Performance Evaluation and Initial Thoughts
Performance EvaluationOur performance evaluation needs to look at all three “modes” that the current HYDRA hardware and software support: N-mode, A-mode and X-mode for cross-vendor support. Another aspect to consider with our HYDRA N-mode results is not just raw performance advantages, but efficiency in those performance increases. For example, Batman at 1920×1200 ran at 160.3 FPS with both GPUs at work while it ran at 115 FPS and 87 FPS on the GTX 285 and GTX 260+. That means that we saw a 50% utilization in the power of the GTX 260+ when added to the GTX 285 card. The math goes like this:
- (160.3 (combined FPS) – 115.1 (GTX 285 FPS)) = 45.2 (gained FPS) / 87.4 (GTX 260+ FPS) = 0.51 = 51%
But, when we look at the Far Cry 2 results, the efficiency isn’t as great. Take the same 1920×1200 test and you’ll see the same equation results in an efficiency of only 12%.
Obviously our A-mode testing results were less than impressive than our NVIDIA-only scaling results. If we look at efficiency once again, you can see that Batman at 2560×1600 utilized the 2nd GPU the best getting as much as 36% of the GPUs performance in the combined configuration. However, there are a couple games like World in Conflict that see a 10% efficiency while HAWX (at 1920×1200 only) sees just over 20%.
While HYDRA’s X-mode is giving us SOME performance scaling it is intermittent at best. You will notice that we did not include Far Cry 2 in our testing here – it simply would not run with the GTX 285 and HD 5870 working together. Resident Evil 5 was also a no-go situation creating enough visual tearing that completely ruin any semblance of a solid gaming performance; and performance was dropped below that of a single card anyway.
The cross-vendor implementation is also less consistent: we saw dramatic performance differences between even different resolutions in the same game. Take Tom Clancy’s HAWX for example: we saw a 35% performance boost over a single card when playing at 1920×1200 and only a 3.5% boost when playing at 2560×1600! At the lower resolution the HYDRA technology was able to get 36% efficiency out of the second GPU while at 2560×1600 that 3.5% overall performance boost only takes advantage of 3.8% of the frame rate of the second card.
When it works, is it good?
This is kind of a loaded question of course, sort of like asking “if someone gave you a million dollars, is that cool?” but I feel that it is worth discussing to view this technology from all angles. And actually, when it does work, the Lucid HYDRA technology is impressive in what it is able to accomplish in terms of a unique multi-GPU scaling solution. The Lucid development team has a lot more experience with the NVIDIA drivers and technology as that was originally the only target goal for HYDRA and it shows in the much better scaling and compatibility we saw when testing the MSI Big Bang Fuzion motherboard in N-mode. When combining the power of the GeForce GTX 285 and the GTX 260+ we saw up to 40% scaling over the performance of only the GTX 285 though the average was probably closer to 28-30%. That is actually pretty reasonable performance improvement for combining these different GPU graphics cards especially considering it is simply impossible to due through any other method.
Current game list with HYDRA’s driver
The AMD results were less exciting and the cross-vendor X-mode results were even bigger downers in our view. There were some cases where it did “just work” like in Batman or HAWX (in A-mode) and only at 1920×1200 in those same titles for X-mode. In other cases the games simply didn’t scale at all or had some kind of rendering error. To be frank, even Lucid knows the list of games supported in cross-vendor gaming mode isn’t as far as long as the NVIDIA-mode or even the ATI-mode. Since we have had this board for the 6-8 weeks or so we have seen Lucid release a steady stream of new drivers that added fixes and new games so there is at least a concerted effort on their part to keep the technology moving in the right direction.
So when the HYDRA scaling does work correctly, it is impressive to see. When we first heard that Lucid was aiming to allow NVIDIA and ATI graphics cards to work in tandem I was incredibly interested and excited by the possibilities and flexibility it would offer gamers. I still think that the technology has a shot but they obviously have an uphill battle to get the performance scaling and game compatibility lists where they need to be.
The MSI Big Bang Fuzion Motherboard
This does lead us to the discussion of whether or not the MSI Big Bang Fuzion motherboard is worth buying. Obviously the Lucid HYDRA technology is not the board’s only feature but it is by far its most differentiating feature. The motherboard shares pretty much everything in common with the Big Bang Trinergy motherboard including support for the latest LGA1156 Intel processors, triple graphics ports, tons of SATA ports and a highly overclockable motherboard and BIOS design. (You can see all of the overclocking features that the MSI Big Bang series offers here!) But the Trinergy motherboard offers not only SLI support but 3-Way SLI support while the Fuzion replaces that feature with the Lucid HYDRA solution. The Fuzion board will be a great enthusiast class motherboard that anyone will be able to take advantage of and I would have no qualm recommending it if…
…it supported NVIDIA SLI. But it doesn’t. The Fuzion motherboard has support for HYDRA multi-GPU scaling technology and for AMD’s native CrossFire technology but NVIDIA and MSI and Lucid, for one reason or another, decided to not allow/permit/purchase the license for SLI technology. Remember that because CrossFire is license free, it can be included on any motherboard with the necessary GPU slots but NVIDIA still requires motherboard vendors to pay a flat fee per board or per series in order to use the SLI technology and logos. Whether or not this is NVIDIA’s fault or MSI’s, it’s a bad move in my view, because had this motherboard offered SLI support, I could easily recommend it. Why would a user NOT want to have the only motherboard on the market that supports all forms of multi-GPU configurations including CrossFire and SLI and this new, maybe questionable and potentially very intriguing, HYDRA technology? Without SLI support, the users are basically taking a risk that HYDRA will pan out to be better than it is in its current state and that the value of it will outweigh what SLI could add.
Initial Thoughts
As you can probably tell from my tone and our testing, I am conflicted on this product. I really wanted it to work at 100% so that I could recommend a GPU-scaling option that takes the best that both NVIDIA and AMD offer and combines them into a single solution but it just isn’t there yet. Lucid’s HYDRA technology still has the potential to be better and if their software team can continue to pump out updated drivers with more performance improvements and compatibility fixes, there is still the chance for it to be successful in its current form. But both NVIDIA and AMD will be glad to sit down and chat with anyone about how difficult writing drivers for all of these games can be and it would appear that Lucid is finding that out as well. As mentioned above, had this motherboard offered SLI support I would still give the MSI Big Bang Fuzion the recommendation as it offers TONS of awesome features and overclocking options with the flexibility of HYDRA down the road as it improves. Without it, I can’t recommend Fuzion for anyone but the most hardcore and curious users (or those that will stick with AMD CrossFire solutions) that want to play with Lucid’s technology as it grows, without the SLI fallback.
You can find the MSI Big Bang Fuzion motherboard at Newegg.com $369.
The MSI Big Bang Trinergy motherboard sells for $329.
