[all-commits] [llvm/llvm-project] 513f9c: [libc][math] Add float-only implementation for sin...

[lntue via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: 513f9c50e74e55b7f8dedf65105bdc1083d4f83d
      https://github.com/llvm/llvm-project/commit/513f9c50e74e55b7f8dedf65105bdc1083d4f83d
  Author: lntue <lntue at google.com>
  Date:   2025-10-21 (Tue, 21 Oct 2025)

  Changed paths:
    M libc/src/__support/FPUtil/double_double.h
    M libc/src/__support/math/CMakeLists.txt
    M libc/src/__support/math/cosf.h
    A libc/src/__support/math/sincosf_float_eval.h
    M libc/src/math/generic/sinf.cpp
    M libc/test/src/math/CMakeLists.txt
    A libc/test/src/math/cosf_float_test.cpp
    M libc/test/src/math/exhaustive/CMakeLists.txt
    A libc/test/src/math/exhaustive/cosf_float_test.cpp
    M libc/test/src/math/exhaustive/exhaustive_test.h
    A libc/test/src/math/exhaustive/sinf_float_test.cpp
    A libc/test/src/math/sinf_float_test.cpp
    M utils/bazel/llvm-project-overlay/libc/BUILD.bazel

  Log Message:
  -----------
  [libc][math] Add float-only implementation for sinf / cosf. (#161680)

Based on the double precision's sin/cos fast path algorithm:

Step 1: Perform range reduction `y = x mod pi/8` with target errors <
2^-54.
This is because the worst case mod pi/8 for single precision is ~2^-31,
so to have up to 1 ULP errors from
the range reduction, the targeted errors should `be 2^(-31 - 23) =
2^-54`.

Step 2: Polynomial approximation
We use degree-5 and degree-4 polynomials to approximate sin and cos of
the reduced angle respectively.

Step 3: Combine the results using trig identities
```math
\begin{align*}
  \sin(x) &= \sin(y) \cdot \cos(k \cdot \frac{\pi}{8}) + \cos(y) \cdot \sin(k \cdot \frac{\pi}{8}) \\
  \cos(x) &= \cos(y) \cdot \cos(k \cdot \frac{\pi}{8}) - \sin(y) \cdot \sin(k \cdot \frac{\pi}{8})
\end{align*}
```

Overall errors: <= 3 ULPs for default rounding modes (tested
exhaustively).

Current limitation: large range reduction requires FMA instructions for
binary32. This restriction will be removed in the followup PR.

---------

Co-authored-by: Petr Hosek <phosek at google.com>



To unsubscribe from these emails, change your notification settings at https://github.com/llvm/llvm-project/settings/notifications