https://gcc.gnu.org/bugzilla/show_bug.cgi?id=89125
--- Comment #3 from Steve Kargl <sgk at troutmask dot apl.washington.edu> --- On Wed, Jan 30, 2019 at 08:06:44PM +0000, dominiq at lps dot ens.fr wrote: > https://gcc.gnu.org/bugzilla/show_bug.cgi?id=89125 > > --- Comment #2 from Dominique d'Humieres <dominiq at lps dot ens.fr> --- > See pr31249. > Don't see the relevance in that I'm not advocating that the middle-end transform the sin(x), cos(x) pair into a complex function cexpi. function foo(x) complex foo real,intent(in) :: x foo = cmplx(cos(x),sin(x)) end gfcx -o - -S -O3 a.f90 .file "a.f90" .text .p2align 4 .globl foo_ .type foo_, @function foo_: .LFB0: .cfi_startproc subq $24, %rsp .cfi_def_cfa_offset 32 movss (%rdi), %xmm2 movaps %xmm2, %xmm0 movss %xmm2, 4(%rsp) call cosf movss 4(%rsp), %xmm2 movss %xmm0, (%rsp) movaps %xmm2, %xmm0 call sinf movss (%rsp), %xmm1 movss %xmm0, 12(%rsp) movss %xmm1, 8(%rsp) movq 8(%rsp), %xmm0 addq $24, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE0: .size foo_, .-foo_ .ident "GCC: (GNU) 9.0.1 20190125 (experimental)" .section .note.GNU-stack,"",@progbits On FreeBSD and suspect with glibc, there is a clear win Single precision, 100M calls for x in [0, 20000) ./testf -s -n 100 -m 0 -M 20000 100M sincosf calls in 4.235 seconds. 100M sinf and 100M cosf calls in 5.330 seconds. Double precision, 100M calls for x in [0, 20000) ./testd -s -n 100 -m 0 -M 20000 100M sincos calls in 5.222 seconds. 100M sin and 100M cos calls in 8.105 seconds.
