[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[Commit-gnuradio] r9123 - gnuradio/branches/releases/3.1/gnuradio-core/s
|
From: |
jcorgan |
|
Subject: |
[Commit-gnuradio] r9123 - gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter |
|
Date: |
Fri, 1 Aug 2008 09:34:14 -0600 (MDT) |
Author: jcorgan
Date: 2008-08-01 09:34:12 -0600 (Fri, 01 Aug 2008)
New Revision: 9123
Modified:
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/3dnow_float_dotprod_really_simple.S
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/3dnow_float_dotprod_simple.S
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/ccomplex_dotprod_3dnow.S
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/ccomplex_dotprod_3dnow64.S
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/ccomplex_dotprod_3dnowext.S
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/ccomplex_dotprod_3dnowext64.S
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/ccomplex_dotprod_sse.S
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/ccomplex_dotprod_sse64.S
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/complex_dotprod_3dnow.S
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/complex_dotprod_3dnow64.S
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/complex_dotprod_3dnowext.S
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/complex_dotprod_3dnowext64.S
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/complex_dotprod_sse.S
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/complex_dotprod_sse64.S
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/fcomplex_dotprod_3dnow.S
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/fcomplex_dotprod_3dnow64.S
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/fcomplex_dotprod_sse.S
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/fcomplex_dotprod_sse64.S
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/float_dotprod_3dnow.S
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/float_dotprod_3dnow64.S
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/float_dotprod_sse.S
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/float_dotprod_sse64.S
Log:
Applied changesets r8669, r8684, and r8685 on trunk to release branch.
Modified:
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/3dnow_float_dotprod_really_simple.S
===================================================================
---
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/3dnow_float_dotprod_really_simple.S
2008-08-01 15:25:08 UTC (rev 9122)
+++
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/3dnow_float_dotprod_really_simple.S
2008-08-01 15:34:12 UTC (rev 9123)
@@ -72,14 +72,14 @@
shll $1, %ecx # count * 2
.p2align 4
-.loop1:
+.Loop1:
movq (%eax), %mm0
pfmul (%edx), %mm0
pfadd %mm0, %mm4
addl $8, %edx
addl $8, %eax
decl %ecx
- jne .loop1
+ jne .Loop1
# at this point mm4 contains partial sums
Modified:
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/3dnow_float_dotprod_simple.S
===================================================================
---
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/3dnow_float_dotprod_simple.S
2008-08-01 15:25:08 UTC (rev 9122)
+++
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/3dnow_float_dotprod_simple.S
2008-08-01 15:34:12 UTC (rev 9123)
@@ -72,7 +72,7 @@
pxor %mm5, %mm5 # mm5 = 0 0
.p2align 4
-.loop1:
+.Loop1:
movq 0(%eax), %mm0
movq 8(%eax), %mm1
@@ -85,7 +85,7 @@
addl $16, %edx
addl $16, %eax
decl %ecx
- jne .loop1
+ jne .Loop1
# at this point mm4 and mm5 contain partial sums
Modified:
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/ccomplex_dotprod_3dnow.S
===================================================================
---
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/ccomplex_dotprod_3dnow.S
2008-08-01 15:25:08 UTC (rev 9122)
+++
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/ccomplex_dotprod_3dnow.S
2008-08-01 15:34:12 UTC (rev 9123)
@@ -91,7 +91,7 @@
#
.p2align 4
-.loop1:
+.Loop1:
# complex prod: C += A * B, w/ temp Z, mmPN=$80000000
#
@@ -165,7 +165,7 @@
.L1_test:
decl %ecx
- jge .loop1
+ jge .Loop1
# We've handled the bulk of multiplies up to here.
# Let's see if original n_2_ccomplex_blocks was odd.
Modified:
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/ccomplex_dotprod_3dnow64.S
===================================================================
---
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/ccomplex_dotprod_3dnow64.S
2008-08-01 15:25:08 UTC (rev 9122)
+++
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/ccomplex_dotprod_3dnow64.S
2008-08-01 15:34:12 UTC (rev 9123)
@@ -91,7 +91,7 @@
#
.p2align 4
-.loop1:
+.Loop1:
# complex prod: C += A * B, w/ temp Z, mmPN=$80000000
#
@@ -165,7 +165,7 @@
.L1_test:
dec %rax
- jge .loop1
+ jge .Loop1
# We've handled the bulk of multiplies up to here.
# Let's see if original n_2_ccomplex_blocks was odd.
Modified:
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/ccomplex_dotprod_3dnowext.S
===================================================================
---
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/ccomplex_dotprod_3dnowext.S
2008-08-01 15:25:08 UTC (rev 9122)
+++
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/ccomplex_dotprod_3dnowext.S
2008-08-01 15:34:12 UTC (rev 9123)
@@ -89,7 +89,7 @@
#
.p2align 4
-.loop1:
+.Loop1:
# complex prod: C += A * B, w/ temp Z
#
@@ -144,7 +144,7 @@
.L1_test:
decl %ecx
- jge .loop1
+ jge .Loop1
# We've handled the bulk of multiplies up to here.
# Let's see if original n_2_ccomplex_blocks was odd.
Modified:
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/ccomplex_dotprod_3dnowext64.S
===================================================================
---
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/ccomplex_dotprod_3dnowext64.S
2008-08-01 15:25:08 UTC (rev 9122)
+++
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/ccomplex_dotprod_3dnowext64.S
2008-08-01 15:34:12 UTC (rev 9123)
@@ -89,7 +89,7 @@
#
.p2align 4
-.loop1:
+.Loop1:
# complex prod: C += A * B, w/ temp Z
#
@@ -144,7 +144,7 @@
.L1_test:
dec %rax
- jge .loop1
+ jge .Loop1
# We've handled the bulk of multiplies up to here.
# Let's see if original n_2_ccomplex_blocks was odd.
Modified:
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/ccomplex_dotprod_sse.S
===================================================================
---
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/ccomplex_dotprod_sse.S
2008-08-01 15:25:08 UTC (rev 9122)
+++
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/ccomplex_dotprod_sse.S
2008-08-01 15:34:12 UTC (rev 9123)
@@ -85,7 +85,7 @@
#
.p2align 4
-.loop1:
+.Loop1:
# complex prod: C += A * B, w/ temp Z & Y (or B), xmmPN=$0x8000000080000000
#
@@ -141,7 +141,7 @@
.L1_test:
decl %ecx
- jge .loop1
+ jge .Loop1
# We've handled the bulk of multiplies up to here.
# Let's sse if original n_2_ccomplex_blocks was odd.
Modified:
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/ccomplex_dotprod_sse64.S
===================================================================
---
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/ccomplex_dotprod_sse64.S
2008-08-01 15:25:08 UTC (rev 9122)
+++
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/ccomplex_dotprod_sse64.S
2008-08-01 15:34:12 UTC (rev 9123)
@@ -85,7 +85,7 @@
#
.p2align 4
-.loop1:
+.Loop1:
# complex prod: C += A * B, w/ temp Z & Y (or B), xmmPN=$0x8000000080000000
#
@@ -141,7 +141,7 @@
.L1_test:
dec %rax
- jge .loop1
+ jge .Loop1
# We've handled the bulk of multiplies up to here.
# Let's sse if original n_2_ccomplex_blocks was odd.
Modified:
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/complex_dotprod_3dnow.S
===================================================================
---
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/complex_dotprod_3dnow.S
2008-08-01 15:25:08 UTC (rev 9122)
+++
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/complex_dotprod_3dnow.S
2008-08-01 15:34:12 UTC (rev 9123)
@@ -100,7 +100,7 @@
#
.p2align 4
-.loop1:
+.Loop1:
pfmul 0(%edx), %mm0
pfadd %mm2, %mm6
@@ -145,7 +145,7 @@
.L1_test:
decl %ecx
- jge .loop1
+ jge .Loop1
# We've handled the bulk of multiplies up to here.
# Now accumulate the final two additions and see if original
Modified:
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/complex_dotprod_3dnow64.S
===================================================================
---
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/complex_dotprod_3dnow64.S
2008-08-01 15:25:08 UTC (rev 9122)
+++
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/complex_dotprod_3dnow64.S
2008-08-01 15:34:12 UTC (rev 9123)
@@ -100,7 +100,7 @@
#
.p2align 4
-.loop1:
+.Loop1:
pfmul 0(%rsi), %mm0
pfadd %mm2, %mm6
@@ -145,7 +145,7 @@
.L1_test:
dec %rax
- jge .loop1
+ jge .Loop1
# We've handled the bulk of multiplies up to here.
# Now accumulate the final two additions and see if original
Modified:
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/complex_dotprod_3dnowext.S
===================================================================
---
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/complex_dotprod_3dnowext.S
2008-08-01 15:25:08 UTC (rev 9122)
+++
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/complex_dotprod_3dnowext.S
2008-08-01 15:34:12 UTC (rev 9123)
@@ -93,7 +93,7 @@
#
.p2align 4
-.loop1:
+.Loop1:
pfmul 0(%edx), %mm0
pfadd %mm2, %mm6
@@ -125,7 +125,7 @@
.L1_test:
decl %ecx
- jge .loop1
+ jge .Loop1
# We've handled the bulk of multiplies up to here.
# Now accumulate the final two additions and see if original
Modified:
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/complex_dotprod_3dnowext64.S
===================================================================
---
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/complex_dotprod_3dnowext64.S
2008-08-01 15:25:08 UTC (rev 9122)
+++
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/complex_dotprod_3dnowext64.S
2008-08-01 15:34:12 UTC (rev 9123)
@@ -93,7 +93,7 @@
#
.p2align 4
-.loop1:
+.Loop1:
pfmul 0(%rsi), %mm0
pfadd %mm2, %mm6
@@ -125,7 +125,7 @@
.L1_test:
dec %rax
- jge .loop1
+ jge .Loop1
# We've handled the bulk of multiplies up to here.
# Now accumulate the final two additions and see if original
Modified:
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/complex_dotprod_sse.S
===================================================================
---
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/complex_dotprod_sse.S
2008-08-01 15:25:08 UTC (rev 9122)
+++
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/complex_dotprod_sse.S
2008-08-01 15:34:12 UTC (rev 9123)
@@ -81,7 +81,7 @@
jmp .L1_test
.p2align 4
-.loop1:
+.Loop1:
pxor %mm0, %mm0
punpcklwd 0(%eax), %mm0
@@ -95,7 +95,7 @@
addps %xmm0, %xmm4
.L1_test:
decl %ecx
- jge .loop1
+ jge .Loop1
# set up for primary loop which is unrolled 4 times
@@ -105,7 +105,7 @@
movaps %xmm5, %xmm7
shrl $2, %ecx # n_2_complex_blocks / 4
- je .cleanup # if zero, take short path
+ je .Lcleanup # if zero, take short path
# finish setup and loop priming
@@ -129,7 +129,7 @@
# hence enter loop at top
.p2align 4
-.loop2:
+.Loop2:
mulps (%edx), %xmm0
addps %xmm2, %xmm6
@@ -169,7 +169,7 @@
addl $0x40, %edx
addl $0x10, %eax
decl %ecx
- jne .loop2
+ jne .Loop2
# OK, now we've done with all the multiplies, but
# we still need to handle the unaccumulated
@@ -188,7 +188,7 @@
# At this point, xmm4 contains 2x2 partial sums. We need
# to compute a "horizontal complex add" across xmm4.
-.cleanup: # xmm4 = r1 i2 r3 i4
+.Lcleanup: # xmm4 = r1 i2 r3 i4
movl 20(%ebp), %eax # @result
movhlps %xmm4, %xmm0 # xmm0 = ?? ?? r1 r2
addps %xmm4, %xmm0 # xmm0 = ?? ?? r1+r3 i2+i4
Modified:
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/complex_dotprod_sse64.S
===================================================================
---
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/complex_dotprod_sse64.S
2008-08-01 15:25:08 UTC (rev 9122)
+++
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/complex_dotprod_sse64.S
2008-08-01 15:34:12 UTC (rev 9123)
@@ -80,7 +80,7 @@
jmp .L1_test
.p2align 4
-.loop1:
+.Loop1:
pxor %mm0, %mm0
punpcklwd 0(%rdi), %mm0
@@ -94,7 +94,7 @@
addps %xmm0, %xmm4
.L1_test:
dec %rax
- jge .loop1
+ jge .Loop1
# set up for primary loop which is unrolled 4 times
@@ -103,7 +103,7 @@
shr $2, %rdx # n_2_complex_blocks / 4
movaps %xmm5, %xmm7
- je .cleanup # if zero, take short path
+ je .Lcleanup # if zero, take short path
# finish setup and loop priming
@@ -127,7 +127,7 @@
# hence enter loop at top
.p2align 4
-.loop2:
+.Loop2:
mulps (%rsi), %xmm0
addps %xmm2, %xmm6
@@ -167,7 +167,7 @@
add $0x40, %rsi
add $0x10, %rdi
dec %rdx
- jne .loop2
+ jne .Loop2
# OK, now we've done with all the multiplies, but
# we still need to handle the unaccumulated
@@ -186,7 +186,7 @@
# At this point, xmm4 contains 2x2 partial sums. We need
# to compute a "horizontal complex add" across xmm4.
-.cleanup: # xmm4 = r1 i2 r3 i4
+.Lcleanup: # xmm4 = r1 i2 r3 i4
movhlps %xmm4, %xmm0 # xmm0 = ?? ?? r1 r2
addps %xmm4, %xmm0 # xmm0 = ?? ?? r1+r3 i2+i4
movlps %xmm0, (%rcx) # store low 2x32 bits (complex) to
memory
Modified:
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/fcomplex_dotprod_3dnow.S
===================================================================
---
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/fcomplex_dotprod_3dnow.S
2008-08-01 15:25:08 UTC (rev 9122)
+++
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/fcomplex_dotprod_3dnow.S
2008-08-01 15:34:12 UTC (rev 9123)
@@ -92,7 +92,7 @@
#
.p2align 4
-.loop1:
+.Loop1:
pfmul 0(%edx), %mm0
pfadd %mm2, %mm6
@@ -129,7 +129,7 @@
.L1_test:
decl %ecx
- jge .loop1
+ jge .Loop1
# We've handled the bulk of multiplies up to here.
# Now accumulate the final two additions and see if original
Modified:
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/fcomplex_dotprod_3dnow64.S
===================================================================
---
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/fcomplex_dotprod_3dnow64.S
2008-08-01 15:25:08 UTC (rev 9122)
+++
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/fcomplex_dotprod_3dnow64.S
2008-08-01 15:34:12 UTC (rev 9123)
@@ -91,7 +91,7 @@
#
.p2align 4
-.loop1:
+.Loop1:
pfmul 0(%rsi), %mm0
pfadd %mm2, %mm6
@@ -128,7 +128,7 @@
.L1_test:
dec %rax
- jge .loop1
+ jge .Loop1
# We've handled the bulk of multiplies up to here.
# Now accumulate the final two additions and see if original
Modified:
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/fcomplex_dotprod_sse.S
===================================================================
---
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/fcomplex_dotprod_sse.S
2008-08-01 15:25:08 UTC (rev 9122)
+++
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/fcomplex_dotprod_sse.S
2008-08-01 15:34:12 UTC (rev 9123)
@@ -81,7 +81,7 @@
jmp .L1_test
.p2align 4
-.loop1:
+.Loop1:
movlps 0(%eax), %xmm0
shufps $0x50, %xmm0, %xmm0 # b01010000
@@ -92,7 +92,7 @@
addps %xmm0, %xmm4
.L1_test:
decl %ecx
- jge .loop1
+ jge .Loop1
# set up for primary loop which is unrolled 4 times
@@ -102,7 +102,7 @@
movaps %xmm5, %xmm7
shrl $2, %ecx # n_2_complex_blocks / 4
- je .cleanup # if zero, take short path
+ je .Lcleanup # if zero, take short path
# finish setup and loop priming
@@ -120,7 +120,7 @@
# hence enter loop at top
.p2align 4
-.loop2:
+.Loop2:
addps %xmm2, %xmm6
movlps 0x10(%eax), %xmm2
@@ -152,7 +152,7 @@
addl $0x40, %edx
addl $0x20, %eax
decl %ecx
- jne .loop2
+ jne .Loop2
# OK, now we've done with all the multiplies, but
# we still need to handle the unaccumulated
@@ -171,7 +171,7 @@
# At this point, xmm4 contains 2x2 partial sums. We need
# to compute a "horizontal complex add" across xmm4.
-.cleanup: # xmm4 = r1 i2 r3 i4
+.Lcleanup: # xmm4 = r1 i2 r3 i4
movl 20(%ebp), %eax # @result
movhlps %xmm4, %xmm0 # xmm0 = ?? ?? r1 r2
addps %xmm4, %xmm0 # xmm0 = ?? ?? r1+r3 i2+i4
Modified:
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/fcomplex_dotprod_sse64.S
===================================================================
---
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/fcomplex_dotprod_sse64.S
2008-08-01 15:25:08 UTC (rev 9122)
+++
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/fcomplex_dotprod_sse64.S
2008-08-01 15:34:12 UTC (rev 9123)
@@ -79,7 +79,7 @@
jmp .L1_test
.p2align 4
-.loop1:
+.Loop1:
movlps 0(%rdi), %xmm0
shufps $0x50, %xmm0, %xmm0 # b01010000
@@ -90,7 +90,7 @@
addps %xmm0, %xmm4
.L1_test:
dec %rax
- jge .loop1
+ jge .Loop1
# set up for primary loop which is unrolled 4 times
@@ -99,7 +99,7 @@
movaps %xmm5, %xmm7
shr $2, %rdx # n_2_complex_blocks / 4
- je .cleanup # if zero, take short path
+ je .Lcleanup # if zero, take short path
# finish setup and loop priming
@@ -117,7 +117,7 @@
# hence enter loop at top
.p2align 4
-.loop2:
+.Loop2:
addps %xmm2, %xmm6
movlps 0x10(%rdi), %xmm2
@@ -149,7 +149,7 @@
add $0x40, %rsi
add $0x20, %rdi
dec %rdx
- jne .loop2
+ jne .Loop2
# OK, now we've done with all the multiplies, but
# we still need to handle the unaccumulated
@@ -168,7 +168,7 @@
# At this point, xmm4 contains 2x2 partial sums. We need
# to compute a "horizontal complex add" across xmm4.
-.cleanup: # xmm4 = r1 i2 r3 i4
+.Lcleanup: # xmm4 = r1 i2 r3 i4
movhlps %xmm4, %xmm0 # xmm0 = ?? ?? r1 r2
addps %xmm4, %xmm0 # xmm0 = ?? ?? r1+r3 i2+i4
movlps %xmm0, (%rcx) # store low 2x32 bits (complex) to
memory
Modified:
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/float_dotprod_3dnow.S
===================================================================
---
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/float_dotprod_3dnow.S
2008-08-01 15:25:08 UTC (rev 9122)
+++
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/float_dotprod_3dnow.S
2008-08-01 15:34:12 UTC (rev 9123)
@@ -86,7 +86,7 @@
#
.p2align 4
-.loop1:
+.Loop1:
pfmul 0(%edx), %mm0
pfadd %mm2, %mm6
movq 16(%eax), %mm2
@@ -107,7 +107,7 @@
addl $32, %eax
.L1_test:
decl %ecx
- jge .loop1
+ jge .Loop1
# We've handled the bulk of multiplies up to here.
# Now accumulate the final two additions and see if original
Modified:
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/float_dotprod_3dnow64.S
===================================================================
---
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/float_dotprod_3dnow64.S
2008-08-01 15:25:08 UTC (rev 9122)
+++
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/float_dotprod_3dnow64.S
2008-08-01 15:34:12 UTC (rev 9123)
@@ -85,7 +85,7 @@
#
.p2align 4
-.loop1:
+.Loop1:
pfmul 0(%rdi), %mm0
pfadd %mm2, %mm6
movq 16(%rsi), %mm2
@@ -106,7 +106,7 @@
add $32, %rsi
.L1_test:
dec %rax
- jge .loop1
+ jge .Loop1
# We've handled the bulk of multiplies up to here.
# Now accumulate the final two additions and see if original
Modified:
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/float_dotprod_sse.S
===================================================================
---
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/float_dotprod_sse.S
2008-08-01 15:25:08 UTC (rev 9122)
+++
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/float_dotprod_sse.S
2008-08-01 15:34:12 UTC (rev 9123)
@@ -79,7 +79,7 @@
jmp .L1_test
.p2align 4
-.loop1:
+.Loop1:
movaps (%eax), %xmm0
mulps (%edx), %xmm0
addl $0x10, %edx
@@ -87,7 +87,7 @@
addps %xmm0, %xmm4
.L1_test:
decl %ecx
- jge .loop1
+ jge .Loop1
# set up for primary loop which is unrolled 4 times
@@ -97,7 +97,7 @@
movaps %xmm5, %xmm7
shrl $2, %ecx # n_4_float_blocks / 4
- je .cleanup # if zero, take short path
+ je .Lcleanup # if zero, take short path
# finish setup and loop priming
@@ -110,7 +110,7 @@
# hence enter loop at top
.p2align 4
-.loop2:
+.Loop2:
mulps (%edx), %xmm0
addps %xmm2, %xmm6
movaps 0x20(%eax), %xmm2
@@ -130,7 +130,7 @@
addl $0x40, %edx
addl $0x40, %eax
decl %ecx
- jne .loop2
+ jne .Loop2
# OK, now we've done with all the multiplies, but
# we still need to handle the unaccumulated
@@ -150,7 +150,7 @@
# to compute a "horizontal add" across xmm4.
# This is a fairly nasty operation...
-.cleanup: # xmm4 = d1 d2 d3 d4
+.Lcleanup: # xmm4 = d1 d2 d3 d4
xorps %xmm0, %xmm0 # xmm0 = 0 0 0 0 (may be unnecessary)
movhlps %xmm4, %xmm0 # xmm0 = 0 0 d1 d2
addps %xmm4, %xmm0 # xmm0 = d1 d2 d1+d3 d2+d4
Modified:
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/float_dotprod_sse64.S
===================================================================
---
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/float_dotprod_sse64.S
2008-08-01 15:25:08 UTC (rev 9122)
+++
gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter/float_dotprod_sse64.S
2008-08-01 15:34:12 UTC (rev 9123)
@@ -77,7 +77,7 @@
jmp .L1_test
.p2align 4
-.loop1:
+.Loop1:
movaps (%rsi), %xmm0
mulps (%rdi), %xmm0
add $0x10, %rdi
@@ -85,7 +85,7 @@
addps %xmm0, %xmm4
.L1_test:
dec %rax
- jge .loop1
+ jge .Loop1
# set up for primary loop which is unrolled 4 times
@@ -94,7 +94,7 @@
movaps %xmm5, %xmm7
shr $2, %rdx # n_4_float_blocks / 4
- je .cleanup # if zero, take short path
+ je .Lcleanup # if zero, take short path
# finish setup and loop priming
@@ -107,7 +107,7 @@
# hence enter loop at top
.p2align 4
-.loop2:
+.Loop2:
mulps (%rdi), %xmm0
addps %xmm2, %xmm6
movaps 0x20(%rsi), %xmm2
@@ -127,7 +127,7 @@
add $0x40, %rdi
add $0x40, %rsi
dec %rdx
- jne .loop2
+ jne .Loop2
# OK, now we've done with all the multiplies, but
# we still need to handle the unaccumulated
@@ -147,7 +147,7 @@
# to compute a "horizontal add" across xmm4.
# This is a fairly nasty operation...
-.cleanup: # xmm4 = d1 d2 d3 d4
+.Lcleanup: # xmm4 = d1 d2 d3 d4
xorps %xmm0, %xmm0 # xmm0 = 0 0 0 0 (may be unnecessary)
movhlps %xmm4, %xmm0 # xmm0 = 0 0 d1 d2
addps %xmm4, %xmm0 # xmm0 = d1 d2 d1+d3 d2+d4
| [Prev in Thread] |
Current Thread |
[Next in Thread] |
- [Commit-gnuradio] r9123 - gnuradio/branches/releases/3.1/gnuradio-core/src/lib/filter,
jcorgan <=