Menu
▾
▴
[Ggnfs-commits] SF.net SVN: ggnfs:[419] trunk/src/experimental/lasieve4_64/x64
|
From: <brg...@us...> - 2011-10-22 22:07:12
|
Revision: 419
http://ggnfs.svn.sourceforge.net/ggnfs/?rev=419&view=rev
Author: brgladman
Date: 2011-10-22 22:07:05 +0000 (Sat, 22 Oct 2011)
Log Message:
-----------
correct Windows assembler handling of RIP relative addressing
Modified Paths:
--------------
trunk/src/experimental/lasieve4_64/x64/asm-zeit.asm
trunk/src/experimental/lasieve4_64/x64/mpqs_gauss.asm
trunk/src/experimental/lasieve4_64/x64/mpqs_sieve.asm
trunk/src/experimental/lasieve4_64/x64/mpqs_td.asm
Modified: trunk/src/experimental/lasieve4_64/x64/asm-zeit.asm
===================================================================
--- trunk/src/experimental/lasieve4_64/x64/asm-zeit.asm 2011-10-22 19:20:40 UTC (rev 418)
+++ trunk/src/experimental/lasieve4_64/x64/asm-zeit.asm 2011-10-22 22:07:05 UTC (rev 419)
@@ -16,16 +16,16 @@
rdtsc
shl rdx, 32
or rax, rdx
- mov rdx, [rip+asmzeitcounter]
- sub [rdx+rdi*8], rax
+ mov rdx, [rel asmzeitcounter]
+ sub [rdx+rcx*8], rax
ret
LEAF_PROC zeitB
rdtsc
shl rdx, 32
or rax, rdx
- mov rdx, [rip+asmzeitcounter]
- add [rdx+rdi*8], rax
+ mov rdx, [rel asmzeitcounter]
+ add [rdx+rcx*8], rax
ret
end
Modified: trunk/src/experimental/lasieve4_64/x64/mpqs_gauss.asm
===================================================================
--- trunk/src/experimental/lasieve4_64/x64/mpqs_gauss.asm 2011-10-22 19:20:40 UTC (rev 418)
+++ trunk/src/experimental/lasieve4_64/x64/mpqs_gauss.asm 2011-10-22 22:07:05 UTC (rev 419)
@@ -28,10 +28,10 @@
FRAME_PROC asm_gauss, 0, reg_save_list
-.1: mov eax, [rip+mpqs_gauss_k]
+.1: mov eax, [rel mpqs_gauss_k]
dec eax
mov r8d, 1
- mov [rip+mpqs_gauss_k], eax
+ mov [rel mpqs_gauss_k], eax
jl .14
mov ecx, eax
and ecx, 31
@@ -39,31 +39,31 @@
mov ecx, eax
shr ecx, 5 ; %r11d
mov r11d, ecx
- mov ecx, [rip+mpqs_gauss_j]
+ mov ecx, [rel mpqs_gauss_j]
mov rdi, mpqs_gauss_c
- mov rsi, [rip+mpqs_gauss_row]
+ mov rsi, [rel mpqs_gauss_row]
mov eax, r11d
- mov edx, [rip+mpqs_gauss_n32]
+ mov edx, [rel mpqs_gauss_n32]
mov rsi, [rsi+rcx*8]
sub rax, rdx
dec ecx
lea rsi, [rsi+rax*4]
- mov eax, [rip+mpqs_gauss_n32]
+ mov eax, [rel mpqs_gauss_n32]
align 16
.2: inc ecx
-.3: cmp ecx, [rip+mpqs_gauss_m]
+.3: cmp ecx, [rel mpqs_gauss_m]
jnc .15
lea rsi, [rsi+rax*4]
mov edx, [rsi]
and edx, r8d
jz .2
-.4: cmp ecx, [rip+mpqs_gauss_j]
+.4: cmp ecx, [rel mpqs_gauss_j]
jz .7
- mov rsi, [rip+mpqs_gauss_row]
+ mov rsi, [rel mpqs_gauss_row]
mov rax, [rsi+rcx*8] ; mpqs_gauss_row[j]
- mov edx, [rip+mpqs_gauss_j]
+ mov edx, [rel mpqs_gauss_j]
mov r9, [rsi+rdx*8] ; mpqs_gauss_row[mpqs_gauss_j]
xor rdx, rdx
.5: movq mm0, [rax+rdx*4]
@@ -71,20 +71,20 @@
movq [r9+rdx*4], mm0
movq [rax+rdx*4], mm1
add rdx, 2
- cmp edx, [rip+mpqs_gauss_n32]
+ cmp edx, [rel mpqs_gauss_n32]
jc .5
-.6: mov ecx, [rip+mpqs_gauss_j]
+.6: mov ecx, [rel mpqs_gauss_j]
.7: mov rsi, mpqs_gauss_d
- mov eax, [rip+mpqs_gauss_k]
+ mov eax, [rel mpqs_gauss_k]
mov [rsi+rax*2], cx
mov [rdi+rcx*2], ax
- inc dword[rip+mpqs_gauss_j]
+ inc dword[rel mpqs_gauss_j]
mov r10d, ecx
- mov rsi, [rip+mpqs_gauss_mat]
+ mov rsi, [rel mpqs_gauss_mat]
mov rdi, mpqs_gauss_col
mov edx, r11d
test ecx, ecx
- mov eax, [rip+mpqs_gauss_n32]
+ mov eax, [rel mpqs_gauss_n32]
lea rsi, [rsi+rdx*4]
mov rdx, 2
jz .9
@@ -102,7 +102,7 @@
lea rsi, [rsi+rax*4]
jc .8
.9: inc ecx
- cmp ecx, [rip+mpqs_gauss_m]
+ cmp ecx, [rel mpqs_gauss_m]
lea rsi, [rsi+rax*4]
jnc .11
@@ -114,14 +114,14 @@
cmovnz r9, rdx
inc rcx
add rdi, r9
- cmp ecx, [rip+mpqs_gauss_m]
+ cmp ecx, [rel mpqs_gauss_m]
lea rsi, [rsi+rax*4]
jc .10
.11: mov r8, mpqs_gauss_col
cmp rdi, r8
jz .1
mov ecx, r10d
- mov rsi, [rip+mpqs_gauss_row]
+ mov rsi, [rel mpqs_gauss_row]
mov rsi, [rsi+rcx*8]
mov edx, r11d
cmp edx, 2
@@ -133,7 +133,7 @@
align 16
.12: movzx r9, word[r8]
- mov rax, [rip+mpqs_gauss_row]
+ mov rax, [rel mpqs_gauss_row]
mov rax, [rax+r9*8]
xor ecx, ecx
.13: mov edx, [rsi+rcx*4]
@@ -150,14 +150,14 @@
EXIT_PROC reg_save_list
.15: mov rdi, mpqs_gauss_d
- mov eax, [rip+mpqs_gauss_k]
+ mov eax, [rel mpqs_gauss_k]
mov r9w, -1
mov [rdi+rax*2], r9w
mov r10d, ecx
jmp .14
.16 movzx r9, word[r8]
movq mm0, [rsi]
- mov rsi, [rip+mpqs_gauss_row]
+ mov rsi, [rel mpqs_gauss_row]
mov rax, [rsi+r9*8]
.17: movq mm1, [rax]
lea r8, [r8+2]
@@ -171,7 +171,7 @@
.18: movq mm0, [rsi]
movzx r9, word[r8]
movq mm2, [rsi+8]
- mov rsi, [rip+mpqs_gauss_row]
+ mov rsi, [rel mpqs_gauss_row]
mov rax, [rsi+r9*8]
.19: movq mm1, [rax]
lea r8, [r8+2]
@@ -189,7 +189,7 @@
movzx r9, word[r8]
movq mm2, [rsi+8]
movq mm4, [rsi+16]
- mov rsi, [rip+mpqs_gauss_row]
+ mov rsi, [rel mpqs_gauss_row]
mov rax, [rsi+r9*8]
.21: movq mm1, [rax]
lea r8, [r8+2]
Modified: trunk/src/experimental/lasieve4_64/x64/mpqs_sieve.asm
===================================================================
--- trunk/src/experimental/lasieve4_64/x64/mpqs_sieve.asm 2011-10-22 19:20:40 UTC (rev 418)
+++ trunk/src/experimental/lasieve4_64/x64/mpqs_sieve.asm 2011-10-22 22:07:05 UTC (rev 419)
@@ -17,8 +17,8 @@
FRAME_PROC asm_sieve, 0, reg_save_list
- movzx rax, word[rip+mpqs_sievebegin]
- movzx r10, word[rip+mpqs_sievelen]
+ movzx rax, word[rel mpqs_sievebegin]
+ movzx r10, word[rel mpqs_sievelen]
mov rcx, mpqs_FB
mov r8, mpqs_FB_start
mov r9, mpqs_FB_log
@@ -27,7 +27,7 @@
lea r9, [r9+rax]
mov r11, r10
shr r10, 2
- mov r13, [rip+mpqs_sievearray]
+ mov r13, [rel mpqs_sievearray]
mov r15, r13
add r15, r11
Modified: trunk/src/experimental/lasieve4_64/x64/mpqs_td.asm
===================================================================
--- trunk/src/experimental/lasieve4_64/x64/mpqs_td.asm 2011-10-22 19:20:40 UTC (rev 418)
+++ trunk/src/experimental/lasieve4_64/x64/mpqs_td.asm 2011-10-22 22:07:05 UTC (rev 419)
@@ -58,10 +58,10 @@
pxor xmm5, xmm5
movzx r9, word[rdi+8] ; %r9
- movzx rcx, word[rip+mpqs_td_begin]
+ movzx rcx, word[rel mpqs_td_begin]
mov rbx, mpqs_FB_inv_info
mov r11, mpqs_FB_start
- mov r13w, [rip+mpqs_nFBk_1]
+ mov r13w, [rel mpqs_nFBk_1]
add r13w, 4
movaps xmm0, [rbx]
movaps xmm1, [rbx+16]
@@ -139,7 +139,7 @@
jmp .1
.2:
mov r10, mpqs_FB
- movzx rdx, word[rip+mpqs_nFBk_1]
+ movzx rdx, word[rel mpqs_nFBk_1]
add rdx, rdx
add rdx, rdx
sub r10, rdx
@@ -163,7 +163,7 @@
inc r9
cmp r9, 27
jnc .16
- mov cx, [rip+mpqs_nFBk_1]
+ mov cx, [rel mpqs_nFBk_1]
mov [rdi+r9*2+8], cx
shr r8, 1
jmp .5
@@ -195,7 +195,7 @@
mul r11d
mov r8d, eax
mov r11, mpqs_FB_inv
- movzx rcx, word[rip+mpqs_nFBk_1]
+ movzx rcx, word[rel mpqs_nFBk_1]
add rcx, rcx
add rcx, rcx
sub r11, rcx
@@ -223,7 +223,7 @@
jz .15
mov r11, mpqs_FBk_inv
mov r10, mpqs_FBk
- movzx rsi, word[rip+mpqs_nFBk]
+ movzx rsi, word[rel mpqs_nFBk]
inc rsi
.10: dec rsi
mov eax, r8d
@@ -247,8 +247,8 @@
jz .15
mov r11, mpqs_FB_A_inv
mov r10, mpqs_Adiv_all
- mov r13w, [rip+mpqs_nFB]
- add r13w, [rip+mpqs_nFBk]
+ mov r13w, [rel mpqs_nFB]
+ add r13w, [rel mpqs_nFBk]
movzx rsi, word[mpqs_nAdiv_total]
inc rsi
.13: dec rsi
This was sent by the SourceForge.net collaborative development platform, the world's largest Open Source development site.
|
