From: | Thomas Lockhart <lockhart(at)alumni(dot)caltech(dot)edu> |
---|---|
To: | Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>, Lamar Owen <lamar(dot)owen(at)wgcr(dot)org> |
Cc: | Mike Mascari <mascarm(at)mascari(dot)com>, pgsql-hackers(at)postgresql(dot)org |
Subject: | Re: Weird function behavior from Sept 11 snapshot |
Date: | 2000-09-12 16:45:28 |
Message-ID: | 39BE5DA8.C50F9CF8@alumni.caltech.edu |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-hackers |
Hmm. I recompiled the current snapshot with the optimizations from my
Mandrake RPM (using the Mandrake defaults, except for disabling
"fast-math"), and get the following:
7.0.2 current test
1.8 5.3 sum(i)
1.95 1.77 sum(f)
2.3 1.9 sum(cast(i as float8))
My previous tests on the current tree were with -O0, asserts enabled,
and few other options specified (mostly, the defaults for the Postgres
Linux build).
The Linux defaults in the Postgres tarball are:
-O2 -Wall -Wmissing-prototypes -Wmissing-declarations
whereas the defaults for Mandrake (with fast-math turned off since it
gives rounding trouble in date/time math):
-O3 -fomit-frame-pointer -fno-exceptions -fno-rtti -pipe -s
-mpentiumpro -mcpu=pentiumpro -march=pentiumpro
-fexpensive-optimizations
-malign-loops=2 -malign-jumps=2 -malign-functions=2
-mpreferred-stack-boundary=2 -fno-fast-math
I'll do some more tests with the default compiler options. The good news
is that the new fmgr interface is apparently as fast or faster than the
old one :)
- Thomas
From | Date | Subject | |
---|---|---|---|
Next Message | Bruce Momjian - CVS | 2000-09-12 18:56:04 | pgsql/src/interfaces/jdbc/org/postgresql/jdbc1 (DatabaseMetaData.java) |
Previous Message | Mario Weilguni | 2000-09-12 16:39:54 | Re: Patch for TNS services |