[ale] Performance

Jeff Hubbs hbbs at comcast.net
Fri Nov 7 12:20:17 EST 2003



On Fri, 2003-11-07 at 11:56, James P. Kinney III wrote:
> On Fri, 2003-11-07 at 11:37, Christopher Fowler wrote:
> > I've got an AMD 1.2GHZ machine and I'm looking to upgrade it to get
> > better performance.  The main problem I have is with compiling our
> > software.  It is just taking too long and I want to make the process
> > faster.
> > 
> > Disk Performance:
> > 
> > [root at cfowler cfowler]# hdparm -tT /dev/hdb
> > 
> > /dev/hdb:
> >  Timing buffer-cache reads:   128 MB in  1.01 seconds =126.73 MB/sec
> >  Timing buffered disk reads:  64 MB in  1.37 seconds = 46.72 MB/sec
> > 
> > Will getting at AMD 2400+ make a huge difference in compile times?
> > Obviously when compiling software some I/O is invloved on the disks.
> 
> Yes. Compiles are very CPU intensive. Dual CPU systems are very nice for
> heavy compilation projects. I lost a CPU in a dual box. The kernel
> compile time doubled (almost. there is small bit of overhead to use -j 2
> on make) after the death of cpu #2.

Oh, heck, yeah!  Also, if you replace mobo, you're probably going to be
looking at a faster FSB and RAM I/O as well.  If you're looking at
Athlon vs. Duron, clock speed for clock speed, look at the difference in
L1/L2 cache as well - there would probably be a notable compile speed
boost with the bigger caches.  

Some years back, I had read that the make -j parameter should be
adjusted so as to use up as much RAM as you've got without going into
paging, but these days everywhere I look/ask, people are saying the
parameter should be the number of CPUs plus one.

Dunno what your situation is, but if you do this A LOT and you have
other people there who are in the same place also doing it A LOT, you
might want to set up a mega-mutha box as a compile server OR, go the
Mosix route and make a compile farm out of a lot of cheapies or a
handful of biggies.

- Jeff



More information about the Ale mailing list