With a tag line of "Any GPU, Unlimited Possibilities", you can imagine our collective geek meters were pegged when we received word from a contact at Israeli startup Lucid Logix about the company's formal debut announcement. And Lucid's roll-out press release went on to tease a bit more... "The real-time distributed processing engine, called HYDRA™, offers a groundbreaking approach to scaling 3D graphics performance in a multi-GPU environment. Outlined for the first time on the new Lucid Web site at www.lucidlogix.com, the novel technology is the industry’s first to work with any GPU, any CPU or chipset and on any application that gamers and professionals may choose... The HYDRA engine combines a system-on-chip with exclusive software technologies that efficiently load-balances graphics processing tasks, delivering another industry breakthrough: near-linear to above-linear performance with two, three or more graphics cards. All together, the technology means people will be able to customize and tweak their PC to achieve the price, performance and power balance they want – whether it is a mainstream system or a high-performance gaming package."
Hang on a second, let's get our heads around this a bit. This thing isn't a GPU and it's not a motherboard chipset but it's a sort of GPU loadbalancer, right? Check, that's the long and short of this new technology but perhaps it's the performance claims that really get our attention. Not only is HYDRA able to co-join GPUs from both NVIDIA and AMD-ATI, as the diagram to the right shows but the claim is that we'll see a "near-liner" to "above linear" performance boost; Hello? Just how can we expect an "above linear" gain from two GPUs working in unison? We've lobbed a few pointed questions over the wall to our contacts at Lucid and will update this story if more detail becomes available.Update, 7/14 1:30PM EST:
We just received word from Lucid's President, Offir Remez on the subject of "near-linear" or "above-linear" performance gains with HYDRA technology.... "Lucid technology is task based. We make sure similar tasks stay coherent to the specific GPU (or GPU core). This in turn creates enhanced GPU cache hits. So in some cases we enjoy from an indirect additional performance boost.”
Coined a "real time distributed processing engine" the Lucid team is developing an SoC (System On a Chip) technology that allows mixing and matching of GPUs with a claimed efficiency that even the likes of NVIDIA and AMD haven't been able to achieve as of yet. Backed by VC firms Giza VC, Genesis Partners and Intel Capital, the company and technology, if it gains any significant traction, could shake up the landscape big time and of course be a prime target for acquisition by any of the major players in the 3D Graphics arena.
Lucid Logix was actually formed back in 2003 by a few industry veterans including former Silicon Graphics pioneer, Efraim Fogel. According to their release, Lucid is already engaged with major OEMs and ODMs targeting platform solutions for either motherboard or graphics card applications. Reportedly the HYDRA engine requires no modification to existing graphics driver software to function with standard graphics applications like DirectX. More on the announcment can be found here