tree d71e12113f9eb31d789d059aa4073b07a44409fb
parent 48c0534c2eafe23fe8efba8c913109f30f6a294c
author Damjan Marion <damarion@cisco.com> 1635415335 +0200
committer Florin Coras <florin.coras@gmail.com> 1635436465 +0000

ip: improve csum fold on x86_64

New code seems to be 1.5 clocks faster.

old:
	mov    eax,edi
	shr    rdi,0x20
	add    rdi,rax
	movzx  edx,di
	shr    rdi,0x10
	add    rdx,rdi
	movzx  eax,dx
	shr    rdx,0x10
	add    rax,rdx
	mov    rdx,rax
	shr    rdx,0x10
	add    eax,edx

new:
	mov    rax,rdi
	shr    rax,0x20
	add    eax,edi
	mov    edi,0x10
	shrx   edi,eax,edi
	adc    ax,di
	adc    ax,0x0

Type: improvement
Change-Id: I3c565812c67ff4c3db197a9d4137a6c131b5b66c
Signed-off-by: Damjan Marion <damarion@cisco.com>