From: ujlh@pool.info.sunyit.edu (James Henrickson) Subject: Re: Benchmarking under Linux (was Re: New 486 Suggestions? Date: 2 Sep 1992 17:54:17 GMT
In article <1992Aug31.210041.21832@novell.com> bboerner@novell.com (Brendan B. Boerner) writes:
>>In article <pdhatchm.715057547@syzygy> of comp.os.linux,
>> In my experience with Linux a faster CPU is a BIG advantage.
>>
>> Someone preveously mentioned it taking a 486-33 12 minutes to do a
>> "make clean ; make dep ; make".
>>
>> My 486-50 does this in just over 4 mins.
>
>This has me wondering: does anyone have any benchmarking or performance
>metric programs for Linux? I've got a 386/16Mhz w/a tad under 5MB of
>RAM and it takes me somewhere between 3-4hours to build the kernel. On
>my girlfriend's machine, a 486/33 w/8MB RAM, it takes under 10
>minutes. I'd like to get an idea of where I could improve
>performance.
>
>Thanks,
>Brendan
>--
>Brendan B. Boerner Phone: 512/346-8380
>Internet: bboerner@novell.com MHS: bboerner@novell
>Please use ^^^^^^^^^^^^^^^^^ if replying by mail.
This is an obvious answer, but...
Kernel compilation times depend on several things: disk speed, amount of
RAM, and processor speed. For starters, you are probably swapping with
5 MB of RAM, and a slow disk will compound the problem. A slow processor
doesn't help, either. I haven't tried a kernel rebuild on my 386SX-25 yet,
but I shouldn't have many complaints. It has 8 MB of RAM and the disk has
a 16 or 18 ms (don't remember) access time. I am also using a swap partition,
which is faster than a swap file. Oh, and another thing: the drive is
an IDE. I have read (but can't confirm) that the current SCSI drivers
are slower.
On my 386DX-33, I noticed a significant speed-up when I increased the RAM
from 4 MB to 8 MB. I believe the buffers help, too, since reloading gcc
is a lot snappier. :-)
-- Jim H. * * James L. Henrickson * ujlh@sunyit.edu "Some day I might have a real .signature!"