How to benchmark CPU and graphics separately?

[redss]

I ran a flash computer speedtest (http://megaswf.com/serve/1087240) which told me the my 1.6 GHz laptop is much FASTER than my 2.53 GHz desktop, which seems backwards to me.

I'm trying to figure out why the laptop scored so much better. Maybe the flash test included graphics card performance so that the desktop's graphics card is inferior to the laptops GPU?

How can I run isolated benchmarks so that each computer's CPU performance and graphics card performance can be compared separately?

BTW, both computers are running Linux (ubuntu 10.04) not windows.

 

[esrever]

what are the 2 processors?

 

[redss]

Ok I was wrong about the 2.53 GHZ, actually it's 2.4 GHz.

In Ubuntu, /proc/cpuinfo tells me about each cpu:

Intel(R) Pentium(R) 4 CPU 2.40GHz

...and...

Intel(R) Celeron(R) M CPU 420 @ 1.60GHz

Do you need info on the graphics cards? I am working on how to get to that info from within Ubuntu...

 

[Kari]

the celeron 420 was launched in -07, P4 2.4 in -02, and they use different microarchitecture. Obviously the newer is better and more efficient so it ends up faster than the older one even when clocked lower...
http://ark.intel.com/products/27438/Intel-Pentium-4-Processor-2_40-GHz-512K-Cache-533-MHz-FSB
http://ark.intel.com/products/29734/Intel-Celeron-Processor-420-512K-Cache-1_60-GHz-800-MHz-FSB


edit
similar thing happened when the core 2 duo launched, even the slowest one (clocked at 1,86GHz) was faster than the previous gen PentiumD clocked over 3GHz

 

[redss]

So two things I noticed from the specs are the newer 1.6 cpu has a faster FSB and also a 64 bit instruction set as opposed to a 32 bit set.

What do you think would be the largest contributing factor to the newer 1.6 cpu beating out the older 2.4 one?

Also, do you think my graphics card impacted that flash test (at http://megaswf.com/serve/1087240 ) ...if so would it be advised for me to buy a newer graphics card for faster flash performance?

 

[esrever]

CPU are more efficient now and can do more in 1 clock cycle. this is due to many things mostly due to better use of cache and better branch predictions. That celeron can probably do 2x as much per clock cycle than the pentium 4 so its really not much of a surprise that it beats the 2.4ghz pentium 4 even with its slower 1.6ghz clock.

 

[Kari]

What do you think would be the largest contributing factor to the newer 1.6 cpu beating out the older 2.4 one?

They dont list the actual microarchitecture used on those specs, it's just some general compatibility stuff... The microarch affects a lot more internall operations that aren't visible on the surface...

 

[redss]

Ok if purchasing a new computer and evaluating the cpu power, how can you compare apples to apples if gigahertz comparisons aren't accurate? Where can you find true benchmark ratings?

 

[esrever]

Some simple benchmarks like the windows experience index or PCmark give a rough estimate of the cpu performance but isn't 100% accurate. If you are looking to buy a now processor, you can check out a review of it and see how it performs.

CPU performance depends a lot on software so know what you run and then decide what kind of performance you want and need.

 

[Kari]

^^yep, you'd better check some performance reviews on sites like Toms hardware, Anandtech, TechPowerUp, HardwareCanucks etc...
And some software/programs prefer more cores and some prefer higher single core performance, whether it's due to higher clocks or more efficient microarch is pretty meaningless, only the end result matters.