This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: Patch for IA64 performance (inline division)
- From: Andreas Schwab <schwab at suse dot de>
- To: sje at cup dot hp dot com
- Cc: gcc-patches at gcc dot gnu dot org
- Date: Sun, 29 Feb 2004 22:26:49 +0100
- Subject: Re: Patch for IA64 performance (inline division)
- References: <200402262303.PAA28340@hpsje.cup.hp.com>
Steve Ellcey <sje@cup.hp.com> writes:
> 2004-02-26 Steve Ellcey <sje@cup.hp.com>
>
> * config/ia64/ia64.h (no-inline-float-divide): New option.
> * config/ia64/ia64.h (no-inline-int-divide): New option.
> * config/ia64/ia64.h (no-inline-sqrt): New option.
> (TARGET_DEFAULT): Add MASK_INLINE_FLOAT_DIV_THR to define.
> * config/ia64/hpux.h (TARGET_DEFAULT): Ditto.
> * config/ia64/ia64.c (ia64_override_options): Modify error
> checking for inlined division/sqrt.
This is causing comparison failures on ia64-linux.
./alias.o differs
./c-typeck.o differs
./combine.o differs
./cpptrad.o differs
./dojump.o differs
./dwarf2out.o differs
./expmed.o differs
./expr.o differs
./flow.o differs
./fold-const.o differs
./function.o differs
./gcc.o differs
./gcov.o differs
./gcse.o differs
./genattr.o differs
./genattrtab.o differs
./gengtype-lex.o differs
./gengtype.o differs
./global.o differs
./insn-extract.o differs
./insn-recog.o differs
./integrate.o differs
./loop-iv.o differs
./loop.o differs
./optabs.o differs
./ra-build.o differs
./ra-colorize.o differs
./ra-rewrite.o differs
./real.o differs
./regmove.o differs
./regrename.o differs
./reload.o differs
./reload1.o differs
cp/mangle.o differs
cp/name-lookup.o differs
f/equiv.o differs
f/expr.o differs
f/info.o differs
f/stb.o differs
java/class.o differs
java/gjavah.o differs
java/jcf-write.o differs
--- /dev/fd/63 2004-02-29 22:25:08.000000000 +0100
+++ /dev/fd/62 2004-02-29 22:25:08.000000000 +0100
@@ -1,5 +1,5 @@
-stage2/alias.o: file format elf64-ia64-little
+alias.o: file format elf64-ia64-little
Disassembly of section .text:
@@ -10536,7 +10536,7 @@
c5f0: 10 00 00 00 01 00 [MIB] nop.m 0x0
c5f6: 70 70 98 0c e8 03 cmp.ltu p7,p6=r14,r38
c5fc: f0 05 00 43 (p07) br.cond.dpnt.few cbe0 <init_alias_analysis+0x760>
- c600: 0c 60 01 02 00 24 [MFI] addl r44=0,r1
+ c600: 0c 58 01 02 00 24 [MFI] addl r43=0,r1
c600: LTOFF22X reload_completed
c606: 00 00 00 02 00 20 nop.f 0x0
c60c: 64 02 48 80 shladd r33=r38,3,r0
@@ -10545,7 +10545,7 @@
c612: GPREL22 .sbss+0x58
c616: 00 00 00 02 00 a0 nop.f 0x0
c61c: 05 08 00 90 addl r45=0,r1;;
- c620: 11 60 01 58 18 10 [MIB] ld8 r44=[r44]
+ c620: 11 58 01 56 18 10 [MIB] ld8 r43=[r43]
c620: LDXMOV reload_completed
c622: PCREL21B xmalloc
c626: a0 03 84 00 42 00 mov r58=r33
@@ -10558,7 +10558,7 @@
c646: 00 00 00 02 00 00 nop.f 0x0
c64c: 08 00 00 50 br.call.sptk.many b0=c640 <init_alias_analysis+0x1c0>;;
c650: 09 00 20 5a 98 11 [MMI] st8 [r45]=r8
- c656: e0 00 b0 20 20 20 ld4 r14=[r44]
+ c656: e0 00 ac 20 20 20 ld4 r14=[r43]
c65c: 00 b8 01 84 mov r1=r55;;
c660: 10 00 00 00 01 00 [MIB] nop.m 0x0
c666: 70 00 38 0c 73 03 cmp4.eq p7,p6=0,r14
@@ -10855,7 +10855,7 @@
cbf6: 10 00 dc 00 42 00 mov r1=r55
cbfc: 10 fa ff 48 br.few c600 <init_alias_analysis+0x180>
cc00: 19 00 00 00 01 00 [MMB] nop.m 0x0
- cc06: e0 00 b0 20 20 00 ld4 r14=[r44]
+ cc06: e0 00 ac 20 20 00 ld4 r14=[r43]
cc0c: 00 00 00 20 nop.b 0x0;;
cc10: 10 00 00 00 01 00 [MIB] nop.m 0x0
cc16: 60 00 38 0e f3 03 cmp4.eq p6,p7=0,r14
@@ -11093,8 +11093,8 @@
d086: 00 00 00 02 00 00 nop.f 0x0
d08c: 08 00 00 50 br.call.sptk.many b0=d080 <init_alias_analysis+0xc00>;;
d090: 08 30 00 10 07 39 [MMI] cmp.eq p6,p7=0,r8
- d096: 10 00 dc 00 42 60 mov r1=r55
- d09c: 05 40 00 84 mov r43=r8
+ d096: 10 00 dc 00 42 80 mov r1=r55
+ d09c: 05 40 00 84 mov r44=r8
d0a0: 18 d8 0d 00 00 24 [MMB] mov r59=3
d0a6: c0 03 00 00 42 03 mov r60=r0
d0ac: d0 04 00 43 (p06) br.cond.dpnt.few d570 <init_alias_analysis+0x10f0>
@@ -11121,12 +11121,12 @@
d116: 01 18 00 80 21 00 (p06) br.cond.dpnt.few d140 <init_alias_analysis+0xcc0>
d11c: 08 00 00 50 br.call.sptk.many b0=d110 <init_alias_analysis+0xc90>;;
d120: 08 00 00 00 01 00 [MMI] nop.m 0x0
- d126: 10 00 dc 00 42 60 mov r1=r55
- d12c: 05 40 00 84 mov r43=r8
+ d126: 10 00 dc 00 42 80 mov r1=r55
+ d12c: 05 40 00 84 mov r44=r8
d130: 19 00 00 00 01 00 [MMB] nop.m 0x0
d136: 70 00 20 0c f2 03 cmp.eq p7,p6=0,r8
d13c: 60 fc ff 4b (p07) br.cond.dpnt.few cd90 <init_alias_analysis+0x910>;;
- d140: 0d 50 21 56 00 21 [MFI] adds r42=8,r43
+ d140: 0d 50 21 58 00 21 [MFI] adds r42=8,r44
d146: 00 00 00 02 00 60 nop.f 0x0
d14c: 17 00 00 90 mov r59=1;;
d150: 0b d0 01 54 18 10 [MMI] ld8 r58=[r42];;
@@ -11152,7 +11152,7 @@
d1b0: 18 70 00 68 18 10 [MMB] ld8 r14=[r52]
d1b6: 00 01 a8 30 20 00 ld8 r16=[r42]
d1bc: 00 00 00 20 nop.b 0x0
- d1c0: 0a 88 08 56 00 21 [MMI] adds r17=2,r43;;
+ d1c0: 0a 88 08 58 00 21 [MMI] adds r17=2,r44;;
d1c6: e0 08 39 24 40 00 shladd r14=r33,3,r14
d1cc: 00 00 04 00 nop.i 0x0
d1d0: 0a 78 00 66 18 10 [MMI] ld8 r15=[r51];;
@@ -11365,7 +11365,7 @@
d586: c0 03 00 00 42 00 mov r60=r0
d58c: 08 00 00 50 br.call.sptk.many b0=d580 <init_alias_analysis+0x1100>;;
d590: 08 00 00 00 01 00 [MMI] nop.m 0x0
- d596: b0 02 20 00 42 20 mov r43=r8
+ d596: c0 02 20 00 42 20 mov r44=r8
d59c: 00 b8 01 84 mov r1=r55
d5a0: 18 00 00 00 01 00 [MMB] nop.m 0x0
d5a6: 70 00 20 0c 72 03 cmp.eq p7,p6=0,r8
Andreas.
--
Andreas Schwab, SuSE Labs, schwab@suse.de
SuSE Linux AG, Maxfeldstraße 5, 90409 Nürnberg, Germany
Key fingerprint = 58CA 54C7 6D53 942B 1756 01D3 44D5 214B 8276 4ED5
"And now for something completely different."