Fix neon cmp* functions for bf16.

The current impl corrupts the comparison masks when converting
from float back to bfloat16.  The resulting masks are then
no longer all zeros or all ones, which breaks when used with
`pselect` (e.g. in `pmin<PropagateNumbers>`).  This was
causing `packetmath_15` to fail on arm.

Introducing a simple `F32MaskToBf16Mask` corrects this (takes
the lower 16-bits for each float mask).
1 file changed
tree: 98aac84d6faf17d02688a2a4478c4c582932a812
  1. bench/
  2. blas/
  3. ci/
  4. cmake/
  5. debug/
  6. demos/
  7. doc/
  8. Eigen/
  9. failtest/
  10. lapack/
  11. scripts/
  12. test/
  13. unsupported/
  14. .gitignore
  15. .gitlab-ci.yml
  16. .hgeol
  17. CMakeLists.txt
  18. COPYING.APACHE
  19. COPYING.BSD
  20. COPYING.GPL
  21. COPYING.LGPL
  22. COPYING.MINPACK
  23. COPYING.MPL2
  24. COPYING.README
  25. CTestConfig.cmake
  26. CTestCustom.cmake.in
  27. eigen3.pc.in
  28. INSTALL
  29. README.md
  30. signature_of_eigen3_matrix_library
README.md

Eigen is a C++ template library for linear algebra: matrices, vectors, numerical solvers, and related algorithms.

For more information go to http://eigen.tuxfamily.org/.

For pull request, bug reports, and feature requests, go to https://gitlab.com/libeigen/eigen.