https://gcc.gnu.org/bugzilla/show_bug.cgi?id=114944
--- Comment #2 from John Platts <john_platts at hotmail dot com> ---
Here is more optimal codegen for SSE2ShuffleI8 on x86_64:
SSE2ShuffleI8(long long __vector(2), long long __vector(2)):
pand xmm1, XMMWORD PTR .LC0[rip]
movaps XMMWORD PTR [rsp-24], xmm0
movd eax, xmm1
movzx eax, al
movzx eax, BYTE PTR [rsp-24+rax]
movaps XMMWORD PTR [rsp-40], xmm1
movd xmm0, eax
movzx eax, BYTE PTR [rsp-39]
movzx eax, BYTE PTR [rsp-24+rax]
movd xmm2, eax
movzx eax, BYTE PTR [rsp-38]
punpcklbw xmm0, xmm2
movzx eax, BYTE PTR [rsp-24+rax]
movd xmm8, eax
movzx eax, BYTE PTR [rsp-37]
movzx eax, BYTE PTR [rsp-24+rax]
movd xmm2, eax
movzx eax, BYTE PTR [rsp-36]
punpcklbw xmm8, xmm2
movzx eax, BYTE PTR [rsp-24+rax]
punpcklwd xmm0, xmm8
movd xmm5, eax
movzx eax, BYTE PTR [rsp-35]
movzx eax, BYTE PTR [rsp-24+rax]
movd xmm2, eax
movzx eax, BYTE PTR [rsp-34]
punpcklbw xmm5, xmm2
movzx eax, BYTE PTR [rsp-24+rax]
movd xmm7, eax
movzx eax, BYTE PTR [rsp-33]
movzx eax, BYTE PTR [rsp-24+rax]
movd xmm2, eax
movzx eax, BYTE PTR [rsp-32]
punpcklbw xmm7, xmm2
movzx eax, BYTE PTR [rsp-24+rax]
punpcklwd xmm5, xmm7
punpckldq xmm0, xmm5
movd xmm3, eax
movzx eax, BYTE PTR [rsp-31]
movzx eax, BYTE PTR [rsp-24+rax]
movd xmm4, eax
movzx eax, BYTE PTR [rsp-30]
punpcklbw xmm3, xmm4
movzx eax, BYTE PTR [rsp-24+rax]
movd xmm6, eax
movzx eax, BYTE PTR [rsp-29]
movzx eax, BYTE PTR [rsp-24+rax]
movd xmm2, eax
movzx eax, BYTE PTR [rsp-28]
punpcklbw xmm6, xmm2
movzx eax, BYTE PTR [rsp-24+rax]
punpcklwd xmm3, xmm6
movd xmm2, eax
movzx eax, BYTE PTR [rsp-27]
movzx eax, BYTE PTR [rsp-24+rax]
movd xmm4, eax
movzx eax, BYTE PTR [rsp-26]
punpcklbw xmm2, xmm4
movzx eax, BYTE PTR [rsp-24+rax]
movd xmm4, eax
movzx eax, BYTE PTR [rsp-25]
movzx eax, BYTE PTR [rsp-24+rax]
movd xmm1, eax
punpcklbw xmm4, xmm1
movdqa xmm1, xmm2
movdqa xmm2, xmm3
punpcklwd xmm1, xmm4
punpckldq xmm2, xmm1
punpcklqdq xmm0, xmm2
ret
.LC0:
.quad 1085102592571150095
.quad 1085102592571150095
