Browse Source

Merge pull request #4190 from martin-frbg/issue4186-2

Allow negative INCX in the ?NRM2 kernels
tags/v0.3.24
Martin Kroeker GitHub 2 years ago
parent
commit
ebc157fcc9
No known key found for this signature in database GPG Key ID: 4AEE18F83AFDEB23
38 changed files with 38 additions and 38 deletions
  1. +1
    -1
      kernel/arm/nrm2.c
  2. +1
    -1
      kernel/arm/znrm2.c
  3. +1
    -1
      kernel/loongarch64/cnrm2.S
  4. +1
    -1
      kernel/loongarch64/dnrm2.S
  5. +1
    -1
      kernel/loongarch64/snrm2.S
  6. +1
    -1
      kernel/loongarch64/znrm2.S
  7. +1
    -1
      kernel/mips/nrm2.c
  8. +1
    -1
      kernel/mips/znrm2.c
  9. +1
    -1
      kernel/mips64/cnrm2.S
  10. +1
    -1
      kernel/mips64/dnrm2.S
  11. +1
    -1
      kernel/mips64/snrm2.S
  12. +1
    -1
      kernel/mips64/znrm2.S
  13. +1
    -1
      kernel/power/cnrm2.S
  14. +1
    -1
      kernel/power/cnrm2_hummer.S
  15. +1
    -1
      kernel/power/cnrm2_ppc440.S
  16. +1
    -1
      kernel/power/dnrm2_hummer.S
  17. +1
    -1
      kernel/power/dnrm2_ppc440.S
  18. +1
    -1
      kernel/power/nrm2.S
  19. +1
    -1
      kernel/power/snrm2.S
  20. +1
    -1
      kernel/power/snrm2_hummer.S
  21. +1
    -1
      kernel/power/snrm2_ppc440.S
  22. +1
    -1
      kernel/power/znrm2.S
  23. +1
    -1
      kernel/power/znrm2_hummer.S
  24. +1
    -1
      kernel/power/znrm2_ppc440.S
  25. +1
    -1
      kernel/riscv64/nrm2.c
  26. +1
    -1
      kernel/riscv64/znrm2.c
  27. +1
    -1
      kernel/sparc/cnrm2.S
  28. +1
    -1
      kernel/sparc/dnrm2.S
  29. +1
    -1
      kernel/sparc/snrm2.S
  30. +1
    -1
      kernel/sparc/znrm2.S
  31. +1
    -1
      kernel/x86/nrm2.S
  32. +1
    -1
      kernel/x86/nrm2_sse.S
  33. +1
    -1
      kernel/x86/znrm2.S
  34. +1
    -1
      kernel/x86/znrm2_sse.S
  35. +1
    -1
      kernel/x86_64/nrm2.S
  36. +1
    -1
      kernel/x86_64/nrm2_sse.S
  37. +1
    -1
      kernel/x86_64/znrm2.S
  38. +1
    -1
      kernel/x86_64/znrm2_sse.S

+ 1
- 1
kernel/arm/nrm2.c View File

@@ -57,7 +57,7 @@ FLOAT CNAME(BLASLONG n, FLOAT *x, BLASLONG inc_x)
FLOAT absxi = 0.0;


if (n <= 0 || inc_x <= 0) return(0.0);
if (n <= 0 || inc_x == 0) return(0.0);
if ( n == 1 ) return( ABS(x[0]) );

n *= inc_x;


+ 1
- 1
kernel/arm/znrm2.c View File

@@ -57,7 +57,7 @@ FLOAT CNAME(BLASLONG n, FLOAT *x, BLASLONG inc_x)
BLASLONG inc_x2;
FLOAT temp;

if (n <= 0 || inc_x <= 0) return(0.0);
if (n <= 0 || inc_x == 0) return(0.0);

inc_x2 = 2 * inc_x;



+ 1
- 1
kernel/loongarch64/cnrm2.S View File

@@ -61,7 +61,7 @@ USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
fmov.d s2, s1
bge $r0, N, .L999
slli.d INCX, INCX, ZBASE_SHIFT
bge $r0, INCX, .L999
beq $r0, INCX, .L999
srai.d I, N, 2
bge $r0, I, .L25
LD a1, X, 0 * SIZE


+ 1
- 1
kernel/loongarch64/dnrm2.S View File

@@ -70,7 +70,7 @@ USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
MTC s1, $r0
bge $r0, N, .L999
slli.d INCX, INCX, BASE_SHIFT
bge $r0, INCX, .L999
beq $r0, INCX, .L999
move XX, X
NOP
LD a1, X, 0 * SIZE


+ 1
- 1
kernel/loongarch64/snrm2.S View File

@@ -61,7 +61,7 @@ USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
fmov.d s2, s1
bge $r0, N, .L999
slli.d INCX, INCX, BASE_SHIFT
bge $r0, INCX, .L999
beq $r0, INCX, .L999
srai.d I, N, 3
bne INCX, TEMP, .L20
bge $r0, I, .L15


+ 1
- 1
kernel/loongarch64/znrm2.S View File

@@ -64,7 +64,7 @@ USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
MTC s1, $r0
bge $r0, N, .L999
slli.d INCX, INCX, ZBASE_SHIFT
bge $r0, INCX, .L999
beq $r0, INCX, .L999
move XX, X
MOV s2, s1
srai.d I, N, 2


+ 1
- 1
kernel/mips/nrm2.c View File

@@ -57,7 +57,7 @@ FLOAT CNAME(BLASLONG n, FLOAT *x, BLASLONG inc_x)
FLOAT absxi = 0.0;


if (n <= 0 || inc_x <= 0) return(0.0);
if (n <= 0 || inc_x == 0) return(0.0);
if ( n == 1 ) return( ABS(x[0]) );

n *= inc_x;


+ 1
- 1
kernel/mips/znrm2.c View File

@@ -48,7 +48,7 @@ FLOAT CNAME(BLASLONG n, FLOAT *x, BLASLONG inc_x)
BLASLONG inc_x2;
FLOAT temp;

if (n <= 0 || inc_x <= 0) return(0.0);
if (n <= 0 || inc_x == 0) return(0.0);

inc_x2 = 2 * inc_x;



+ 1
- 1
kernel/mips64/cnrm2.S View File

@@ -77,7 +77,7 @@
blez N, .L999
mov.d s2, s1

blez INCX, .L999
beqz INCX, .L999
dsll INCX, INCX, ZBASE_SHIFT

dsra I, N, 2


+ 1
- 1
kernel/mips64/dnrm2.S View File

@@ -81,7 +81,7 @@
blez N, .L999
MTC $0, s1

blez INCX, .L999
beqz INCX, .L999
dsll INCX, INCX, BASE_SHIFT

move XX, X


+ 1
- 1
kernel/mips64/snrm2.S View File

@@ -77,7 +77,7 @@
blez N, .L999
mov.d s2, s1

blez INCX, .L999
beqz INCX, .L999
dsll INCX, INCX, BASE_SHIFT

bne INCX, TEMP, .L20


+ 1
- 1
kernel/mips64/znrm2.S View File

@@ -80,7 +80,7 @@
blez N, .L999
MTC $0, s1

blez INCX, .L999
beqz INCX, .L999
dsll INCX, INCX, ZBASE_SHIFT

move XX, X


+ 1
- 1
kernel/power/cnrm2.S View File

@@ -99,7 +99,7 @@
cmpwi cr0, N, 0
ble- LL(9999)
cmpwi cr0, INCX, 0
ble- LL(9999)
beq- LL(9999)

fmr f0, f1
fmr f2, f1


+ 1
- 1
kernel/power/cnrm2_hummer.S View File

@@ -119,7 +119,7 @@
cmpwi cr0, N, 0
ble LL(99)
cmpwi cr0, INCX, 0
ble LL(99)
beq LL(99)

andi. r0, X, 2 * SIZE - 1
bne LL(100)


+ 1
- 1
kernel/power/cnrm2_ppc440.S View File

@@ -104,7 +104,7 @@
cmpwi cr0, N, 0
ble- LL(999)
cmpwi cr0, INCX, 0
ble- LL(999)
beq- LL(999)

fmr f0, f1
sub X, X, INCX


+ 1
- 1
kernel/power/dnrm2_hummer.S View File

@@ -134,7 +134,7 @@
cmpwi cr0, N, 0
ble LL(99)
cmpwi cr0, INCX, 0
ble LL(99)
beq LL(99)

mr XX, X



+ 1
- 1
kernel/power/dnrm2_ppc440.S View File

@@ -111,7 +111,7 @@
cmpwi cr0, N, 0
ble- LL(999)
cmpwi cr0, INCX, 0
ble- LL(999)
beq- LL(999)

mr NN, N
mr XX, X


+ 1
- 1
kernel/power/nrm2.S View File

@@ -113,7 +113,7 @@
cmpwi cr0, N, 0
ble- LL(9999)
cmpwi cr0, INCX, 0
ble- LL(9999)
beq- LL(9999)

mr NN, N
mr XX, X


+ 1
- 1
kernel/power/snrm2.S View File

@@ -97,7 +97,7 @@
cmpwi cr0, N, 0
ble- LL(9999)
cmpwi cr0, INCX, 0
ble- LL(9999)
beq- LL(9999)

fmr f0, f1
fmr f2, f1


+ 1
- 1
kernel/power/snrm2_hummer.S View File

@@ -119,7 +119,7 @@
cmpwi cr0, N, 0
ble LL(99)
cmpwi cr0, INCX, 0
ble LL(99)
beq LL(99)

cmpwi cr0, INCX, SIZE
bne LL(100)


+ 1
- 1
kernel/power/snrm2_ppc440.S View File

@@ -105,7 +105,7 @@
cmpwi cr0, N, 0
ble- LL(999)
cmpwi cr0, INCX, 0
ble- LL(999)
beq- LL(999)

fmr f0, f1
fmr f2, f1


+ 1
- 1
kernel/power/znrm2.S View File

@@ -105,7 +105,7 @@
cmpwi cr0, N, 0
ble- LL(9999)
cmpwi cr0, INCX, 0
ble- LL(9999)
beq- LL(9999)

mr NN, N
mr XX, X


+ 1
- 1
kernel/power/znrm2_hummer.S View File

@@ -134,7 +134,7 @@
cmpwi cr0, N, 0
ble LL(99)
cmpwi cr0, INCX, 0
ble LL(99)
beq LL(99)

mr XX, X



+ 1
- 1
kernel/power/znrm2_ppc440.S View File

@@ -112,7 +112,7 @@
cmpwi cr0, N, 0
ble- LL(999)
cmpwi cr0, INCX, 0
ble- LL(999)
beq- LL(999)

mr NN, N
mr XX, X


+ 1
- 1
kernel/riscv64/nrm2.c View File

@@ -57,7 +57,7 @@ FLOAT CNAME(BLASLONG n, FLOAT *x, BLASLONG inc_x)
FLOAT absxi = 0.0;


if (n <= 0 || inc_x <= 0) return(0.0);
if (n <= 0 || inc_x == 0) return(0.0);
if ( n == 1 ) return( ABS(x[0]) );

n *= inc_x;


+ 1
- 1
kernel/riscv64/znrm2.c View File

@@ -57,7 +57,7 @@ FLOAT CNAME(BLASLONG n, FLOAT *x, BLASLONG inc_x)
BLASLONG inc_x2;
FLOAT temp;

if (n <= 0 || inc_x <= 0) return(0.0);
if (n <= 0 || inc_x == 0) return(0.0);

inc_x2 = 2 * inc_x;



+ 1
- 1
kernel/sparc/cnrm2.S View File

@@ -76,7 +76,7 @@
FMOV c1, t4

cmp INCX, 0
ble .LL20
beq .LL20
sll INCX, ZBASE_SHIFT, INCX

cmp N, 0


+ 1
- 1
kernel/sparc/dnrm2.S View File

@@ -107,7 +107,7 @@
FMOV fzero, c1

cmp INCX, 0
ble .LL99
beq .LL99
sll INCX, BASE_SHIFT, INCX

add %sp, -8, %sp


+ 1
- 1
kernel/sparc/snrm2.S View File

@@ -76,7 +76,7 @@
FMOV c1, t4

cmp INCX, 0
ble .LL20
beq .LL20
sll INCX, BASE_SHIFT, INCX

cmp N, 0


+ 1
- 1
kernel/sparc/znrm2.S View File

@@ -107,7 +107,7 @@
FMOV fzero, c1

cmp INCX, 0
ble .LL99
beq .LL99
sll INCX, ZBASE_SHIFT, INCX

add %sp, -8, %sp


+ 1
- 1
kernel/x86/nrm2.S View File

@@ -78,7 +78,7 @@
testl M, M
jle .L999
testl INCX, INCX
jle .L999
je .L999

sall $BASE_SHIFT, INCX
fldz


+ 1
- 1
kernel/x86/nrm2_sse.S View File

@@ -69,7 +69,7 @@
jle .L999
pxor %xmm1, %xmm1
testl INCX, INCX
jle .L999
je .L999

leal (, INCX, SIZE), INCX
cmpl $SIZE, INCX


+ 1
- 1
kernel/x86/znrm2.S View File

@@ -78,7 +78,7 @@
testl M, M
jle .L999
testl INCX, INCX
jle .L999
je .L999

sall $ZBASE_SHIFT, INCX
fldz


+ 1
- 1
kernel/x86/znrm2_sse.S View File

@@ -69,7 +69,7 @@
jle .L999
pxor %xmm1, %xmm1
testl INCX, INCX
jle .L999
je .L999

sall $ZBASE_SHIFT, INCX



+ 1
- 1
kernel/x86_64/nrm2.S View File

@@ -58,7 +58,7 @@
testq M, M
jle .L999
testq INCX, INCX
jle .L999
je .L999

salq $BASE_SHIFT, INCX



+ 1
- 1
kernel/x86_64/nrm2_sse.S View File

@@ -57,7 +57,7 @@
jle .L999
pxor %xmm1, %xmm1
testq INCX, INCX
jle .L999
je .L999

pxor %xmm2, %xmm2
leaq (, INCX, SIZE), INCX


+ 1
- 1
kernel/x86_64/znrm2.S View File

@@ -58,7 +58,7 @@
testq M, M
jle .L999
testq INCX, INCX
jle .L999
je .L999

salq $ZBASE_SHIFT, INCX



+ 1
- 1
kernel/x86_64/znrm2_sse.S View File

@@ -58,7 +58,7 @@
jle .L999
pxor %xmm1, %xmm1
testq INCX, INCX
jle .L999
je .L999

xorq FLAG, FLAG



Loading…
Cancel
Save