[libc-commits] [libc] [libc][math][c23] Add exp2m1f16 C23 math function (PR #105690)

[Joseph Huber via libc-commits]

jhuber6 wrote:

```[2220/2586] Running hermetic test libc.test.src.math.smoke.exp2m1f16_test.__hermetic__
FAILED: libc/test/src/math/smoke/libc.test.src.math.smoke.exp2m1f16_test.__hermetic__-cmd /home/jhuber/Documents/llvm/llvm-project/build/runtimes/runtimes-amdgcn-amd-amdhsa-bins/libc/test/src/math/smoke/libc.test.src.math.smoke.exp2m1f16_test.__hermetic__-cmd 
cd /home/jhuber/Documents/llvm/llvm-project/build/runtimes/runtimes-amdgcn-amd-amdhsa-bins/libc/test/src/math/smoke && /home/jhuber/Documents/llvm/llvm-project/build/bin/amdhsa-loader /home/jhuber/Documents/llvm/llvm-project/build/runtimes/runtimes-amdgcn-amd-amdhsa-bins/libc/test/src/math/smoke/libc.test.src.math.smoke.exp2m1f16_test.__hermetic__.__build__
[==========] Running 3 tests from 1 test suite.
[ RUN      ] LlvmLibcExp2m1f16Test.SpecialNumbers
/home/jhuber/Documents/llvm/llvm-project/libc/test/src/math/smoke/exp2m1f16_test.cpp:38: FAILURE
Failed to match (__llvm_libc_20_0_0_git::exp2m1f16(neg_zero)) against LIBC_NAMESPACE::testing::getMatcher< LIBC_NAMESPACE::testing::TestCond::EQ>((neg_zero)).
Expected floating point value: 0x8000 = (S: 1, E: 0x0000, M: 0x0000)
Actual floating point value: 0x0000 = (S: 0, E: 0x0000, M: 0x0000)
```
Hm, works fine on NVPTX but fails on AMDGPU. Wonder why the sign is off.

https://github.com/llvm/llvm-project/pull/105690