When i try to grep an 1gb file in fedora core 1 it takes almost 10 minutes

but the same same hardware configuration on suse 9 it takes 30 seconds

can any one can suggest some changes in fedora so that grep will be

faster in Fedora too.

How about some other information like Hardware, kernel, filesystem type?

Hi,

The command line is pretty slow in FC1, particularly grep, the tech boffin at work says that it may have something to do with the filesystem setup (whatever that means), but an "outside the square" solution is to upgrade to Fedora Core 1 from Redhat 7.3 (not RH8 or RH9). This works (and is what I have done on my system to overcome the slow command line problem), but I have found that the overall installation is a bit of a mess with dependencies etc.

I am running a PIII 733 with 256Mb RAM, oldish hardware but not that old! Debian 3 is lightning quick on this system...

I hope that someone has been able to fix this issue for FC2.

cheers,

jdw

my suse default installation with reiserfs (filesystem)
Fedora is default installation with ext3 filesystem

hardware for both are identical

the size log file is also identical on both the system

is there any changes can be made on kernel parameter side ????

reiserfs is supposed to be faster then ext3

maybe upgrade to a faster file system (theres reiserfs, xfs, and i never hear anything on jfs (from the specs i found somewhere its better then ext2/3 when it comes to file sizes and such)

Even though reiser IS faster than ext3, that
shouldn't make that much of a difference ...

I'd look at DMA as a possible source. Or
DeadRat checking every single file you're
accessing ...

What does hdparm /dev/hda (or whichever
drive it's installed on) say in either system?


Cheers,
Tink

If my Linux version 2.6.7-rc3 fedora2 setup is representative,
the problem is with grep (or, more likely, the regexp library),
not with the file system. I base this on the observation that
"anchored" greps, like "^#X" are more than 50 times slower
than "unanchored" greps, like "#X" on the same file. On assorted
other systems, the anchored patterns are faster, as one might
expect from a well-implemented Boyer-Moore pattern matcher.

With apologies for following up my own followup
(I am, as the title shows, a newbie), a bit of googling
has revealed that the real culprit is UTF8. My default LANG is
en_US.UTF-8
If I simply change that to
en_US.iso885915
grep suddenly starts running 100 times faster!!!
I still regard this as a flaw in the regexp processing,
but it's certainly an easy "fix". Give it a shot.