FPSLabs : CrossFireX Scaling Take 2

We got our hands on the second revision of AMD’s first CrossFireX enabling drivers, and its time to revisit our scaling results- Now with 100% more Crysis.

As lowly hardware reviewers, we really don’t know too much about what goes into revising video card drivers. One time a very smart and eccentric software engineer from AMD tried his best to get us to understand what changing certain variables from one version to the next actually ended up doing in the long run, but that conversation was so ridiculously far over our heads that we prefer to forget it ever happened then to try to make sense of it. Obviously the folks at AMD know what they are doing when it comes to driver revisions, as over the past three years now they have been absolutely on the ball with creating performance gains out of nothing but simple software update.

When we were graced with the chance of getting an exclusive (and completely un-sanctioned) behind the scenes look at various revisions of what could very well be the most significant graphics driver release in AMD’s history, we were pretty excited. Like kids in a candy store, we fired up the new driver on a system with borrowed hardware and produced some results that looked pretty good for AMD. However, the driver we used was the very first to enable CrossFireX, and we made absolutely sure that readers understood the exclusivity and probable inaccuracies inherent with the reported results. Now, a few weeks after that initial review, we have managed to get our hands on the second revision of that evaluation driver, and have run a couple of tests on another system with borrowed hardware. Luckily specs are pretty similar, so drawing conclusions between the previous data set and the one you are about to see isn’t necessarily out of the question.

[ArticleAdvert]

Unfortunately this second revision of the driver, if anything, seems more raw than the first, and consequently had dropped support for various rudimentary features, like the resolutions we wanted to run. We were unable to run our tests at the monitor’s native resolution of 1680×1050, because the eval driver would not let us run Windows at that resolution, no matter how many times we forced it to do so. This problem appears to be specific to our system only, as we have seen a couple of unofficial reports on forums stating that 1680×1050 resolution is supported in the same configuration. We did, however, manage to run a couple of tests at 1600×1200, which unfortunately was the maximum resolution we had the resources to test.

Again we must stress that the numbers you will see in this report should not be compared directly to any other numbers you can find in any other article. It is possible that the scaling performance figures can be compared to others on the net, but system configurations must be taken into heavy consideration due to potential component bottlenecks (CPU limiting, for instance).

Lastly, anyone unfamiliar with the standalone performance of the Radeon HD 3870 X2 should check out the full review we published last month before proceeding. The results of our first CrossFireX testing that took place earlier this month are also recommended prerequisite reading.
{mospagebreak heading=Introduction&title=Test Setup and Methods}
Test Setup and Methods

Please note that this hardware setup is very much different from what we usually test video cards with. The reason for this is that we had to borrow a CrossFire capable system, as our normal testbed utilizes NVIDIA’s nForce 680i SLI chipset. Pretty much all components are dissimilar between the two systems, so under no circumstances should conclusions be drawn relating scores obtained in articles using different systems. It is only appropriate to draw conclusions from each set of CrossFireX or non-CrossFireX data independently.

Hardware Setup

[ArticleAdvert]

Case: Cooler Master Stacker 830
Power Supply: Cooler Master Real Power Pro 1000
Motherboard: MSI X48 Platinum
Processor: Intel Core 2 Extreme QX9770 @ 4.0GHz (400×10)
Hard Drive: 2x Western Digital Raptor 150 10,000RPM
Video: 2x MSI Radeon HD 3870 X2 (R3870X2-T2D1G-OC 850/1800)
Memory: 2048MB (2×1024MB) Kingston HyperX PC3 14400
Optical Drive: ASUS DRW-2014L1
Cooling: Cooler Master Hyper 212

Software Setup

Operating System: Microsoft Windows Vista Home Premium 32-bit
Video Driver: AMD Quad Vista Evaluation Software (Feb 12 release)

Our methodology for testing video cards typically requires time quantified in weeks. For the testing of this card, we had about one day. For this reason, we were only able to run four of our tests in accordance with our testing methodology. Unfortunately we do not have the hardware in our labs for supplementary testing, which to this point we have done as visitors at an ex-situ location. We half expect that this driver revision will be the last before CrossFireX becomes officially supported in Catalyst 8.3, which should be available sometime next month. However, if a new driver becomes available in the next couple of days, we will make every effort to run more tests to reveal CrossFireX scaling performance.

All tests were conducted at least three times, with the average score being that reported in the graphs. For Futuremark’s 3DMark06, our only synthetic test, settings were left on default out-of-the-box values. In our game tests, built-in benchmarks were used for F.E.A.R., Call of Juarez, and Crysis. The benchmarks presented in this review are Vista exclusive, as we were unable to secure XP drivers in the short time we had with the card (they exist, but we had one day to test… what do you want from us, people!?!?).

Vsync was turned off for all tests. If there were any special kind of graphical quality settings used in any game, they will be noted in the section for that game or a screenshot of the settings will be provided.
{mospagebreak title=3DMark06}
Tests

3DMark06

Since Futuremark’s 3DMark Vantage is not yet out (though the first screenshots have recently been made available), 3DMark06 remains the go to utility for a quick benchmark. The numbers it produces aren’t necessarily indicative of real-world gameplay or gameplay in general, and for that reason we really dislike using it. However, as long as the rest of the world uses it to evaluate hardware performance, we will too.

{multithumb popup_type=none enable_thumbs=0} {multithumb popup_type=default enable_thumbs=default}

[ArticleAdvert]

As was the case in our previous article, it is fairly evident here that scaling in 3DMark06 is pretty abysmal. Luckily for AMD, 3DMark06 is just a stupid benchmark that rarely results in impressive scaling numbers. We actually got a slightly less impressive 14% scaling in 3DMark06 when enabling the second Radeon HD 3870 X2, a decrease we can easily attribute to the slightly decreased processor speed, as this particular test is heavily dependent on CPU performance. Please note that the above screenshots are from our previous testing, as they relay the settings used for the test. The numbers and specs are obviously different this time around, but we didn’t feel a new screen capture was different given the relatively similar results.
{mospagebreak title=Call of Juarez}
Call of Juarez

Call of Juarez was another game we had on hand and could be benchmarked fairly quickly, which was our primary concern while testing these cards. Whereas many games offer pretty poor built-in benchmarks, the Call of Juarez DirectX 10 test incorporates a lot of different gameplay elements that make it a pretty good indication of in-game performance. The game is also ideal for testing high-end graphics cards, as in DirectX 10 in looks quite pretty and brings even the best graphics subsystems to their knees at higher resolutions. Call of Juarez certainly isn’t the best-looking game around anymore thanks to Unreal Engine 3, updates to the Source Engine, and Crysis, but it will remain in our benchmarking suite until it proves to be useless.

{multithumb popup_type=none enable_thumbs=0} {multithumb popup_type=default enable_thumbs=default}

[ArticleAdvert]

Call of Juarez scaling remains impressive on the new driver version. However, as we mentioned before, we had a very difficult time getting our Radeon HD 3870 X2 setup under the new CrossFireX driver to operate at anything over 1680×1050 resolution. The built-in DX10 benchmark for Call of Juarez typically allows for testing at all resolutions well above native for the monitor in use, but attempting to run at 1920×1200, as we did in the previous test, resulted in the game locking up completely. Consequently, the resolution we were able to test at was 1440×900, and expectedly, CPU limitation played a much larger factor than we would have wanted while running this test. Even with the decreased weight of graphics performance, the 47% scaling in Call of Juarez is still very good and certainly far better than what the competition has to offer. Also, please note that the pictures included above are from our previous CrossFireX scaling article.
{mospagebreak title=F.E.A.R.}
F.E.A.R.

By now F.E.A.R. is a pretty old game, and many might question why we would even incorporate it into our benchmark suite. Well, despite several drawbacks, F.E.A.R. was actually played competitively for a while and for good reason: it is damn fun to play in multiplayer. The built-in benchmark is probably a bit too slow to represent real gameplay, but due to the incredible range of explosions and shots and particles that can be flying around while playing the game, it is definitely the most reliable method. F.E.A.R. has also always exhibited incredible scaling performance in multi-GPU configurations for graphics cards from both AMD and NVIDIA, so it will be interesting to see what four GPUs can do against the game.

{multithumb popup_type=none enable_thumbs=0} {multithumb popup_type=default enable_thumbs=default}

[ArticleAdvert]

Unfortunately it appears that graphics performance is at the point now that even while playing under the highest possible settings in FEAR, CPU limitation has a pretty drastic effect on performance. With one Radeon HD 3870 X2 enabled, scores with this driver revision are very similar to those achieved with the last. However, when both cards are enabled, we see a full 20 FPS drop thanks to the 200MHz processor speed decrease this time around. Consequently, the scaling numbers we are reporting here are worse than those we showed you earlier. This shouldn’t be interpreted as a decrease in performance from the previous driver revision, because the primary factor here is the slightly different system configuration.
{mospagebreak title=Crysis}
Crysis
Last time around we really wanted to get some Crysis benchmarks in to see what this powerful graphics subsystem could do against the infamous slayer of video cards and high-end systems. Due to time constraints we were unable to run any tests on the game, but made sure that we would be able to in the future should the opportunity to test later revisions present itself. All of our Crysis tests were performed in DirectX10 with no anti-aliasing enabled whatsoever. The only variable between tests was the use of all “Very High” detail settings or all “High” detail settings.

{multithumb popup_type=none enable_thumbs=0}
{multithumb popup_type=default enable_thumbs=default}

[ArticleAdvert]

To be perfectly honest we were not expecting much out of CrossFireX in Crysis, considering the game is currently the poster child for NVIDIA’s “The Way It’s Meant To Be Played” propaganda. However, we were quite impressed that the Radeon HD 3870 X2 configuration was able to produce a 36% increase at Very High settings. The scaling at lower detail was only slightly lower at 33%, but still impressive considering the relatively high frame rates the cards were producing in general. Granted, Crysis tends to run a hell of a lot better on NVIDIA cards in the first place, so you can expect (actually we guarantee *wink wink*) that cards like the upcoming GeForce 9800GX2 will demolish the scores you see here. Overall there is definitely room for improvement, but a 36% performance gain in an NVIDIA title is not something AMD should be unhappy with.
{mospagebreak title=Conclusion}
Final Thoughts and Conclusions

Again, we have to state that the numbers you see here should not be compared to any others that you might come across anywhere else, either due to the fact that those tests will not be using the same driver or they will not have been run on identical hardware configurations. The only other test results that should even be considered fair game to compare with are those previously published in our initial look at CrossFireX scaling back at the beginning of this month. Further, in no way should these numbers be considered directly indicative of the performance that will be available when the final CrossFireX driver is launched to the public next month.

{multithumb popup_type=none enable_thumbs=0} {multithumb popup_type=default enable_thumbs=default}

[ArticleAdvert]

Echoing what we said last time, CrossFireX technology seems to be in pretty good shape even though it is still in its infancy. Although the quality of AMD drivers has increased dramatically over the past three years, it still typically takes about three driver updates past the initial hardware launch to get performance numbers up to the expected level. Granted, this is a hell of a lot better cadence than the competition has offered lately, but we don’t expect to see fully realized performance gains with CrossFireX until sometime in Q2 of this year.

Since these numbers are very similar to those we achieved earlier, with small discrepancies due to the less-powerful supplementary hardware and software used this time around, there is not much else we can say about the current state of CrossFireX that we haven’t said already. What we can definitely say is that there was full compatibility with this evaluation driver across all of the games we tested, which is more than can be said for current beta drivers from another graphics company that will remain unnamed. We promise to keep you updated should any more driver updates be released between now and the official launch of Catalyst 8.3, though we do not really expect that to happen.

Popularity: 35% [?]

Support FPSLabs! Share and Enjoy: