM, but I want @manopapad to take a look at this as well.

[manopapad]

LGTM

[lightsighter]

Shouldn't we really ping the TBLIS library developers to make sure that they are masking their intrinsic headers correctly? Do we know if they even support PowerPC vector intrinsics?

[eddy16112]

Let me create a issue on their github repo to see what do they say about this error.
devinamatthews/tblis#45

[manopapad]

The ppc build for TBLIS has not yet materialized, so this is still necessary. However, the make-based build workflow has been removed since this PR was submitted, so we need to port this to the cmake-based build.

@trxcllnt can we specifically pass -DNO_WARN_X86_INTRINSICS through to the tblis compilation when on powerpc? I have been manually passing it through CFLAGS/CXXFLAGS on the relevant platforms that we frequent. https://github.com/nv-legate/quickstart/blob/branch-22.12/common.sh#L53-L58

[martin-frbg]

I'm now trying to build cunumeric on POWER9 myself, hitting the same issue and wondering what happened to this PR (and topic in general) ? tblis still has not changed anything, and even trying to feed a local build of a modified tblis to install.py does not appear to work. install.py --with-tblis <location> is documented in install.py, but the build still downloads a fresh copy of tblis everytime instead of using the (patched) sources from the provided location. (Reapplying the patch to the downloaded copy in _skbuild and rerunning install.py with the --no-clean option also does not help - still overwritten with a fresh copy...) This is on a shared machine, so I cannot easily edit the system headers

[manopapad]

I have opened a tblis PR that fixes the build on ARM. I believe it also fixes the build on PowerPC, but I don't have access to a PowerPC machine to test it myself. You can try telling cuNumeric to use the PR branch directly during compilation and see if that fixes it for you:

diff --git a/cmake/thirdparty/get_tblis.cmake b/cmake/thirdparty/get_tblis.cmake
index b02afbd7..d7f1d289 100644
--- a/cmake/thirdparty/get_tblis.cmake
+++ b/cmake/thirdparty/get_tblis.cmake
@@ -171,11 +171,11 @@ function(find_or_configure_tblis)
 endfunction()

 if(NOT DEFINED cunumeric_TBLIS_BRANCH)
-  set(cunumeric_TBLIS_BRANCH master)
+  set(cunumeric_TBLIS_BRANCH arm-build)
 endif()

 if(NOT DEFINED cunumeric_TBLIS_REPOSITORY)
-  set(cunumeric_TBLIS_REPOSITORY https://github.com/devinamatthews/tblis.git)
+  set(cunumeric_TBLIS_REPOSITORY https://github.com/manopapad/tblis.git)
 endif()

 find_or_configure_tblis(VERSION          1.2.0

[martin-frbg]

Thank you - I can confirm now that this fixes the build on PowerPC as well (tested on host cfarm120 in the gcc compile farm, a 24core/192threads POWER10 machine running AlmaLinux9.3)