search

conservation-laws / ryujin

High-performance high-order finite element solver for hyperbolic conservation equations
https://conservation-laws.org
Other
112 stars 27 forks source link

SIMD: use generic `std::pow` overloads for ARM Neon instructions #83

Closed tamiko closed 6 months ago

tamiko commented 6 months ago

@ejtovar Would you mind to test this PR?

ejtovar commented 6 months ago

@tamiko Code compiles and links, but the test common/fast_pow fails. This is expected, yes?

tamiko commented 6 months ago

@ejtovar We will need a test output variant for M{1,2,3} Neon.

Edit: I have guesstimated the output and added common/fast_pow.output.osx-m1-neon. Can you test again? If it fails, please update this output variant and push your changes to the branch.

ejtovar commented 6 months ago

Test passes. Merging now.