[PATCH] D99741: [RISCV][Clang] Add some RVV Floating-Point intrinsic functions. (vfclass, vfmerge, vfrec7, vfrsqrt7, vfsqrt)

Tue Apr 27 20:47:12 PDT 2021

craig.topper added a comment.

In D99741#2721596 <https://reviews.llvm.org/D99741#2721596>, @thakis wrote:

> In D99741#2721235 <https://reviews.llvm.org/D99741#2721235>, @craig.topper wrote:
>
>> I believe we had reductions at builds 7601 and 8175.
>>
>> It looks like there may have been an increase between builds 7979 and 7991 the build failed for a while there. I suspect "e951b04 <https://reviews.llvm.org/rGe951b045bfbf7974d66a4d5c040bbcf4d513c861> [AArch64][SVE] Regression test all ACLE tests with C++" as it added a quite a few RUN lines.
>>
>> I think we also caused an increase at build 8202 when these 4 commits went in 
>> bd32c2d <https://reviews.llvm.org/rGbd32c2d64842c8d474bcce0e73dcd8918b3e0f2f> [RISCV] Implement the vwcvt{u}.x.x.v/vncvt.x.x.w builtin.                                                                                                                                                                                                               
>> 645c5f2 <https://reviews.llvm.org/rG645c5f27a84d795fc658b49c702d77cc6145bfec> [RISCV] Implement the pseudo compare builtin.                                                                                                                                                                                                                           
>> bfb3fca <https://reviews.llvm.org/rGbfb3fca7e110a367c9589495575913d83bea7095> [RISCV] Implement the vfabs.v/vfneg.v builtin.                                                                                                                                                                                                                          
>> 4b24341 <https://reviews.llvm.org/rG4b2434172cc15ccd32c5c9110a3a4df18b8fba5e> [RISCV] Implement the vmmv.m/vmnot.m builtin.
>
> Any disagreement about reverting riscv builtin changes until there's some way to test them without increasing check-clang time by 20%?

We're preparing a patch to remove to stop testing both rv32 and rv64 on every test. That should reduce the time by half. What is an acceptable number?

Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D99741/new/

https://reviews.llvm.org/D99741