This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: Patch to improve scan_loop
- To: John Wehle <john at feith dot com>
- Subject: Re: Patch to improve scan_loop
- From: Toon Moene <toon at moene dot indiv dot nluug dot nl>
- Date: Sun, 20 Aug 2000 17:52:16 +0200
- CC: gcc-patches at gcc dot gnu dot org
- Organization: Moene Computational Physics, Maartensdijk, The Netherlands
- References: <200008182348.TAA04144@jwlab.FEITH.COM>
John Wehle wrote:
> Checking for constant and pure calls (which always return) allows
> more insns to be considered for moving. This patch passes make
> bootstrap and make check on Compaq Tru64 UNIX 4.0f.
Ya betcha - this definitely helps, if I combine it with my (upcoming)
declare-Fortran-intrinsics-const patch:
Timings before (with my patch, but without yours):
HL_ref23215950: FORECAST TOOK 8.0010
SECONDS
HL_ref23215950: FORECAST TOOK 132.3877
SECONDS
HL_ref23215950: FORECAST TOOK 381.9346
SECONDS
HL_ref23215950: FORECAST TOOK 1050.5430
SECONDS
HL_ref23215950: FORECAST TOOK 5.7773
SECONDS
HL_ref23215950: FORECAST TOOK 130.9160
SECONDS
HL_ref23215950: FORECAST TOOK 380.6230
SECONDS
HL_ref23215950: FORECAST TOOK 1057.1553
SECONDS
Timings after (with both your patch and mine):
HL_ref20153249: FORECAST TOOK 7.0547
SECONDS
HL_ref20153249: FORECAST TOOK 131.7363
SECONDS
HL_ref20153249: FORECAST TOOK 338.0977
SECONDS
HL_ref20153249: FORECAST TOOK 931.3125
SECONDS
HL_ref20153249: FORECAST TOOK 4.9443
SECONDS
HL_ref20153249: FORECAST TOOK 132.5449
SECONDS
HL_ref20153249: FORECAST TOOK 348.8555
SECONDS
HL_ref20153249: FORECAST TOOK 928.5674
SECONDS
[ this is all on alphaev6-unknown-linux-gnu ]
HURRAY !
--
Toon Moene - mailto:toon@moene.indiv.nluug.nl - phoneto: +31 346 214290
Saturnushof 14, 3738 XG Maartensdijk, The Netherlands
GNU Fortran 77: http://gcc.gnu.org/onlinedocs/g77_news.html
GNU Fortran 95: http://g95.sourceforge.net/ (under construction)