PerformanceTest Version 7.0 Beta Release

[{SAS}TB]

Thanks for the reply and feedback.

I fully understand the dual core GPU issue as its not uncommon, but in relation to the windowed 3D GPU test I have run the test and tried various monitoring software and (im assumed due to it being windowed) the GPU stays at 2D clock speeds, therefore not actually testing the 3D GFX?

Is the simplest way not to make all the 2D tests full screen ?

Thanks again



p.s. out of interest what are the component weightings for brta 7 ?

Thanks

[Michael (Passmark)]

We've made a change to the score calculation and put up a new beta. The new formula will severely limit the effect of a few sub scores being in the extremely high ranges. The formula is basically like this.

each sub score is used as a divisor to its weighting

weighting (eg. 0.29(29%) for CPU) / cpu sub score

These are than added together and inverted.

1 / addition of all sub-score formula outputs.

There's a few other things we do in order to keep the scores roughly equivalent to the old V6 ones so you can't apply this formula directly to your sub-scores and get the actual Passmark rating but this is the gist of it.

[{SAS}TB]

Cool, downloading now to give it a try

[{SAS}TB]

OK, i've given it a whirl "out the box" and got:



So a drastic reduction in the overall score. I have also tried (as a comparison) with and without the RAM disc and Image mount:

NO RAM DISC (USING EVERYDAY HDD) AND DISC IN THE DVD PLAYER:



1897.0

Exactly the same setup, ran straight after the above, but with

RAM DISC AND ISO IMAGE MOUNTED FROM THE RAM DISC



2716.7

So still a good increase using these "tweaks", but not the huge increases seen on the previous versions

[{SAS}TB]

Best score to date (only done three properly) is

[David (PassMark)]

With the previous V6 algorithm having a infinitely fast CD, or hard disk would result in a infinitely large PassMark Rating.

But of course in the real world even if your CD or HD drive was infinitely fast, you would not get an infinite amount of work done on the computer. Other components would become a bottleneck instead.

With the new algorithm you can only get a extremely high Passmark rating if ALL components are extremely fast. The corollary of this is however means that having very slow component will drag down the overall score much more than before.

And this also aligns better with real life. In that if you had a infinitely slow CD or HD, then the amount of work you would get done on the PC would effectively be 0. And the PassMark rating will also now approach zero as ANY component comes close to being non functional.

Regarding the comments on the Windows 3D test. Windowed 3D is common in applications like CAD/CAM, 3D design and some games. If the hardware manufacturers choose to provide less acceleration for windowed applications, then this is something interesting and worth measuring. I don't think we should avoid doing a test, just because some manufactuers get poor results in that test.

Your new results look good and demonstrate what we were trying to achieve. Switching to a DRAM disk will increase your overall system performance by 20% - 30% and not the massive ~300% increases reported previously. So thanks for posting your results.

[{SAS}TB]

Thanks for the reply - glad to be of assistance.

I would, however, like to point out with regards to the 3D tests that the clock staying at 2D speed is with a variety of reference GFX board and manufacturer derivatives of both ATI and nVidia cards.