[Date Index] [Thread Index] [Author Index]

Gigahertz Benchmarks in a world of multicores

• To: mathgroup at smc.vnet.net
• Subject: [mg58856] Gigahertz Benchmarks in a world of multicores
• From: "Hobbs, Sylvia (DPH)" <Sylvia.Hobbs at state.ma.us>
• Date: Thu, 21 Jul 2005 03:08:02 -0400 (EDT)
• Sender: owner-wri-mathgroup at wolfram.com

Hi!

Yes Yasvir, definitely an amusing epiphany for your little laptop. Sounds
like a Tortoise and Hare matrix transposition race. Gigahertz benchmark
races are going to become stranger and stranger (or become obsolescent),
what with the new Pentium D true multicore having two Pentium 4 processor
cores on a single die (or rather multichip). Software needs to progress to
exploit multicores to notice a difference. Also, won't we live in an even
more exciting world when the new Longhorn Operating System has her
Debutante's Ball in Fall 2006?

Sylvia Hobbs

PS: If your cat is always getting near my computer screen, especially when
you are transposing matrices, consider this website for how to clean your
monitor: http://www.legrady.hu/sc.html



-----Original Message-----
From: Yasvir A. Tesiram [mailto:tesiramy at omrf.ouhsc.edu] 
To: mathgroup at smc.vnet.net
Subject: [mg58856] [mg58835] Re: [mg58808] Re: [mg58786] "2p" in BenchmarkReport[] ?

Hi,
This Benchmark test has had me intrigued for some time, especially in 
Linux systems where system libraries change continuously. And I was 
especially amused when my little 2.0GHz Pentium M laptop (Debian 
Linux kernel 2.6.12) outdid a Redhat 9 Pentium4 3.2GHz and the Athlon 64 
3200+ (Suse Linux) when it came to Transposing matrices. I guess thats 
where the buck stopped, because it was thoroughly thrashed on all other 
accounts.

Cheers
Yas


On Tue, 19 Jul 2005, Murray Eisenberg wrote:

> That's plausible -- but a bit surprising to me given benchmark results
> for my machine vs. some of the reference systems:
>
>         2p 2.4 GHz Opteron 250 (Sun Solaris 10):            1.45
> mine -> 3.2 GHz Pentium 4 Extreme Edition (Windows XP Pro): 1.42
>         2p 2.5 GHz G5 Apple Mac OS 10.41):                  1.15
>
> I guess raw processor speed really makes a difference with these
benchmarks.
>
>
> Sseziwa Mukasa wrote:
>> My guess would be dual processor.  With the option for more, eg 4p  for
>> a quad processor.
>>
>> On Jul 18, 2005, at 3:21 AM, Murray Eisenberg wrote:
>>
>>> Mathematica's version 5.2 StandardPackages Utilities`Benchmark`
>>> identifies several of the reference systems as "2p" -- e.g., "2p  1.4
GHz
>>> Itanium 2", "2p 2.5 GHz G5".
>>>
>>> What does the "2p" mean?
>>> --
>>> Murray Eisenberg                     murray at math.umass.edu
>>> Mathematics & Statistics Dept.
>>> Lederle Graduate Research Tower      phone 413 549-1020 (H)
>>> University of Massachusetts                413 545-2859 (W)
>>> 710 North Pleasant Street            fax   413 545-1801
>>> Amherst, MA 01003-9305
>>>
>>>
>>
>>
>
> -- 
> Murray Eisenberg                     murray at math.umass.edu
> Mathematics & Statistics Dept.
> Lederle Graduate Research Tower      phone 413 549-1020 (H)
> University of Massachusetts                413 545-2859 (W)
> 710 North Pleasant Street            fax   413 545-1801
> Amherst, MA 01003-9305
>