best way to measure overclocking...?

pjgtech

pjgtech

pjgtech

Associate
Joined
30 Apr 2007
Posts
40
Location
UK
Hi guys... I have been overclocking many pcs for many years now, (mainly to play games) and I like to think I know the basics fairly well.

However, whenever I use futuremark benchmarks (3Dmark06, etc) I never seem to score that highly compared to other people, my scores always seem to be about average. bearing in mind my systems are usually overclocked, I would expect them to be faster than average.

As an example my current 2 x pc's both score around 10000 using 3DMark06.

1) Abit AB9 Pro:E6600(clocked to .2Ghz):2GbBallistix:8800GTS(320Mb):Sata HD:using Win XP home (SP2)
2) Abit IP35:E6550(clocked to 3.0Ghz):2Gb OCZ:AsusGeForce EN8800GS(384Mb):Sata HD:using Win XP home(SP2)

Generally, I only have my anti virus (McAfee) running in the background when running 3DMark (evrything else is off). Both GFX cards are also slightly overcocked (about 5% increase)

So my question is this.... is 3Dmark not an accurate benchmarking tool to use? and if so, what tools do other people use to measure the performance of their pcs (mainly relevant to games, framerates, etc)

How do people get those higher scores? are there some tweaks, overclocks that I'm not aware of? or am I doing something wrong with my own overclocks?

Just interested in what other people do.... Any info appreciated.... 8-)
 
Replying to my own messages again...!

As I have nVidia GFX cards inboth my pc's, I wondered if Futuremark gives better results for ATI cards than nVidia?

Also, I tend to overclock my pcs until I can use them every day and they are totally stable, then I benchmark them for framerates, etc.

However, I presume the guys who go for really high scores in programs like 3DMark are only interested in speed (even if the pc is highly unstable), so whilst they may get a good score, the pc may not be relaibel at that speed?

Just a few more thoughts..... 8-)
 
I find a overclock of 600mhz on my q6600 from stock to 3ghz gives an extra 2000ish points in 3dmark06 with a 4870 so a cpu overclock definately boosts the score. Have you ran the benchmark at stock to see what the increase is?

SuperPi is often used as a quick benchmark, 1M being the common one.

Some reviews use PCmark and other benchmarks which you could also run to see how your system compares. Also other synthetics like everst are often used to test the effect on memory factors.

Others can probably provide more insight...

ps. in relation to your last post, I personally have seen the ati cards perform very well in 3dmark benchmarks fortunately now that translates to performance in games with the 4870 being very much on par with the new nvidia cards. However, looking at 3dmark results for the 3870x2 showed pretty dominant scores in the synthetic aspects which didn't always translate to superior performance in games. So what I'm trying to say is that people often prefer in-game benchmarks over synthetics which are often not a true reflection of in-game performance.
 
Last edited:
Well, I just updated the driver for my graphics card in my Abit AB9 Pro pc, the GFX card is a nVidia 8800GTS (320Mb) and the 3DMark06 score went from a whisker over 10000 to 10600...!

Gonna try the mb bios next, as I haven't updated that for a while...

8-)
 
People who benchmark tend to do a lot of optimisation. Don't use other peoples scores as a reference of what your pc should score, but what it could score.
 
You must log in or register to reply here.
Back
Top Bottom