LBRYCommunity/cpuminer
1
0
Fork 0
Code Issues 3 Pull requests Projects Releases Packages Wiki Activity

Switch to RIP-relative addressing on x86-64

Browse source
This commit is contained in:
pooler 2012-02-28 20:51:11 +01:00
parent 03d012e86c
commit 6fc88331d6
1 changed files with 11 additions and 9 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

20
scrypt-x64.S
View file

@ -117,18 +117,18 @@ _SHA256_InitState_4way:
pushq %rdi pushq %rdi
movq %rcx, %rdi movq %rcx, %rdi
#endif #endif
movdqa sha256_4h+0, %xmm0 movdqa sha256_4h+0(%rip), %xmm0
movdqa sha256_4h+16, %xmm1 movdqa sha256_4h+16(%rip), %xmm1
movdqa sha256_4h+32, %xmm2 movdqa sha256_4h+32(%rip), %xmm2
movdqa sha256_4h+48, %xmm3 movdqa sha256_4h+48(%rip), %xmm3
movdqu %xmm0, 0(%rdi) movdqu %xmm0, 0(%rdi)
movdqu %xmm1, 16(%rdi) movdqu %xmm1, 16(%rdi)
movdqu %xmm2, 32(%rdi) movdqu %xmm2, 32(%rdi)
movdqu %xmm3, 48(%rdi) movdqu %xmm3, 48(%rdi)
movdqa sha256_4h+64, %xmm0 movdqa sha256_4h+64(%rip), %xmm0
movdqa sha256_4h+80, %xmm1 movdqa sha256_4h+80(%rip), %xmm1
movdqa sha256_4h+96, %xmm2 movdqa sha256_4h+96(%rip), %xmm2
movdqa sha256_4h+112, %xmm3 movdqa sha256_4h+112(%rip), %xmm3
movdqu %xmm0, 64(%rdi) movdqu %xmm0, 64(%rdi)
movdqu %xmm1, 80(%rdi) movdqu %xmm1, 80(%rdi)
movdqu %xmm2, 96(%rdi) movdqu %xmm2, 96(%rdi)
@ -192,6 +192,7 @@ _SHA256_Transform_4way:
p2bswap_rsi_rsp 14 p2bswap_rsi_rsp 14
jmp sha256_transform_4way_extend jmp sha256_transform_4way_extend
.p2align 5
sha256_transform_4way_block_copy: sha256_transform_4way_block_copy:
movdqu 0*16(%rsi), %xmm0 movdqu 0*16(%rsi), %xmm0
movdqu 1*16(%rsi), %xmm1 movdqu 1*16(%rsi), %xmm1
@ -306,10 +307,11 @@ sha256_transform_4way_extend_loop:
movdqu 96(%rdi), %xmm9 movdqu 96(%rdi), %xmm9
movdqu 112(%rdi), %xmm10 movdqu 112(%rdi), %xmm10
leaq sha256_4k(%rip), %rcx
xorq %rax, %rax xorq %rax, %rax
sha256_transform_4way_main_loop: sha256_transform_4way_main_loop:
movdqa (%rsp, %rax), %xmm6 movdqa (%rsp, %rax), %xmm6
paddd sha256_4k(%rax), %xmm6 paddd (%rcx, %rax), %xmm6
paddd %xmm10, %xmm6 paddd %xmm10, %xmm6
movdqa %xmm0, %xmm1 movdqa %xmm0, %xmm1