[PATCH] D26128: [X86][AVX512] Removing llvm x86 intrinsics for _mm_mask_move_{ss|sd} intrinsics.

[Ayman Musa via llvm-commits]

aymanmus added a comment.

This change has been reverted due to failure in build.
compilers do not accept nesting levels greater that 127, and the file AutoUpgrade.cpp contains a long sequence of "if else" statements which the compiler considers too deep.

The error I got:
"FAILED: lib/IR/CMakeFiles/LLVMCore.dir/AutoUpgrade.cpp.obj 
C:\Buildbot\Slave\llvm-clang-lld-x86_64-scei-ps4-windows10pro-fast\llvm.src\lib\IR\AutoUpgrade.cpp(1529): **fatal error C1061: compiler limit: blocks nested too deeply**"

A heavy refactor to this piece of code should be considered as an effort to reduce the use of llvm x86 intrinsics is being initiated, and as a result many new cases will be added to the same control flow in the future.


Repository:
  rL LLVM

https://reviews.llvm.org/D26128