[clug-talk] CPU benchmark - Help.

"J. Rafael Sánchez" rsanchez at itres.com
Wed Feb 16 14:31:08 PST 2005


Much appreciated Trever.

R.

Trever Miller wrote:

> J. Rafael Sánchez wrote:
>
>> can I  trust the bogomips value to be accurate.
>
>
> Yes. In that the bogomips will be roughly the same value every time
> you boot or query them from the system.  They are _not_ an accurate
> representation of a processor's capabilities though.
>
>> And another question, is there some tools out there that I can use to
>>  run on the systems to get a more accurate comparison?
>
>
> The best benchmark is to run your actual application, with a real
> dataset that approximates actual intended usage.  If you can't do that,
> then the next best thing would be to find an appropriate benchmark and
> run that on the target hardware.
>
> http://www.fourmilab.ch/fbench/fbench.html
>
> If you need to run tests under windows, maybe SiSoft's Sandra?
> http://www.sisoftware.net/
>
> Note that you can't simply compare 3rd party benchmarks unless
> they are using the exact same server make/model as you will have
> and will be buying.  All sorts of factors such as chipset, bus
> configuration, etc come into play.
>
> If at all possible, burn a cd with your app and sample data set,
> run it on your existing hardware, then take it to the hardware
> vendor and run it on their equipment.  They should be happy
> to let you do this if it means they could make a sale on
> some of that pricey 64bit gear.
>

-- 
.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.
J. Rafael Sánchez
Systems Administrator

Airborne Hyperspectral Remote
Sensing Systems & Solutions
(p) 403.250.9944
(f) 403.250.9916
W: www.itres.com 
.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.





More information about the clug-talk mailing list