GPU Shoot-Out – Part II – Setting New Benches
3DMark06
3DMark06 still remains the number one utility for a system benchmark. The numbers it produces aren’t necessarily indicative of real-world gameplay or any gameplay in general, and for that reason we really dislike using it to compare different systems. However, as long as the rest of the tech world uses it to evaluate gaming performance, we will too. We find it mostly useful for tracking changes in a single system. There are four “mini-games” that it uses for benchmarking graphics, as well as two CPU tests. The scores are “weighed” and added together to give an overall number and there is a further breakdown possible with these mini games that we are now showing to you.
Above is Canyon Flight, a scene from one of the four benchmark “mini games” used to benchmark GPU performance. It will give your PC a real workout even though the default resolution is only 12×10 (as pictured)!
Now we see our top GPUs “stretch their legs” as it were, as we overclock our e8600 to almost 4Ghz – within 10Mhz! We note our HD4870 – which was getting 13023 with Cat 8-8 and e8600@3.33Ghz – is now getting 14548 with half of it coming from the CPU score 2932 to 3654. But in an exact comparison, using Cat 8-9 both times, we see 4870X2 going from 16514 with e8600@3.33Ghz to 19395 with e8600@3.99Ghz!! The CPU score only increased by the same amount as with 4870/512 but now the graphics score also increases with this fast multi-GPU solution. The GTX280 shows a more modest increase – rather in-between the 4870 and the x2 and closer to the X2’s performance leap with the CPU overclock. As with Catalyst 8-8, CrossfireX-3 does not offer much over 4870X2 with Cat 8-9 drivers. Later we will see if it has anything to do with the crossfired 2nd card being in a 4xPCIe slot; and also note that the second 4870 has 512MB VRAM compared to the 4870X2’s 2GB.
Here are the results of our 3DMark06 benchmark comparison:
As with Part I of our testing, the results are pretty self explanatory. We can see the ranking and we note that HD4870X2 scales well but crossfireX-3 barely scales with this combination of drivers/HW in this synthetic benchmark.
With 3-D Mark06 – in Part 1 – when we briefly tried our older Intel CPU, e4300 – even when it was overclocked from its stock clock of 1.8 GHz, all the way to 3.33 GHz – it still fell rather short of e8600 at its stock clock of 3.33Ghz. This is probably because of e8600’s much larger cache and newer optimizations. It is noticeable even with a single HD4870-512MB video card. We can also noted that when you are using a card of the 4870X2 class, it starts to make a significant performance difference. So we decided to do all of our video card testing with e8600 at its stock clock for the last article. Since we are seeing even further significant improvement with our e8600 overclocked to near 4 GHz, we will now begin to compare all video performance with this new overclock.
Interesting. We are starting to include the 3DMark06 mini-games as they specifically compare the video cards performance. We note that it is more significant than the final 3DMark overall score might suggest and we will use them in all of our benching as long as we use this benchmark. Here we also note that 4870-X3 outperforms 4870-X2 in every test! Also, we can see that the 4870X2 – simply by being paired with the faster CPU – almost catches up to the crossfired X-3 configuration. From this, we would have to suggest that if you are using a 4870X2 – or even a GTX280 – as mindboggling as it may appear – the fastest stock Core 2 Duo CPUs barely keep up with them, and if you want total performance, you might consider overclocking it safely, which is beyond the scope of this article.
