Lucid's GPU-mixing HYDRA Engine gets previewed, shows real promise
After a bit of a delay Lucid is back on our radar -- and we're pleased to say that the company's technology for combining multiple GPUs seems to be well on course. Indeed, a few HYDRA 200 units appear to be out and about, with preliminary tests showering love on the device. When the kids at Hot Hardware put theirs through its paces they encountered drivers issues (not surprising for something still in development) but still reported running a mixed ATI / NVIDIA combo for "impressive scaling" most of the time, peaking at 85% in dual-GPU mode. And PC Perspective also had positive things to say, with the HYDRA pushing performance "up by 83% -- definitely a competitive solution to SLI!" But that ain't all -- for the real nitty gritty you'll have to check out the links below.























Before anyone asks if it runs crysis...
Will it ever get old 0_o ?
100 years from now, people will be asking that same question on computers
You misspelled "graphing calculators" :)
Its because computers in a 100 years will never run it perfectly because of poor programming and bad product support from EA.
This looks great
I can barely wait for this to hit the market ....
That is quite amazing. Not terribly practical (well maybe in a rare few cases) but still damn impressive.
I personally would find this nice, since I usually out of habit build a new machine with a mid-range nVidia card (9600GT), and then the high-end, cheap ATI cards come out a few months later (5850/5870) and I don't see a reason to junk my perfectly good GT just for a little boost from the ATI card. This would give me the best of both worlds, plus freedom to switch between XFire and SLI as I saw fit, and yeah, I know there are boards that already do that, but they are limited in other ways more than likely.
So, the idea of being able to throw any two cards into your system, or upgrading an existing card by adding a second unmatched card isnt the least enticing? This means no more card matching, no more ATI or nVidia only boards, no Crossfire or SLI connectors. This is the way it should be, drop in a card and it will work, no matter what.
Actually yes I retract my previous thought. Wonder if there will be compatibility issues though? e.g NVIDIA and ATI launching new drivers, trying to block it out and/or break it etc.
If i was a GPU manuf. i would embrace this, then people who are Pro Nvidia might actually buy an ATI card or vise versa, it uses the best of both worlds and will result in more gpu sales for both companies.
Also if a 3rd party does it it looks less like a monopoly.
Actually Raptor, you logic sounds decent but I look at it this way. In order to get an SLI setup going, you have to buy two cards, no? If you use an existing card from your setup, you could use this and only buy one card. Really, I think would drive down sales just because one can mismatch cards and run the software and get good results.
I don't see the manufacturers embracing this. I expect them to find new ways to block it with firmware updates.
At this point, AMD/ATI will not block it since they have lower marketshare. Nvidia likely will block it, as they have majority marketshare (for discrete graphics cards at least), and have done stuff like this in the past (hardware PhysX is blocked if an ATI card is detected).
Also, keep in mind for those who buy motherboards usually have to choose between CrossFire and SLI (few motherboards support both, but it's usually one or the other). These chipset makers have to pay ATI or nvidia license fees for xfire/sli support. HYDRA may cut into that as well.
so is that 85% of top speed per card? or is that 85% of the cards speed in SLI or Crossfire?
The first statement suggests that it's 85% of the combined performance of the two cards. The second statement makes it sound like it's 85% better than running one card or the other. Maybe I'm reading them wrong, but that's how it appears to be worded.
Its an 85% improvement over using 1 card. This is about on par with Crossfire and SLI. So long as it scales with the performance of the cards used, thats expected.
@Brad Green
SLI/X-fire doing scaling by 85%? Yeah right... Maybe 50%....
Oh this could make some sick possibilities. Now to see what they can do by putting 4 cards in one system and see the total output.
(wonder if it degrades for each card added)
Kick ass! I had my hopes high for Hydra... now I'm certain my next motherboard will be one with Lucid's Hydra solution. Hope to get one with 3 PCI-E slots for epic frame rates.
Lot of skeptics in the last article- I say power to Lucid for some seriously groundbreaking technology.
Now, what are the odds that AMD/Nvidia attempt a buyout to keep the technology out of the hands of the competition?
Or maybe Intel will buy them out to encourage gamers to go with Intel chipsets?
Can I run GT 220 for PhyX and Radeon 5870 for the rest?
You can do that right now without any extra chip.
@TylerDurden
You have to hack nvidia's drivers to support hardware PhysX if there is an ATI card in the system.
Incredible. I'd have never thought I'd see the day an ATi and Nvidia GPU played together nicely!
Dear Diary,
Jackpot.
Haha! Quagmire
This is awesome, especially for upgrading. Just buy a single card then buy a newer card single card and use them both! Amazing!
Ditto...
So with my GTX 285, I could use my antique 7900GT to boost framerates even by a couple of frames? Pretty cool.
I may get an HD5870/HD5890 to compliment the GTX 285.
I really hope this becomes a reality in a year or so and Nvidia or ATI doesn't try and block it out with driver updates.
I am very much interested in this tech. I currently have an asus 5870 as my display and a evga 9600gt oc as a physx card and secondary. I have this in windows 7 x64 and it works flawess up to a degree. I can only use fw 182.50 vista. Now the draw back I can see in hydras aproach is with direct x and open gl support. You see as we all should know each gen of gc's iplement new api and graphics standards.
In light it would be redundant to use lets say a a 7800gt and a 8800gt in union due to the differences of the nature of the designs. so in a fact I would say it is a better choice to keep it to where the direct x versions match. ie dx9-dx9 dx10-dx10 dx11-dx11. It just would not makes sense to hamper the performance of the more advanced dx version just so you can say "look what I can do".
I will add I love the idea of getting my hands nvidia dx11 part to add to my 5870 and hving all the perks of both cuda and stream as well as full dx 11 support with out driver issues and lockouts. (points finger at nvidia)Both companies should embrace this tech and AMD needs to get some boards out with the hydra on them.
And again mixing direct x generations can only be trouble for pefromance. To take full advantage of the tech keep it the same as far as the direct x generation.
This is just fan-bloody-tastic.
Lucid, I'll be drinking to you this evening.
I'm not sure if I need this chip, They should've also included a review of Solitaire and/or minesweeper
This is not multiple systems but the exact some one used by both at Lucid's lab
http://www.brightsideofnews.com/news/2009/11/11/is-lucids-hydra-200-really-ready2c-or-are-they-pulling-a-fermi.aspx
All these years we thought about it. Now It's about freaking time.
Nvidia and ATI Tag-Team Tournament.