Multi-GPU World Tour Part 7: Uncommon Benchmarks 2

Final Thoughts and Conclusions

This has been a fun and interesting learning experience. Not only did we get some quality time with some serious hardware but we got to enjoy some games that we had never seen before. It’s also been great working with the other sites around the globe on this Multi-GPU World Tour and hope our readers have enjoyed it as much as we have.

Now, on to our experiences with the hardware. First and foremost I can’t help but be impressed at just how easy to set up and well mannered the SLI systems are. You just pop in the cards, install the driver, click the balloon to enable SLI and you’re off and running. Until I bought my own Dell 2405FPW I had no desire to delve into Multi-GPU systems. Now that I have the monitor though it’s always the most important thing that I look for in a video card review. Believe me when I say that the tests that I ran on the pages before you were a bit of an eye opener. Those of you that have been following the tour have no doubt seen how well SLI has performed. NVIDIA has been hard at work squashing bugs to deliver the best Multi-GPU experience and it shows in every one of the articles in this series.

While CrossFire is an extremely competitive platform it is still not to the level of SLI in “user friendliness.” Is it getting better? Absolutely! ATI has come a long way with CrossFire. These guys are willing to work with you on finding the reason for the problems and sorting it out in quick order. As a testament to just how far CrossFire has come we now have bridgeless/dongle-less X1900 systems. A year ago CrossFire was best known for being limited to a max resolution of 1600×1200. The X1600 Pro and X1900 GT in CrossFire seem more polished than the high-end X1900XT with master card and that is extremely encouraging. Without getting myself into too much trouble, CrossFire has some big things on the horizon and I believe it won’t be long until CrossFire and SLI will be on equal ground.

The good, the bad, and the ugly:

The first thing that must be to mentioned is that your CrossFire experience is directly related to the version of driver you are using. The other sites did not have the timeline that we had and were only able to test with a BETA Catalyst 6.6 for the X1900GT’s. As noted in some of the other articles we were provided with a BETA 6.7 driver just about the time the first articles went live, and that driver now now a public WHQL driver that can be downloaded on ATI’s website. This driver was a huge improvement in my testing and we felt it was important to show the performance that users will experience once this driver is released.

Major performance issues were observed in ATI cards in CrossFire as they do not render the Tiger Woods 2006 introduction videos correctly. At the same time a knock against NVIDIA is that SLI is not able to complete replays correctly in that there is strange corruption on the top and bottom of the screen. (see the above image)When playing Tiger Woods 2006 playing in CrossFire or SLI ruins the gaming expierence with the constant flaws to the point where one graphics card would be better than two. Even though SLI and CrossFire are both now mature platforms they still have small issues that keep both companies from saying “it just works”.

In terms of testing methodology the one thing I noticed the day this article was to be published is that the mid-range platform for NVIDIA uses the XFX GeForce 7950 GX2 M570 1GB DDR3 XXX video card, which is currently sold out at Newegg, Zipzoomfly, and MWave. This card is overclocked to 570MHz on the core and 1600MHz on the memory, which is a far cry from the factory clocks of 500MHz core and 1200MHz memory. This card also retails for roughly $660, where the standard 7950 GX2 can be picked up for under $550. If one is considering one of these cards for a mid-range system it should be noted that you are limited to these select motherboards due to BIOS requirements to make sure the video card works. We put the GeForce 7950 up against a pair of ATI Radeon X1900 GT’s that we purchased on our own at a local Best Buy store for $330 each, which means they cost $660 for the pair. In terms of pricing the pair of X1900 GT’s cost exactly the same as our single GeForce 7950 by XFX, so even though the test platform doesn’t seem fair the pricing was identical.

So, if we tally up the results we end up with the following:

• Low End: Nvidia Wins 30 – 3
• Mid-Range: Nvidia Wins 27-4
• High-End: ATI Wins 9-8

So that’s our take on the state of the Mult-GPU systems available today. We hope that you’ve enjoyed your stay and invite you to the forums to give us your take on how we’re doing. Please do look for the next installment in this series from Rage3D as they finish up testing with a final look at Uncommon Benchmarks. Thanks to abit USA, AMD, ATI Technologies, BFG Technologies, ECS, NVIDIA, Sapphire, Super Talent and XFX for sending out hardware to complete the testing. Another thanks goes to the gang at ASCARON Entertainment for helping us benchmark their upcoming game title DarkStar One with developer tools! This article could not have been down without massive support from the industry.

• Introduction – PenstarSys – Live!
• GPUs – bit-tech – Live!
• Motherboards – Bjorn3D – Live!
• Common Benchmarks – HardInfo – Live!
• Xfire most played – Neoseeker – Live!
• Uncommon Benchmarks 1 – NVNews – Live!
• Uncommon Benchmarks 2 – Legit Reviews – You’ve just Read It!
• Uncommon Benchmarks 3 – Rage3D – Pending
• The Verdict – Guru3D – Pending

We hope you enjoyed this look at uncommonly benchmarked game titles and look forward to hearing your thoughts in the forums, which are of course free and open to all!