After a bit of ballyhoo and a bit more of a delay, NVIDIA is finally ready to launch their competitor to AMD’s triple-monitor Eyefinity technology: 3D Vision Surround.

As a quick refresher, in September of 2009 AMD launched their multi-monitor Eyefinity technology alongside the Radeon HD 5000 series. With Eyefinity AMD could present a Single Large Surface to games and applications, allowing them to draw to 3 monitors as if they were a single monitor. It allowed for computing and gaming at a very wide field of view approaching the limits of human vision.

Not to be left out, NVIDIA decided to counter back with their own take on the technology: 3D Vision Surround. We first learned about 3D Vision Surround at CES 2010, where NVIDIA was officially announced the technology and was offering both public and private demonstrations of the technology. At the time they had it running on both GTX 200 series cards and what would become the GTX 400 series. 3D Vision Surround was to be NVIDIA’s competition to AMD’s Eyefinity technology and then-some: not only would NVIDIA match AMD’s Eyefinity triple-monitor capabilities in the 2D space, but they would extend the concept by merging it with their 3D Vision technology for 3D Vision Surround.

NVIDIA has previously told us that they’ve been sitting on the concept for some time with no apparent market for it, with the success of Eyefinity and Matrox’s TripleHead2Go finally motivating NVIDIA to move forward with the technology. The result of this delayed plan is an interesting technology that in many ways is NVIDIA’s version of Eyefinity, and in other ways is entirely different. In a nutshell: it’s not just 3D Eyefinity.

Today’s Launch

This morning NVIDIA is launching their 258.69 beta driver, the first public driver to offer 3D Vision Surround functionality. NVIDIA did not sample this driver to the general press ahead of this launch so we will not be providing a review for today’s launch. We will have our own review in the coming weeks, as we’re still working on acquiring a complete set of 120Hz LCD monitors to properly test both NVIDIA Surround (2D) and 3D Vision Surround (3D).

In lieu of that we have already been provided a technical briefing for the technology which for the time being enables us to answer some of the biggest questions we had about the technology prior to today’s launch.

First and foremost, as we’ve discussed in previous articles, NVIDIA is technically launching two different technologies today. The first is NVIDIA Surround, the name NVIDIA is giving to their Eyefinity-alike 2D multi-monitor technology. The second is 3D Vision Surround, which is the infusion of 3D Vision in to NVIDIA Surround. Admittedly the naming could use some work (“NVIDIA Surround” does not roll off the tongue quite like “Eyefinity”) but it’s fairly straightforward in conveying which one is for 3D. For the sake simplicity in this article, we’ll be referring to the overall technology as NVIDIA Single Large Surface (NVSLS) when discussing matters that apply to both NVIDIA Surround and 3D Vision Surround.

Although NVIDIA may have been sitting on NVSLS for quite some time, the fact of the matter is that by the time they decided to launch it, they were already too far along in the design process of GF100 to do anything about it on the hardware level. Whereas AMD could make hardware changes to facilitate Eyefinity – primarily by enabling more display outputs – NVIDIA could not. This has some drawbacks and some benefits.

GTX 480: Only 2 outputs can be used at once, requiring SLI for NVSLS

In terms of drawbacks, the lack of dedicated hardware means that virtually none of NVIDIA’s cards have enough display outputs for NVSLS. With the exception of a single model of the GeForce GTX 295 that has an HDMI output on the daughter card, 2+ cards operating in SLI are required to take advantage of NVSLS. This is due to the fact that the second card’s display outputs are needed to drive the 3^rd monitor. This gives NVSLS a higher setup cost than Eyefinity, which can be done for up to 6 monitors on a single card. Along those lines is NVIDIA’s other current limitation: they can only do 3 monitors right now while AMD can do 6.

However there are also benefits of NVIDIA’s software implementation. While AMD relied on hardware and limited Eyefinity to the Radeon HD 5000 series as a result, a pure software solution allows for the technology to be backported to older cards. Along with the GTX 400 series, the last-generation GTX 200 series will also be gaining NVSLS capabilities today – this is for both NVIDIA Surround and 3D Vision Surround. There are a couple more limitations at the moment (3-way SLI is not supported on the GTX 200 series) but the fundamental technology is there. Furthermore in this brute-force manner NVIDIA also tidily bypasses any reliance on DisplayPort, so unlike Eyefinity NVSLS will work without an active DP-to-DVI adapter.

The biggest remaining question right now will be whether a pure-software approach differs from AMD’s hardware + software approach in terms of performance and game compatibility. NVIDIA’s own internal benchmarks have a SLI GTX 480 setup beating a CF 5870 2GB setup, but the GTX 480 is already faster than the Radeon HD 5870 so this wouldn’t be wholly surprising. As for compatibility we do know that NVIDIA is still fighting with the issue much like AMD has been, as NVIDIA is suggesting the use of the 3^rd party Widescreen Fixer to fix the aspect ratio of several games.

The Next Step: 3D

Moving on, NVIDIA’s ace in the hole here is clearly 3D Vision Surround, offering a feature that AMD is still months off from being able to match. By combining their existing 3D Vision technology with NVSLS NVIDIA can offer ultra-widescreen 3D, and having seen it in person at CES we’ll be the first to profess that it definitely looks impressive. With respect to 3D Vision there’s nothing new here – it’s just the same glasses now looking at more than 1 monitor – but it’s a natural extension of the technology.

For those of you interested in the nuts & bolts of how 3D Vision Surround will work, NVIDIA also released some additional technical details on the feature. With 3D Vision Surround NVIDIA is faced with a great deal of rendering to do: not only do they need to render a very large frame to cover 3 monitors, but then they need to render it again for the other eye. In doing this, they have taken an interesting approach to dividing up work – this image from their press kit pretty much says it all:

In short NVIDIA has opted to stick with Alternate Frame Rendering while at the same time having 1 GPU render both the left and right eye versions of any individual frame, rather than having each GPU work on each eye. It’s truly alternate frame rendering rather than alternate eye rendering. Under normal circumstances having the same GPU render two images in a row would increase input lag, but when it comes to 3D Vision there’ s no penalty since the second image represents the same gamestate as the first image, meaning the pre-rendered frame count isn’t actually higher as it would initially appear.

Meanwhile 3D Vision Surround also puts further restrictions on the hardware compared to NVIDIA Surround. A big difference will of course be performance due to rendering another image for the 3D effect, but there’s also a matter of monitors. For NVIDIA Surround the monitor requirements are analogous to Eyefinity: 3 monitors at the same resolution, refresh rate, and sync polarity. However for 3D Vision Surround, monitors must be more than similar: they must be identical. This is because 3D Vision is heavily reliant on V-sync timing to match up a frame with blocking the correct eye, and different monitors can have slightly different refresh timings even though they operate at the same refresh rate. As a result all 3 monitors must be the same to ensure they all refresh at the exact same moment.

The other interesting quirk when it comes to 3D Vision Surround and monitors is portrait orientation. For NVIDIA Surround, NVIDIA holds parity with AMD straight down to the support of landscape and portrait orientations. But with 3D Vision, horizontal linear polarization comes in to play: because both the monitor and the glasses are polarized for glare reduction and image blocking respectively, they have to be properly aligned. Anyone who has tilted their head when viewing 3D through a linear system has seen what happens if the screen and glasses are not aligned: the polarization blocks the entire image. As  a result 3D Vision Surround is not currently usable in portrait mode when used in conjunction with an LCD monitor – only projectors are supported.

Last but not least, there’s a matter of software. While NVIDIA is 9 months behind AMD overall when it comes to triple-monitor gaming, they’re starting off in a better position than AMD did. It wasn’t until March of this year that AMD delivered on bezel correction for Eyefinity, meanwhile NVIDIA is launching with it today. Even on this timeline NVIDIA is still behind AMD, but with this taken in to account they’re not as far back as it would first appear. Grouping groupies may be disappointed however – while we don’t have the software in hand to confirm this, it doesn’t look like NVIDIA has any monitor grouping features at this time.

First Thoughts

Without the software in hand there’s not much more we can say about NVIDIA Surround and 3D Vision Surround at this time. We are of course interested in the performance of NVIDIA’s solution, not only in comparison to AMD’s Eyefinity, but also comparing the GTX 200 series to the GTX 400 series and seeing the performance hit to moving to 3D Vision Surround from NVIDIA Surround modes. Teething issues will also bear watching as this is NVIDIA’s first beta driver : we already know GTX 200 series 3-way SLI isn’t supported, and that anti-aliasing modes above 2x on 3D Vision Surround are also unsupported – both things we would hope to see NVIDIA fix down the line.

Perhaps the best news for the moment though is that this should help to further legitimize the concept of triple-display gaming with game developers. While it’s not a difficult technology to work with, having only 1 GPU manufacturer support it made it yet another manufacturer-specific feature. With NVIDIA on board this will provide further incentive for developers to take the technology in to consideration. Since the biggest thorn in the side triple-display gaming continues to be the lack of proper aspect ratio support, any progress here in converting developers will be of benefit for both sides.

In the meantime stay tuned for our full review of NVIDIA’s 3D Vision Surround later this month.