I do wish this thing support multi GPU-enabled Super Sampling AA like GeForce's SLI AA or Radeon's Super AA.

http://techreport.com/forums/viewforum.php?f=3

This looks like another 3Dfx style company! Will keep an eye on them.

Good work lads!

Woaw...

SLI and Crossfire done right?!... errr... multiple GPU's done right

This looks pretty sweet. Even though they showed this up and running at IDF it doesn't answer the question of if it's comparable to the actual manufacturers GPU schemes. It should be, but...

Few questions are - Will this have issues with talking to another bridge chip? IE linking Hydras for a almost unlimited amount of GPUs through a theoretical hierarchical tiered scheme.

Why exactly can't this work with both AMD and NVidia cards? I know it was mentioned in here that it was due in part because of drivers issues, but since it works on the bit level that really shouldn't matter... should it? That said, AMD and NVidia cards both have their ups and downs in certain areas... using one of each could provide a optimal setup to be all round best in all areas.

Exactly how old are the cards that you can use with this thing before the second card actually starts hindering the better one? Rx4870 and Rx800...

How smart is the card? Do the cards need to support the same DX level or will it actually feed different parts of detail level to different cards? So you can have a DX9 card and a DX10 card working together and the DX10 card will render all the DX10 content and the DX9 will try to help on everything else.

Does this card employ any sort of methods to appear seamless to GPUs to avoid being blacklisted by video card developers (besides taking the approach that it isn't detectable now)? If I was AMD or NVidia I would see this card as a huge threat, not to be negative... but I don't doubt that they will do that in the future. Especially if this could eventually end up combining AMD and NVidia cards to work together.

Could this technology potentially use system memory to supplemental graphical memory to a older graphics card to keep up with newer ones? Turbomemory has been around for awhile, being smartly implemented to help aid a crippled graphics card or help overall performance could be extremely beneficial. Of course this once again depends on how smart the chip is.

While impressive, I doubt this is going to end in a product launch. I'm pretty sure Nvidia will disable this functionality on their drivers just like they killed SLI on Intel chipsets ages ago. Or as other suggested, they could also buy/sue them and be done with it.

Also, I have no idea how is this going to interact with PhysX functionality, so it could be that you would have to sacrifice one for the other.

YAY, now i can haul two boxes around instead of just one. Neat idea, but I'm not entirely sold on having a separate box just for my video cards.

Why would a PCIe Gen 1 accelerator with enhanced performance be attractive when most users and suppliers have already moved to Gen 2?

The moment you've been waiting for, Crysis at very high framerates.

So it is double on crysis. TR reviewed the 9800 GTX in SLI at 1920x1200 at average 35 fps with low of 23. Yet at the demo of Lucid's chip, you guys observed 40-60 fps?

Here's the review:
http://www.techreport.com/articles.x/14524/9

Here's the TR quote from the demo:
"we got a demo of Crysis running at 1920x1200 at the highest quality levels available in DirectX 9. The test system was using a pair of GeForce 9800 GTX cards, and performance ranged between 40 and 60 FPS on the game's built-in frame rate counter."

I'm really curious to see whether Nvidia whips out a patent infringement suit over this. Major aspects of what this chip is doing are almost certainly comparable to what 3dfx was attempting to do with VSA-100, and Nvidia is now:

1) Holding those patents.
2) Confronted by a product that could break their vendor lock-in on SLI.

They almost certainly have lawyers looking at this one, and it would be surprising if they do absolutely nothing.

now the game developers can code for 3d

AMD shoul buy this tech, and stick the lucid chip on its 4870 x2 cards instead of whatever they use now. Could even get another GPU or 2 on there too with the way this tech works :)

My bad, should read all posts before commenting

This technology sounds like an outgrowth of work done in 1993 at UNC-CH called PixelPlanes 5.

Very exciting tech.

Scott, any word on use for laptops?

This is really what Multi-GPU setups have needed. I want to be able to combine my old 7600 GT with a new card and have perfect scaling. Is this technology going to find its way into any boards any time soon?

a) Single-slot 9800 GTX's with reversed coolers?

b) How much is this going to cost? (I'm guessing exorbitant amounts for the mobo, and am curious about how ATi can implement this instead of their PLX chip).

About the RISC core:
http://tensilica.blogspot.com/2006/04/lucid-licenses-diamond-212gp....

Does this card employ any sort of methods to appear seamless to GPUs to avoid being blacklisted by video card developers (besides taking the approach that it isn't detectable now)? If I was AMD or NVidia I would see this card as a huge threat, not to be negative...

Why would nVidia or AMD care? All this thing means is that people will have a bigger incentive to buy more GPUs because now it won't matter what MB you have or if your buying a matching card.

Now what would be nice is if this Hydra chip could somehow manage power to the video cards as well, and maybe power one off completely when not in 3d mode. Or if they could get this thing to work with a MB integrated (or integrated into the Hydra, would it matter where it is?) chip so that it could power down both high end cards.

I see a lot of potential for this tech. Good work Lucid.

one of the more interesting things this chip raises is what happens next
specifically in regards to things like this:
http://www.theinquirer.net/en/inquirer/news/2007/06/09/nvidia-denie...

(non-charlie article for those wondering)

any chipset maker would not need to license sli nor crossfire
which then represents a loss of income stream for amd/nvidia
why pay a license for sli & crossfire when you can buy one chip instead?
intel no longer would ned to pay for crossfire techfrom amd
amd drops crossfire r&d, gets sli and deliver a platform like intel
nvidia is the loser in this one as SLI is a draw-card for them
therefore expect to see nvidia pay a huge sum to buyout lucid

Does anyone know if their chip is hard silicon or an FPGA?

All I have to say is that if this works as advertised then it is very impressive. I find it very hard to believe that they can so easily break up API calls and then put it all back together in a short enough period of time as to not slow down the game. As games get more complex and have more and more calls to the cards, the hydra chip has to at some point be a limiting factor. I am also surprised that they could pull this off with ATI or Nvidia's help considering it seems that hydra chip has to essentially have their drivers on it. There is definitely room here for some copyright issues. I am very excited however!

I fear, one of the big players like Intel, Nvidia or AMD might buy them out...

Unless this is smoke and mirrors (it doesn't seem to be), this seems like the next logical development in GPU technology.

If we compare this idea to multiple CPU cores, this seems more like running two instances of one program, or two separate programs, which is easy, compared to running one program that uses multiple threads to accomplish one task, which is much harder.

Of course all types of load balancing are happening in a modern system, but AFAIK, software that starts out with multi-threaded logic is much more likely to be able to use 100% of available execution resources than trying to split a work load later on. Lucid seems to have caught onto the idea that if the work load is split an earlier stage, better and more _consistent_ results can be achieved.

That's why this is probably a better solution than trying to "split" what boils down to hardware instructions once the API calls are already made, which is what SLI drivers need to do; it's better to make multiple API calls to separate devices via the driver, and then combine the results.

But if this really does work, how long until something similar becomes part of the API itself? I mean, if the API calls can be intercepted, and re-interpreted into two work loads, it makes sense that the API should be able to deal with multiple GPUs in the first place.

With the explosion of multi-core CPUs, general purpose software (should be) starting to be designed with this in mind. Up to now, since 3d graphics is so easily parallelizable (is that a word?) in hardware no one has really though about multiple GPUs as anything other than a niche market.

But sooner or later, it seems transistor density, and therefore die size (NVIDIA was rumoured to already be facing yield issues because of die size), the GPU computing market will have to face the "multicore" reality more seriously...

AMD is probably already on the right path with developing smaller, cheaper cores, and slapping them on the same board that they still can sell relatively cheaply. One issue here is the extra memory resources needed, but memory, even graphics memory continues to both increase in density and fall in price (for now)... What turned out to be a better design, Netburst or Core2 ?

But I digress; Lucid's approach might be GPU agnostic, but it is not API agnostic. The biggest challenge Lucid probably faces is API changes, which in the case of Microsoft, are not altogether infrequent, although this could very well slow as the API becomes more flexible, allowing for greater and greater flexibility by virtue of shader programs. The hardware and driver/software of their solution has to be able to deal with every single possible API call without exception. I'm not a developer, but this seems that this will get very complicated. I guess the advantage is that if the RISC chip that powers this is powerful enough, and flexible enough, the software end of it can be tweaked to do whatever it needs to do.

Here's hoping! (and I think that was my longest comment ever; give my regards to WaltC)

Hmm I've got. 1 1900xt, 2 2600xt, 2 3870's, and 1 4850. I wonder how Crysis will run with these six video cards.

From looking at their patent ( http://www.google.com/patents?id=TE6AAAAAEBAJ ) it looks like instead of splitting up the screen by pixels, they do it by depth (BSP tree?).

By that logic, I can see these advantages compared to SLI/CF:

- The chip looks to be able to divide up geometry by depth, so not all of them has to be duplicated, and T&L tasks can also be split up.
- There is an speed up of Z-lookups since there are fewer depth levels
- Fewer pixels have to drawn since there is an early Z-test

And these disadvantages:

- Pixel shaders effects that read back framebuffer values may not work properly if the needed pixels are in the other graphics card's memory.
- Edge pixels may not blend perfectly
- Much of the Z logic are already present in existing graphics cards, and would be wasted.

Intel finance behind it huh....wonder how much the Larabee team knows about it and whether it might be part of their secret sauce....?

btw, I'm still waiting to hear UberGerbil's take on this.

If this thing really works as good as a lot of people have said they are going to sell a huge amount of them. Slap it in the middle of a 4870X2 card instead of the pci-e switch and the performance would be fantastic. Buy a motherboard with the chip built in and dont worry about wonky SLI and Crossfire problems.

Crysis benchmarks or bust

I wonder if this could be CUDA aware too? That would be really, really cool.

brb, gotta change my pants