Wednesday, September 23rd 2009

LucidLogix Fires up Multi-GPU Computing With Faster, More Flexible HYDRA 200 Chip

LucidLogix (Lucid) today introduced the HYDRA 200 real time distributed processing engine designed to bring multi-GPU computing to the masses.

For the first time ever, motherboard, graphics card manufacturers and users can have the flexibility to use different combinations of GPUs from AMD (ATI) and NVIDIA in notebooks and PCs. The solution delivers faster 3D graphics at consumer price points.

The new HYDRA 200 SoC is Lucid’s second generation parallel graphics processor that works with any GPU, CPU or chipset to provide scalable 3D graphics performance in a multi-GPU computing environment. On display for the first time at IDF 2009 (booth 213) HYDRA 200 is faster, more flexible, smaller and more power-efficient than its predecessor silicon, the HYDRA 100.

“We’ve further refined our HYDRA engine and made it faster and more flexible, allowing for a near limitless combination of GPU’s,” said Lucid vice president of research and development, David Belz. “HYDRA 200 allows the consumer to get more ‘bang for their GPU buck’ by extending the life of their current GPU investment, providing even faster graphics performance and later upgrading their system with whatever card they choose.” Until now, multi-GPU systems have been graphics vendor specific and generally require the consumer to be fairly technically savvy. With Lucid HYDRA 200, OEMs can offer custom configurations at different price/performance targets, and consumers will be able to easily add graphics hardware to achieve an overall performance boost without the worry of compatibility.

Gamers with a need for speed now have a solution that’s optimized for their performance requirements by allowing them to choose the right chip for the job, or simply upgrade without throwing away the old one. HYDRA 200 currently supports Windows Vista and Windows 7 operating systems as well as DirectX 9.0c and 10.1 standard APIs and is DirectX 11 ready.

HYDRA 200 Tech Specs
The adaptive and dynamic parallel graphics load-balancing scheme of HYDRA 200 resolves bottlenecks in the system and optimizes the usage of GPU resources with minimal power consumption. HYDRA 200 is a 65nm PCIe compatible SoC that also features:
  • Low power use of under 6W, making it ideal for graphic cards, notebooks or desktops as there is no need for a special heatsink
  • A small footprint (18-22mm) that allows for a compact design
  • GPU connector free, making it easy to integrate into systems
  • Supports multiple display configurations
  • Universal GPU, CPU and chipset support
  • For systems using dual, tri or quad GPU combinations
“I cannot say the graphics wars are over, but technology demonstrated by Lucid clearly levels the playing field when it comes to GPU scaling,” said Dr. Jon G Peddie. “And with multi-core graphics growing at a rate of ten percent year over year, it will be interesting to see what kind of momentum a mass market M GPU can have in the marketplace.”
HYDRA 200 Availability
With universal GPU support and a variety of configurations, the HYDRA 200 makes system integration swift and worry-free. HYDRA 200 is available now for reference designs in three models:

Source: LucidLogix
Add your own comment

38 Comments on LucidLogix Fires up Multi-GPU Computing With Faster, More Flexible HYDRA 200 Chip

#1
TheMailMan78
Big Member
I cannot compute just how awesome this truly is. This could usher in a new era of folding.
Posted on Reply
#4
PP Mguire
This is totaly awesome, and an awesome company has it on an awesome board.
Posted on Reply
#5
erocker
Well then. My 4890 Toxic should look nice next to a 5870 Toxic and perhaps a Nvidia card thrown in there for good measure.. and PhysX. :D I can't wait to see other motherboard manufacturers pick this up.
Posted on Reply
#6
TheMailMan78
Big Member
by: erocker
Well then. My 4890 Toxic should look nice next to a 5870 Toxic and perhaps a Nvidia card thrown in there for good measure.. and PhysX. :D I can't wait to see other motherboard manufacturers pick this up.
I wonder if we will see some lawsuits with this. Also I wonder if your performance will only be as good as your lowest powered card. Hmmmmm??
Posted on Reply
#7
Jizzler
(Going off their earlier PR...)

All cards take on what they can handle in proportion to the others. 4350 + 4550 + 4770 + 4890 should be the cumulative performance (less overhead) of the total.


First company to throw a lawsuit out there... I'm buying from your competitor :)
Posted on Reply
#8
TheMailMan78
Big Member
by: Jizzler
(Going off their earlier PR...)

All cards take on what they can handle in proportion to the others. 4350 + 4550 + 4770 + 4890 should be the cumulative performance (less overhead) of the total.


First company to throw a lawsuit out there... I'm buying from your competitor :)
So this will make the cards use combined power instead of sharing the lower performance? Is this correct?
Posted on Reply
#9
Error 404
So this means I could keep my 4850 and add a GT300 or a 5850/70 to a motherboard with this chip? That would rock so hard that I'd have to get a straightjacket... :roll:

I really hope there aren't any lawsuits over this device, because it must've taken them ages to get this up and running and it will be brilliant to see in the market. Any idea on how much they add to the cost of a motherboard?
Posted on Reply
#10
TheMailMan78
Big Member
by: Error 404
So this means I could keep my 4850 and add a GT300 or a 5850/70 to a motherboard with this chip? That would rock so hard that I'd have to get a straightjacket... :roll:

I really hope there aren't any lawsuits over this device, because it must've taken them ages to get this up and running and it will be brilliant to see in the market. Any idea on how much they add to the cost of a motherboard?
That board would be worth 600 bones easy. However I doubt they will retail for that much :rockout:
Posted on Reply
#11
PP Mguire
by: erocker
Well then. My 4890 Toxic should look nice next to a 5870 Toxic and perhaps a Nvidia card thrown in there for good measure.. and PhysX. :D I can't wait to see other motherboard manufacturers pick this up.
This dosent mean that at driver level it will make it possible to run Nvidia and ATI in the same loop. Its just a load balancing chip to scale "perfectly" with any game.
My question is how well will the load balancing between tri-sli and quad crossfire will be. In tri-sli it should actualy make that 3rd card usefull even in the games that dont scale well.
Posted on Reply
#12
DaMulta
My stars went supernova
I wonder when Nvidia will buy them out......
Posted on Reply
#13
tigger
I'm the only one
They should tell nvidia to feck off.
Posted on Reply
#14
Kenshai
Currently this won't work on the driver level. So I don't really see how this benefits games, perhaps something other than games though.
Posted on Reply
#15
1Kurgan1
The Knife in your Back
by: TheMailMan78
I cannot compute just how awesome this truly is. This could usher in a new era of folding.
I gotta say, that sounds exactly like Dwight, lol. :laugh:

It really is awesome though, I wonder when we will actually be seeing this on mobo's.
Posted on Reply
#16
Jizzler
by: TheMailMan78
So this will make the cards use combined power instead of sharing the lower performance? Is this correct?
Yup.

by: LucidLogix
The HYDRA engine analyzes the frame before rendering and intelligently distributes the rendering tasks between the GPUs on board. The frame decision mechanism resolves bottlenecks and inter-frame dependencies prior to rendering, in real time, such that there is near zero latency.

The HYDRA engine is a generic solution for different games, as well as rendering methods and an auto-correct load-balancing scheme for scaling. For GPUs that are not identical in performance or in the manufacturer name; the HYDRA engine allocates the resources appropriately during processing for optimization of the GPU rendering power.

Lucid is unique in finding a way to deploy high-level parallelism of graphics application with the best load balancing of the platform GPUs. It can decompose a complex scene, and then recompose each task into the correct final image. Achieving such challenges is what makes the solution "generic" and compatible with chipsets, GPUs and applications from any vendor.
I hope that LucidLogix has really pulled it off.
Posted on Reply
#17
DaMulta
My stars went supernova
by: tigger
They should tell nvidia to feck off.
You already know it's going to happen soon or later lol.
Posted on Reply
#18
enzolt
this sounds so promising!! i cannot wait!!
Posted on Reply
#20
spy2520
Are there even any grounds for a lawsuit? Its not really like an nForce chip and they arent circumventing drivers or anything. Maybe they will be, i dunno...

I'm just wondering what nVidia could say if they were going to say something.
Posted on Reply
#21
theubersmurf
by: DaMulta
I wonder when Nvidia will buy them out......
They won't be, one of the venture capitalists who invested in the company was intel's venture captial company.
Posted on Reply
#22
pr0n Inspector
errrr... you guys are worrying about lawsuits instead of the latency this thing adds?
Posted on Reply
#23
Kenshai
by: pr0n Inspector
errrr... you guys are worrying about lawsuits instead of the latency this thing adds?
Why would it be adding any latency. It potentially does the same thing as nvidia's chip.
Posted on Reply
#25
PP Mguire
It shouldnt be any less exciting than x58. And besides, if this chip does what its supposed to do, the p55 boards will rape.
Posted on Reply
Add your own comment