Howdy, stranger! Ready to join the community? [log in]

Remember the “SLI-killer” HYDRA chip? It’s back!

In August and December of last year, PC enthusiasts were in a tizzy over the LucidLogix HYDRA chip which allowed for SLI or CrossFire configurations with much better scaling. At the time, it was considered “dead at creation” technology because no mobo maker would knock their SLI/CF kickbacks to give it a try. But the industry has changed quite a bit since then, and so has LucidLogix, which returned to the 2009 IDF with an all-new HYDRA 200 chip.

The HYDRA 200 is a dedicated load-balancer that assumes the burden of splitting a game’s workload amongst the GPUs. Where once you needed an SLI or CrossFire bridge so the cards could talk to one another, now you only need HYDRA. And because HYDRA works outside of the GPU driver model, the chip can offer nearly linear scaling. HYDRA is the end of SLI and CrossFire configurations that only improve by 30% at tremendous resolutions. Instead, HYDRA offers about 90% improvement when pairing two like cards together at any resolution.

Different cards from the same maker? No problem. If one card is only 30% as fast as the lead GPU, total performance is very near to 1.3x the most powerful card. Different GPU manufacturers? Not a problem. While Windows Vista’s driver model limited GPU drivers to one architecture, Windows 7 allows multiple GPU drivers, and so HYDRA now supports mixed-GPU environments.

Yes, that is a GeForce and a Radeon (Image credit: Hexus.net)

Yes, that is a GeForce and a Radeon. (Image credit: Hexus.net)

The whole kit is based on a Tensilica Diamond RISC chip that Lucid has attached to PCI Express lanes. The bigger, badder HYDRA 200 is manufactured by TSMC on a 65nm process and consumes about 6W of power in operation.

The technology isn’t a pipedream this time, either. MSI has signed on to produce a motherboard called the Big Bang based on Intel’s new P55 chipset and it will launch on September 29.

The MSI Big Bang (Image credit: Iopanel.net)

The MSI Big Bang (Image credit: Iopanel.net)

It's the real deal! (Image Credit: Iopanel.net)

It's the real deal! (Image Credit: Iopanel.net)

A shot of the HYDRA 200 chip (Image credit: Iopanel.net)

A shot of the HYDRA 200 chip (Image credit: Iopanel.net)

HYDRA is an amazing technology, but it creates as many problems as it solves: How will AMD and NVIDIA respond to a product that renders their technology redundant? How many GPUs can the HYDRA support? How many mobo makers will break ranks from the SLI/CrossFire ecosystem? How expensive is the chip? Only time will tell, but we’re excited.

Share |

46 Comments:

  1. ardichoke
    King Banana Spanner

    Oh man, this technology makes me drool. Do effing want! If a manufacturer comes out with an AM3 board with a HYDRA chip I WILL upgrade (cost depending). Unless of course the HYDRA chip ONLY works for Intel architecture, you didn't really touch on that aspect of it.

  2. To clarify:

    Any motherboard manufacturer can implement HYDRA for any CPU architecture. Its proliferation depends strictly on the motherboard makers' collective will to shirk convention.

  3. ardichoke
    King Banana Spanner

    Thank you for the clarification sir. I will now return to my regularly scheduled drooling. Followed by an all new episode of WANT!

  4. RWB
    Thank God for Blue Monsters

    Thrax mentioned at the end the one thing I kept thinking... will NVidia and/or ATi try to circumvent this chip by creating future drivers or GPU's that "for some unfortunate reason" doesn't play nice with HYDRA.

  5. Gargoyle
    We can't stop here...

    I wonder if one of them will try to license it so it will benefit them exclusively? It would be a nice selling point for either NVIDIA or ATI to say "our multi-gpu setups scale way better than the competition." In my mind, they were fools not to buy this tech up by now. And I'm glad, because I'll giggle like a schoolgirl if I can get a GeForce and a Radeon to play nice together.

  6. ardichoke
    King Banana Spanner

    Oh man, could you imagine having HYDRA integrated into the northbridge? That might send me off into a full-fledged nerdgasm.

  7. Cliff_Forster
    Keepin it real

    Who seriously could not love that? Even ATI and Nvidia, what potential harm will it do to them? Zero that I can see, it just provided a greater value to having a 2nd card, so its win/win the way I see it.

  8. RWB
    Thank God for Blue Monsters
    Who seriously could not love that? Even ATI and Nvidia, what potential harm will it do to them? Zero that I can see, it just provided a greater value to having a 2nd card, so its win/win the way I see it.

    I'm only speculating, but do they licence out SLI and Crossfire to MB manufacturers? If so and MB decide to go with HYDRA instead they potentially loose out on said licencing.

  9. Yes, SLI and CF is a cross-licensing thing. Mobo and chipset makers pay a fee to get their goods on the support list cooked into drivers, and in return they receive a kickback for driving SLI/CF sales.

    It's a lucrative ecosystem, and HYDRA threatens that. This is potentially a big problem, most of all for LucidLogix, but for the incumbents as well.

  10. UPSLynx
    The Dean of Computer Graphics

    Dangit Thrax, I was going to write this! /writernerdrage

    This tech makes my mouth water. The way it only renders certain aspects of a scene per card to keep the loads manageable and then reconstructs the image is the stuff dreams are made of.

  11. DrLiam
    FoxtoN

    Glad to see HYDRA back! I think this is going to rock the boat.

  12. Obsidian
    Way hotter than Fox n' Bush.

    Good luck getting support from ATI and Nvidia.

  13. Snarkasm
    The Photographer.

    They don't NEED support from ATi or nVidia.

    I am so frigging pumped I can't stand it. My next build must contain Hydra 200 if these claims pan out.

  14. Joe
    Guest

    Well there are two ways to look at this, yet they will loose money on the licenses, but lets be honest, how many people buy a second card for a 30% increase. I think they would sell a lot more for a 90% step up and not having to find the exact same card is great. I am afraid Ati/Nvidia will buy them out or sue them to death.

  15. Komete
    DIY Haxx0r

    Now that it's coming out, I wouldn't be surprised if AMD and nvidia have a solution that'll deliver the same type of performance increase. This has been lurking around for 2 years now, they would have to be idiots to not develope anything. Up until now there just wasn't any sort of incentive to do so. This has the potential to destroy their mid to high range products.

    That said, I doubt amd or nvidia would do a driver update that would disable hydra. They may do something that won't let it work with each others cards. If Nvidia were to somehow disable hydra with a driver, it would be a huge boost to AMD's sales and visa versa. However, if they jointly decide to do something about it, that's a different story.

    I do have some question. Games look differently on a comparable AMD card vs a Nvidia card, how exactly do they work that out if you can mix cards? Also what about dx11 and dx10 card playing a dx11 game? There has to be a catch somehere.

  16. Gargoyle
    We can't stop here...

    I wonder if they (ATI/NVIDIA) might license the chip to put on dual GPU boards? Am I correct in assuming that a dual GPU board is typically about as powerful than two single GPU boards in SLI?

  17. Snarkasm
    The Photographer.

    That's the idea, I believe - something like a 4870X2 is actually just two cores in SLi on a single PCB.

  18. Sort of... Two cards use the SLI/CrossFire bridge (the short dongle that connects the cards). The X2-style cards use an onboard bridge chip that allows for direct communication. The problem, really, is that the load balancing is handled by the drivers... The drivers are bad at it, which is why they're constantly adding profiles and improving existing ones.

  19. Cliff_Forster
    Keepin it real

    So many fascinating possibilities with this. As an enthusiast I would replace my motherboard in a heartbeat. I think its going to be up to us to pressure Nvidia and ATI to do whats right for us on this one.

  20. RWB
    Thank God for Blue Monsters

    This just came to mind but AFAIK ATi doesn't have a Physics PPU like NVidia has. Does Hydra give us complete abilities of both cards? IE playing a game on several monitors using eyefinity, while letting the game use the Physics PPU on an NVidia card? So not only are you getting better performance with the two cards combines but also their benefits that the other card may not have.

  21. lordbean
    404 Brain Not Found

    Oh mah god, I can't believe I missed this article. Does this mean if I buy a 5870 and upgrade to a mobo with a Hydra chip, I can pair it off with my GTX 285 as the secondary card? That would be sweet beyond belief.

    Edit - however expecting the answer to be no. It occurs to me that what I'm considering is a DX11 and a DX10 card. May not work well.

  22. If basic assumptions about mix'n'matched generational parts bears out with HYDRA, then you'd be limited to the oldest DX tech.

  23. (Two concurrent posts because it's not easy to edit on my phone). It also stands to reason that you would indeed be able to use the benefits of both cards as each card's drivers are loaded and, therefore, working to intercept api calls like physics.

  24. AlexDeGruven
    I am Geek. Hear me... type?

    I think, rather than GPU manufacturers trying to license Hydra, one of the major chipset or motherboard vendors needs to jump on this quick.

    If Intel could snag this and be (for a while) the exclusive vendor of Hydra-enabled kit, it would be huge for their MoBo business. Then, when they've gotten a bit of saturation, they can license it out to MSI, ASUS, and the like, reaping cash benefits on both sides of the business.

  25. RWB
    Thank God for Blue Monsters

    On another note, I wonder how long it will take for some unknown person/company to sue over Hydra because they claim it's too similar to their patent. Sorry, but it seems there's always someone to do this, legit or not.

  26. kodemunke
    A Force to Be Reckoned With

    Only advantage I see to this would be to test different gpus on one motherboard. Also, you could use a ATI gpu for graphics and nvidia card for physX but I would never do that. If I am going to use a gpu its going to be either ATI or nVidia... Not both. But that's just me.

  27. Snarkasm
    The Photographer.

    The real advantage to this is getting better performance than a 5850 Crossfire or GTX295 SLi setup with MUCH cheaper cards. If you can pair a couple of 4770s and get actual 1:1 scaling - meaning twice the performance of a single 4770 - the $600 cards start looking a lot less necessary.

  28. damninhell
    Guest

    snarkasm,you make a great point.by the way intel has invested money on hydra.hopefully they're not influencing them not to make chips for amd any time soon.i was planning on getting a asus formula 3 but i'll be saving for an am3 hydra instead.

  29. AMD calls Hydra a "tiny paper dragon." We'll have more information for you on this soon.

  30. Snarkasm
    The Photographer.

    Oh man. I really fucking want a tiny paper dragon now.

  31. ardichoke
    King Banana Spanner

    Is it at least a cute tiny paper dragon? One that would look good on my desk? Something that would tie a room together?

  32. Gargoyle
    We can't stop here...

    AMD can say what they want, but I'm rendering my tiny paper dragon with a GeForce and a Radeon at the same time.

  33. Cliff_Forster
    Keepin it real

    Fermi is what AMD called the "paper dragon"

    We can't nail AMD down for this. Sure, I've read a few things that look like FUD generators, like claiming that NVIDIA is totally abandoning the PC gaming market which even I find hard to believe. Ultimately NVIDIA's lack of transparency is to blame. If NVIDIA would have met the demands of the market and had a DX11 product at Windows 7 launch then this would be a non issue. If NVIDIA would just divulge the details of its new product, then it might only be a temporary issue.

    NVIDIA was not forthcoming at the GPU developers conference where the unenvied Fermi. They lead you to believe that Fermi availability was not far away, and that for certain you would be able to get your hands on product by the holiday. Who here really thinks NVIDIA is going to do more than a paper launch in 2009? Hell, who thinks they will even have product that is widely available by Q2 of 2010?

    Thats why Fermi is a "paper dragon". Its a fair criticism. NVIDIA will waive the promise in front of gamers in hopes that they will hold off on 58xx purchases long enough for them to finally get their act together. If your prepared to wait until the middle of next year to find out if that product delivers, knock yourself out, me, I'll enjoy my 5870 today.

  34. Why are you talking about NVIDIA when everyone is talking about Hydra, a chip made by LucidLogix, a company which is in no way connected to NVIDIA?

  35. mondi
    dot.
    Why are you talking about NVIDIA when everyone is talking about Hydra, a chip made by LucidLogix, a company which is in no way connected to NVIDIA?

  36. Mondi. Have I told you lately how much you rule? You rule.

  37. ardichoke
    King Banana Spanner

    If I'm reading what he said correctly, people above said that AMD called Hydra a paper-dragon. He was pointing out that AMD was actually calling NVIDIA's Fermi the paper-dragon, not hydra.

  38. You read what Cliff said correctly, but AMD also called Hydra a "tiny paper dragon" directly to our face, which is why I said we would have more information soon.

  39. primesuspect
    The Icrontic Guy

    AND HE MEANS OUR FACE

    (..thrax and I share a face)

  40. chrisWhite
    Polygons

    Oh hell's yes, I for see this being the basis for my next system depending on the timing, seriously hot stuff.

  41. Cliff_Forster
    Keepin it real
    You read what Cliff said correctly, but AMD also called Hydra a "tiny paper dragon" directly to our face, which is why I said we would have more information soon.

    Well you gotta understand, I'm not privileged to your face, its a decent one and all....

    I just had seen the marketing materials where AMD had refereed to Fermi as the paper dragon. I guess calling things paper dragons is all the rage at AMD. I do think it applies to Fermi. Hydra I'm not as sure?

  42. Sledgehammer70
    Sledgehammer70

    Nvidia ditch gaming? lawls!!! But to the above comment, who is to say Hydra wouldn't increase the performance of a GTX 295?

  43. Cliff_Forster
    Keepin it real
    Nvidia ditch gaming? lawls!!! But to the above comment, who is to say Hydra wouldn't increase the performance of a GTX 295?

    Judge for yourself.

  44. Gargoyle
    We can't stop here...

    I desperately want that desktop - nay, poster sized. I think I found the original, and it's only 799px wide. If I had money to burn, I'd totally try to clear the rights to it.

  45. ardichoke
    King Banana Spanner

    Give me something that isn't a quote from an AMD rep and I will.

  46. Sledgehammer70
    Sledgehammer70

    My friends who work at NVIDIA say otherwise. I trust them over an AMD exec

Hey, be nice. Icrontic is full of good people, we promise.

New Features on Icrontic: