Lucid Hydra 200: Vendor Agnostic Multi-GPU, Available in 30 Days

[AzraelDarkangel]

http://www.anandtech.com/video/showdoc.aspx?i=3646

A year ago Lucid announced the Hydra 100: a physical chip that could enable hardware multi-GPU without any pesky SLI/Crossfire software, game profiles or anything like that.

At a high level what Lucid's technology does is intercept OpenGL/DirectX commands from the CPU to the GPU and load balance them across any number of GPUs. The final buffers are read back by the Lucid chip and sent to primary GPU for display.

The technology sounds flawless. You don't need to worry about game profiles or driver support, you just add more GPUs and they should be perfectly load balanced. Even more impressive is Lucid's claim that you can mix and match GPUs of different performance levels. For example you could put a GeForce GTX 285 and a GeForce 9800 GTX in parallel and the two would be perfectly load balanced by Lucid's hardware; you'd get a real speedup. Eventually, Lucid will also enable multi-GPU configurations from different vendors (e.g. one NVIDIA GPU + one AMD GPU).

[apt403]

Wow, it's been ages since the first press release came out. I figured this thing would never materialize.

[Chumbucket843]

the rebirth of the coprocessor. you gotta love it.


$36 a chip are you for rizzle?

[G0ldBr1ck]

been saving this quote for quite a while. Im gonna have to break it out now just as he said to do.

You can QUOTE me on this, THIS WILL NEVER BE A SUCCESS. As a matter of fact this will never even be mass produced. I say it will dissapear from the face of this world very shortly.

[Mechromancer]

been saving this quote for quite a while. Im gonna have to break it out now just as he said to do.

It seems GAR underestimated the power of Intel.

Let me be the first to add the obligatory "Now pair this with two AMD 5870 GPUs" remark. Lucid+AMD+Intel = Menage a tois of PWNAGE.

[Xoulz]

Finally..!




Been waiting on more info about the MSI's Big Bang and Lucid's Hydra chip...

What more needs to be said, no more SLI... and this:




From the moment I heard about this technology and Company (15 months ago), I've been a close follower. Just looking at the people who are on the board @ Lucid gave me hope this Hydra chip was for real.

I can't wait for the bench reviews, but this looks pleasing..

[Chumbucket843]

this is multi-gpu blasphemy!!

[cegras]

I wonder if there will be any appreciable latency with an added middleman chip?

[LC_Nab]

Mmmm , No love for AMD boards eh ? T_T

[Mechromancer]

Mmmm , No love for AMD boards eh ? T_T

+1

Lucid has to get these on AMD boards of we'll cry "bloody murder!" Intel has their fingers in Lucid, but I hope they wouldn't deprive AMD users of this man-made miracle.

I don't think computing life will ever be the same after this. Today has been a momentous day in the world of hardware betweem AMD's 5870, Lucid's Hydra, and Intel's IDF. I will sleep with a smile on my face.

[deathman20]

Why would they need 2 VGA connectors hooked up?

BTW Thumbs up. Glad they moved forward with this, and im sure Intel is helping push this all the way as well.

[Computurd]

As i understand lucid is owned now by a subsidiary of intel, or at least funded by them indirectly somehow, which is great, because this is the best thing since the gpu!!!!
My next rig...
sandy bridge
sata 6gb/s ssd
AND lucid hydra, doesnt really matter which gpu now!

[villa1n]

Finally..!




Been waiting on more info about the MSI's Big Bang and Lucid's Hydra chip...

What more needs to be said, no more SLI... and this:




From the moment I heard about this technology and Company (15 months ago), I've been a close follower. Just looking at the people who are on the board @ Lucid gave me hope this Hydra chip was for real.

I can't wait for the bench reviews, but this looks pleasing..

I m confused on the 4X the 8x pcie on a p55... since the cpu only has 16 pci-e lanes dedicated to gfx...but the hydra is issuing instructions via 32 lanes..?

[m0da]

I m confused on the 4X the 8x pcie on a p55... since the cpu only has 16 pci-e lanes dedicated to gfx...but the hydra is issuing instructions via 32 lanes..?

I think this quote from the article will help:

And the highest end solution, the one being used on the MSI board, has a x16 to the CPU and then a configurable pair of x16s to GPUs. You can operate this controller in 4 x8 mode, 1 x16 + 2 x8 or 2 x16. It's all auto sensing and auto-configurable.

The way I read it is that the chip communicates with CPU with a single x16 lane; It subsequently communicates with the GPUs with a 4x8, 2x16, or 1x16 interface.

DAMN THIS IS COOL

[villa1n]

I think this quote from the article will help:



The way I read it is that the chip communicates with CPU with a single x16 lane; It subsequently communicates with the GPUs with a 4x8, 2x16, or 1x16 interface.

DAMN THIS IS COOL

I read that, but thats what i m asking, if it communicates to the cpu at pci-e 16... doesnt that necessarily limit the amount of instructions it can send to the gpus at 16lanes.. ie 1x16, 2x8, or 4x4?

[saaya]

the rebirth of the coprocessor. you gotta love it.

thats not really what this is... its a smart plx chip sortof... but i would not call it a co-processor...

[zalbard]

Personally I don't think this will do any good for using similar cards compared to existing SLI and Crossfire. Using Radeon and GeForce cards at the same time... hmm, dunno, probably not worth it...

[Calmatory]

Nvidia will disable the ability for this in their drivers, sooner or later.

[Xoulz]

I read that, but thats what i m asking, if it communicates to the cpu at pci-e 16... doesnt that necessarily limit the amount of instructions it can send to the gpus at 16lanes.. ie 1x16, 2x8, or 4x4?

You are not going to saturate a PCIe x16 connection with a Hydra chip.


It's own technology talks to as many GPU as you want given those lanes... how the PCIe bus works doesn't change... thats the function of the Hydra chip. Seamless!

[Mechromancer]

Nvidia will disable the ability for this in their drivers, sooner or later.

You're right. Sad thing is if Nvidia does this, they will undermine the adoption of their own tech. It will be a horrible business decision on their part.

[Xoulz]

Nvidia will disable the ability for this in their drivers, sooner or later.

Really bro.. watch a few videos, or read some stuff. It's been mentioned that how it works, is agnostic to the OS and drivers.

Here: http://www.xtremesystems.org/forums/...d.php?t=235138Video

[jmke]

is agnostic to the OS and drivers.

that's what I thought too; there is a software side of things though which might be used to limit

[Vinas]

Just like the disablement of Physix when an ATi card is present in the system... But I hope they don't and if nvidia does I hope they're unsuccessful.

[Kaldor]

You're right. Sad thing is if Nvidia does this, they will undermine the adoption of their own tech. It will be a horrible business decision on their part.

They already did this with crappy SLI chipsets on the 6XX and 7XX mobos. X58 is the best thing that ever happened to Nvidia honestly.

I hope it is a winner, however I buy my graphic cards in pairs. It will be interesting to see if there is any performance hit when using it with 2 of the same cards.

[SKYMTL]

Guys, as has already been stated the Hydra operates independently from the OS and thus is not affected by drivers or at least not much anyways. As such, Nvidia / ATI cannot block it on the software side of things. They may be able to limit it however. At least this is how it works to my understanding.