Single page Print

A look at Lucid's Virtu MVP Mobile


...on Origin's 11.6-inch speed demon
— 2:44 PM on October 18, 2012

If you read our news section, you might have seen us talk briefly about the latest from Origin. No, I'm not talking about the Steam substitute to which EA keeps tying its new games. I'm talking about Origin, the Florida-based boutique gaming PC vendor founded by ex-Alienware employees three years ago.

In August, Origin became the first company to offer Lucid's Virtu MVP virtualization technology on a mobile platform. All of its Eon-series laptops, from the diminutive 11.6" offerings to the 17.3" monsters, became available with Lucid MVP software as a free, no-strings-attached option.

The touted benefits are very much the same as for Virtu MVP on the desktop. Lucid's GPU virtualization mojo purportedly provides power-saving benefits by dynamically assigning tasks to the most ideal graphics hardware. The MVP software also includes Virtual Vsync, which is meant to eliminate screen tearing without compromising input responsiveness, and HyperFormance, which is supposed to boost responsiveness further. It all sounds rather enticing, especially on a notebook.

We were intrigued.

Happily, the folks at Origin offered to send us their EON11-S, a compact gaming notebook with an 11.6" 1366x768 display and some surprisingly fast hardware under the hood. The base config starts at $955 with a Pentium B960 processor, 4GB of RAM, GeForce GT 650M 2GB discrete graphics, and a 320GB 7,200-RPM hard drive. Our review sample has been optioned up to include a faster processor, the quad-core Core i7-3616QM, as well as double the RAM and a 256GB Samsung 830 Series solid-state drive in place of the mechanical hard drive. Nevertheless, the machine has the same 11.2" x 8.1" footprint and 1.4" thickness as the base model, and it tips our postal scale at 3.72 lbs—rather light for a quad-core gaming rig with a relatively speedy GPU.

Could this diminutive gaming laptop be further enhanced by the addition of Virtu MVP Mobile? That's what we set out to discover.

How exactly does Virtu MVP work?
Virtu MVP is, as its name indicates, is a virtualization technology that adds an additional software layer between the operating system and the graphics processor(s). Lucid's driver intercepts DirectX calls from games and other applications and fulfills them with the help of the GPUs, to whom it doles out work. On Intel-powered desktop systems, one of Virtu MVP's biggest selling points is that it lets games run on a discrete GPU without walling off access to the Intel integrated graphics' Quick Sync video encoding block. Without Virtu, plugging in a discrete GPU disables the IGP and makes QuickSync unavailable.

Virtu MVP offers power efficiency benefits, as well. On a desktop, users can plug their display into the output port for integrated graphics and still have access to the discrete GPU. That discrete GPU is called upon by the virtualized driver to run games and other demanding applications, and frames are forwarded to the IGP for output to the display. When it's not needed, the discrete GPU sits at idle, and the IGP handles menial graphics duties on the Windows desktop. Lucid calls this configuration i-Mode. Users also have the option of connecting the display to the discrete GPU, a setup Lucid refers to as d-Mode. This approach doesn't present the same power-saving benefits as i-Mode, but it does enable full access to the proprietary control panels and driver optimizations of GPU vendors like AMD and Nvidia.

On notebooks like the EON11-S, there's no way to switch manually been i-Mode and d-Mode. Instead, Origin tells us Virtu MVP Mobile runs on top of Nvidia's Optimus dynamic switchable graphics technology. Optimus determines whether the IGP or the discrete GPU drives the display, calling upon the discrete GPU only when the user runs graphically demanding tasks like games. Origin says Virtu MVP Mobile switches between i- and d-Modes accordingly, all on the fly.

Virtu MVP and its mobile implementation combine the aforementioned virtualization technology with two rather intriguing features. Those are dubbed Virtual Vsync and HyperFormance.

We'll tell you what we know about them, but we should explain up front that piecing together a crystal-clear picture of these features' inner workings has proven to be rather difficult. We've spoken with Lucid about them multiple times, and whenever we press for details, the company's explanations tend to be either not entirely clear or a little bit evasive. We get the sense that this vagueness is somewhat intentional. Perhaps Lucid simply doesn't want to get down too deep in the nitty-gritty of things with journalists, lest competitors figure out the recipe to its special sauce.

We do have a rough notion of what the features do, though, and Lucid has been explicit about the expected benefits. So, we can at least enlighten you on that front.

Both features rely on knowledge of the display refresh cycle to determine when and where screen tearing occurs. Screen tearing, for the uninitiated, refers to instances where the display ends up showing parts of two or more different frames during the same refresh cycle. Tearing can happen whenever the GPU's frame rate doesn't match the display's refresh rate. In cases where the GPU is too fast, the display might receive frame B before it's finished drawing frame A, and then frame C might arrive before frame B has been fully drawn. So, you might see the top third of frame A, the middle third of frame B, and the bottom third of frame C during that refresh cycle. If the GPU is too slow, the display might have to re-draw part of a stale frame until the next frame comes along. The result is visually similar.

Virtual Vsync works to prevent tearing when the GPU delivers too many frames. Lucid's software determines which frames would make it to the display before the end of each refresh cycle, and it simply discards surplus frames. The GPU still does the same amount of rendering work, but surplus frames are never sent to the display. Traditional vsync, by contrast, forces the GPU to match the display's refresh rate, which usually prevents more than 60 frames per second from being rendered (since most LCD panels have a refresh rate of 60 Hz).

Traditional vsync may seem like the more efficient approach. According to Lucid, however, vsync has the side-effect of reducing how rapidly the game responds to keyboard, mouse, or controller input. That's because limiting the frame rate can also slow down the game's main program loop, which processes all of those inputs every time it runs. A fast-paced multiplayer shooter might run at 120 FPS on a modern gaming rig, which would mean its game loop refreshes every 8.3 ms. Enabling traditional vsync would lengthen those refresh cycles to 16.7 ms, since the frame rate would be limited to 60 FPS. Using Virtual Vsync, on the other hand, would preserve the rapid refreshes and eliminate screen tearing.

Virtual Vsync can be complemented by HyperFormance. When both features are enabled, Virtual Vsync discards surplus frames as usual, and HyperFormance jumps in to ensure those surplus frames aren't fully rendered by the GPU. Some rendering tasks, like texture fetches, must still be accomplished in order to keep GPU caches warm and to fulfill inter-frame dependencies. However, tasks like post-processing and some shader effects can be skipped altogether. That means unseen frames are processed faster, the game loop runs faster still, and input responsiveness is, hopefully, even higher.

HyperFormance can also be used on its own—without Virtual Vsync. In that case, Lucid tells us the software allows screen tearing to occur, but it ensures partially displayed frames aren't fully rendered. The exact mechanics involved elude us. However, the goal seems to be the same as for HyperFormance with Virtual Vsync: speed up the game loop in order to boost input responsiveness. Interestingly, Lucid says using HyperFormance on its own can actually amplify input lag in some titles.

Outside of such fringe cases, the theory is that using HyperFormance, Virtual Vsync, or both technologies together shouldn't incur a performance hit. Lucid's timing algorithms run on the Intel IGP, so the discrete GPU doesn't need to do any extra work to keep track of frames on the display; it's free to devote all of its attention to rendering. Sounds pretty much like a free lunch.