EliasFleckenstein03
/
plan9front
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

libmp: 386/amd64 mpvec*(): replace conditional branches with ADC/SBB instructions

front
cinap_lenrek 2015-11-20 06:25:01 +01:00
parent 55d3e11f0f
commit 2559e19e19
6 changed files with 21 additions and 50 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

7
sys/src/libmp/386/mpvecadd.s
View File

@ -35,11 +35,8 @@ _add1:
/* sum[alen] = carry */
_addend:
JC _addcarry
MOVL $0,(DI)(BP*4)
RET
_addcarry:
MOVL $1,(DI)(BP*4)
ADCL $0, CX
MOVL CX, (DI)(BP*4)
RET
/* sum[blen:alen-1],carry = a[blen:alen-1] + 0 */

8
sys/src/libmp/386/mpvecdigmuladd.s
View File

@ -36,13 +36,9 @@ _muladdloop:
MOVL (SI)(BP*4),AX /* lo = b[i] */
MULL BX /* hi, lo = b[i] * m */
ADDL CX,AX /* lo += oldhi */
JCC _muladdnocarry1
INCL DX /* hi += carry */
_muladdnocarry1:
ADCL $0, DX /* hi += carry */
ADDL AX,(DI)(BP*4) /* p[i] += lo */
JCC _muladdnocarry2
INCL DX /* hi += carry */
_muladdnocarry2:
ADCL $0, DX /* hi += carry */
MOVL DX,CX /* oldhi = hi */
INCL BP /* i++ */
JNZ _muladdloop

20
sys/src/libmp/386/mpvecdigmulsub.s
View File

@ -33,21 +33,15 @@ _mulsubloop:
MOVL (SI)(BP*4),AX /* lo = b[i] */
MULL BX /* hi, lo = b[i] * m */
ADDL 0(SP),AX /* lo += oldhi */
JCC _mulsubnocarry1
INCL DX /* hi += carry */
_mulsubnocarry1:
ADCL $0, DX /* hi += carry */
SUBL AX,(DI)(BP*4)
JCC _mulsubnocarry2
INCL DX /* hi += carry */
_mulsubnocarry2:
ADCL $0, DX /* hi += carry */
MOVL DX,0(SP)
INCL BP
LOOP _mulsubloop
MOVL 0(SP),AX
SUBL AX,(DI)(BP*4)
JCC _mulsubnocarry3
MOVL $-1,AX
RET
_mulsubnocarry3:
MOVL $1,AX
MOVL CX, AX
MOVL 0(SP),BX
SUBL BX,(DI)(BP*4)
SBBL CX, AX
ORL $1, AX
RET

9
sys/src/libmp/amd64/mpvecadd.s
View File

@ -9,7 +9,6 @@ TEXT mpvecadd(SB),$0
MOVL alen+8(FP),DX
MOVL blen+24(FP),CX
/* MOVL a+0(FP),SI */
MOVQ RARG, SI
MOVQ b+16(FP),BX
SUBL CX,DX
@ -35,12 +34,10 @@ _add1:
LOOP _addloop2
/* sum[alen] = carry */
_addend:
JC _addcarry
MOVL $0,(DI)(BP*4)
RET
_addcarry:
MOVL $1,(DI)(BP*4)
ADCL $0, CX
MOVL CX,(DI)(BP*4)
RET
/* sum[blen:alen-1],carry = a[blen:alen-1] + 0 */

9
sys/src/libmp/amd64/mpvecdigmuladd.s
View File

@ -22,7 +22,6 @@
*/
TEXT mpvecdigmuladd(SB),$0
/* MOVQ b+0(FP),SI */
MOVQ RARG,SI
MOVL n+8(FP),CX
MOVL m+16(FP),BX
@ -37,13 +36,9 @@ _muladdloop:
MOVL (SI)(BP*4),AX /* lo = b[i] */
MULL BX /* hi, lo = b[i] * m */
ADDL CX,AX /* lo += oldhi */
JCC _muladdnocarry1
INCL DX /* hi += carry */
_muladdnocarry1:
ADCL $0, DX /* hi += carry */
ADDL AX,(DI)(BP*4) /* p[i] += lo */
JCC _muladdnocarry2
INCL DX /* hi += carry */
_muladdnocarry2:
ADCL $0, DX /* hi += carry */
MOVL DX,CX /* oldhi = hi */
INCQ BP /* i++ */
JNZ _muladdloop

18
sys/src/libmp/amd64/mpvecdigmulsub.s
View File

@ -22,8 +22,6 @@
*
*/
TEXT mpvecdigmulsub(SB),$0
/* MOVL b+0(FP),SI */
MOVQ RARG,SI
MOVL n+8(FP),CX
MOVL m+16(FP),BX
@ -34,20 +32,14 @@ _mulsubloop:
MOVL (SI)(BP*4),AX /* lo = b[i] */
MULL BX /* hi, lo = b[i] * m */
ADDL R8,AX /* lo += oldhi */
JCC _mulsubnocarry1
INCL DX /* hi += carry */
_mulsubnocarry1:
ADCL $0, DX /* hi += carry */
SUBL AX,(DI)(BP*4)
JCC _mulsubnocarry2
INCL DX /* hi += carry */
_mulsubnocarry2:
ADCL $0, DX /* hi += carry */
MOVL DX,R8
INCL BP
LOOP _mulsubloop
MOVL CX, AX
SUBL R8,(DI)(BP*4)
JCC _mulsubnocarry3
MOVQ $-1,AX
RET
_mulsubnocarry3:
MOVQ $1,AX
SBBQ CX, AX
ORQ $1, AX
RET