[libc-commits] [libc] [libc][math] Implement double precision sin correctly rounded to all rounding modes. (PR #95736)

[via libc-commits]

zimmermann6 wrote:

On a ntel(R) Xeon(R) Silver 4214 I get:
```
zimmerma at croissant:~/svn/core-math$ LIBM=$L CORE_MATH_PERF_MODE=rdtsc ./perf.sh sin
GNU libc version: 2.38
GNU libc release: stable
[####################] 100 %
Ntrial = 20 ; Min = 61.961 + 0.506 clc/call; Median-Min = 0.548 clc/call; Max = 62.771 clc/call;
[####################] 100 %
Ntrial = 20 ; Min = 32.978 + 0.803 clc/call; Median-Min = 0.627 clc/call; Max = 35.687 clc/call;
[####################] 100 %
Ntrial = 20 ; Min = 30.249 + 0.946 clc/call; Median-Min = 0.809 clc/call; Max = 32.471 clc/call;
zimmerma at croissant:~/svn/core-math$ LIBM=$L CORE_MATH_PERF_MODE=rdtsc PERF_ARGS=--latency ./perf.sh sin
GNU libc version: 2.38
GNU libc release: stable
[####################] 100 %
Ntrial = 20 ; Min = 88.747 + 0.985 clc/call; Median-Min = 0.854 clc/call; Max = 90.962 clc/call;
[####################] 100 %
Ntrial = 20 ; Min = 58.363 + 0.368 clc/call; Median-Min = 0.270 clc/call; Max = 59.351 clc/call;
[####################] 100 %
Ntrial = 20 ; Min = 62.464 + 0.795 clc/call; Median-Min = 0.687 clc/call; Max = 64.072 clc/call;
```

https://github.com/llvm/llvm-project/pull/95736