2001-10-31 23:29:41

by Dave Fuller

[permalink] [raw]
Subject: OSDL Scalable Test Platform test for linux_2_4_13

OSDL is pleased to announce the release of the Scalable
Test Platform tests against kernel release linux_2_4_13.

Summary:

These tests were performed against 2, 4, and 8 CPU
IA-32 (Intel) servers.

Tests run include:

Bonnie++, lmbench, dbench, UNIXbench, and bash-memory.

To view the results, see:

http://www.osdl.org/stp/kernel_suite_2_4_13.html

Contacts:

About the form or content of this message: [email protected]
About the Scalable Test Platform or this release: [email protected]

Regards,

Dave Fuller
Technical Marketing
Open Source Development Labs

http://www.osdlab.org





2001-11-01 03:52:35

by Dmitry Volkoff

[permalink] [raw]
Subject: Re: OSDL Scalable Test Platform test for linux_2_4_13

> These tests were performed against 2, 4, and 8 CPU
> IA-32 (Intel) servers.

UnixBench results are somewhat funny.
4-CPU machine is slower than the one with 2 CPU (238.4 vs 260.0)?
Well, on UP Athlon 1.4Ghz with only 512 Mb RAM I get 460...

--

DV

2001-11-01 19:11:03

by Timothy D. Witham

[permalink] [raw]
Subject: Re: OSDL Scalable Test Platform test for linux_2_4_13

A couple of items explain these differences. The 1st is that
UnixBench is a single stream number and is almost 100% cached even
with a smaller (128KB) cache. Also the benchmark uses a very small
amount of memory. This means that for any given version
of the compiler and OS the best absolute performance number will
be produced by the machine with the fastest single CPU. And
you won't see any improvement from adding additional CPUS.

So what you are seeing is the difference between a single CPU
running at 866MHz on the 2 way and a single CPU running at
700 MHz on the 4 way. And of course your 1.4GHz would give better
numbers as you are looking at 700Mhz, 866MHz and then 1.4 GHz and
so the numbers should be

The idea behind this test setup isn't to compare machine x
verses machine y. (That's what marketing departments do.)
But to be able to compare the same hardware against its self
before and after a kernel change.

wookie

On Wed, 2001-10-31 at 19:51, Dmitry Volkoff wrote:
> > These tests were performed against 2, 4, and 8 CPU
> > IA-32 (Intel) servers.
>
> UnixBench results are somewhat funny.
> 4-CPU machine is slower than the one with 2 CPU (238.4 vs 260.0)?
> Well, on UP Athlon 1.4Ghz with only 512 Mb RAM I get 460...
>
> --
>
> DV
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to [email protected]
> More majordomo info at http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at http://www.tux.org/lkml/
--
Timothy D. Witham - Lab Director - [email protected]
Open Source Development Lab Inc - A non-profit corporation
15275 SW Koll Parkway - Suite H - Beaverton OR, 97006
(503)-626-2455 x11 (office) (503)-702-2871 (cell)
(503)-626-2436 (fax)