Friday, November 6th 2009

NVIDIA Shuns Lucid Hydra

A promising new technology from LucidLogix, the Hydra, has perhaps hit its biggest roadblock. The Hydra multi-GPU engine allows vendor-neutral and model-neutral GPU performance upscaling, without adhering to proprietary technologies such as NVIDIA SLI or ATI CrossfireX. NVIDIA, which is staring at a bleak future for its chipset division, is licensing the SLI technology to motherboard vendors who want to use it on socket LGA-1366 and LGA-1156 motherboards, since Intel is the only chipset vendor. On other sockets such as LGA-775 and AM3, however, NVIDIA continues to have chipsets that bring with them the incentive of SLI technology support. NVIDIA's licensing deals with motherboard vendors are particularly noteworthy. For socket LGA-1366 motherboards that are based on Intel's X58 Express chipset, NVIDIA charges a fee of US $5 per unit sold, to let it support SLI. Alternatively, motherboard vendors can opt for NVIDIA's nForce 200 bridge chip, which allows vendors to offer full-bandwidth 3-way SLI on some high-end models. For the socket LGA-1156 platform currently driven by Intel's P55 Express chipset, the fee is lower, at US $3 per unit sold.

The Lucid Hydra engine by design is vendor-neutral. It provides a sort of abstraction-layer between the OS and the GPUs, and uses the available graphics processing resources to upscale resulting performance. This effectively kills NVIDIA's cut, as motherboard vendors needn't have the SLI license, and that users of Hydra won't be using SLI or Crossfire anymore. Perhaps fearing a loss of revenue, NVIDIA is working on its drivers to ensure that its GeForce GPUs don't work on platforms that use Hydra. Perhaps this also ensures "quality control, and compatibility", since if the customer isn't satisfied with the quality and performance of Hydra, NVIDIA for one, could end up in the bad books. This could then also kick up warranty issues, and product returns.

MSI has the industry's first release-grade motherboard, the Big Bang Fuzion P55 that uses Hydra to power multiple GPUs, while also allowing users to mix and match various PCI-Express GPUs to suit their needs, something new particularly for NVIDIA users. Earlier expected to be announced around this time, MSI's Big Bang Fuzion, as it is called by its maker, has been indefinitely delayed up to Q1 2010. Apparently to fill the void created by months of hype, MSI rushed in its cousin, a similar-looking motherboard, that uses the nForce 200 chip, to provide 3-way SLI support, called the Big Bang Trinergy P55, which will stay on as the company's top offering for the P55 platform. One can only hope that Hydra doesn't end up stillborn because of corporate strategy by much larger companies.
Source: Overclock3D.Net
Add your own comment

230 Comments on NVIDIA Shuns Lucid Hydra

#26
FordGT90Concept
"I go fast!1!11!1!"
InnocentCriminalHe is right, the company is called nVIDIA.
NVIDIA is NVIDIA in all their legal documents. I haven't seen NVIDIA say what it actually stands for or what their inspiration was for the name (if there was any).
Benetanegia- From what I understand Hydra takes DirectX calls and converts them into smaller instructions that the GPUs can understand. As I see it that pretty much bypasses most of the vendor or GPU specific driver work. Where do driver level optimizations go? At least 25% of todays performance comes from GPU and game specific driver level optimizatios. Is Hydra really able to take advantage of those optimizations, when it purposedly breaks original DX calls and breaks them up into smaller pieces? Or are they going to make their own optimizations (do they have the manpower to do so)?
Hydra sits on the HAL--above drivers. What NVIDIA is going to do is search for the Hydra HAL and, if found, basically brick the NVIDIA GPUs just for the sake of not letting anyone use NVIDIA cards on a Hydra platform.

As far as normal GPU operation is concerned, Hydra never existed. Hydra is transparent being on the HAL.
Posted on Reply
#27
btarunr
Editor & Senior Moderator
mechtechYes, but I mean you cant crossfire 2 ATI cards in a Nvidia chip mobo, can you??
Nobody is going to make nForce 790i SLI motherboards all over again, just with Hydra in place. I am fairly sure about that. nForce 700i, 700a are circa 2007. nForce 980a is a very small market. Only two motherboards so far. Nobody will bother giving it Hydra.
Posted on Reply
#28
theubersmurf
BenetanegiaYou saying that is just hilarious. :roll:

- Do we know if Hydra really works as well as they say it works? What they showna year ago or so was not very promising IMO. If SLI is significantly faster, I can understand Nvidia disallowing this for their cards. Yes Hydra allows Nvidia + Ati, but if it comes at the cost of not working as well as SLI/Crossfire when both cards are the same, Nvdia is in their right to not allow it on their cards, specially if <read next 2 questions>

- Has Lucid worked with Nvidia/AMD in order to improve compatibility or make optimizations? Have they shared key technology info, so that optimizations can be made from both sides? IMO no, they don't. They probably don't want to share it with them, because that could let them learn the technology and implement it by themselves.

- From what I understand Hydra takes DirectX calls and converts them into smaller instructions that the GPUs can understand. As I see it that pretty much bypasses most of the vendor or GPU specific driver work. Where do driver level optimizations go? At least 25% of todays performance comes from GPU and game specific driver level optimizatios. Is Hydra really able to take advantage of those optimizations, when it purposedly breaks original DX calls and breaks them up into smaller pieces? Or are they going to make their own optimizations (do they have the manpower to do so)?
-part one doesn't matter, if the consumer is willing to sacrifice performance (and I doubt that's the issue for invidia) for a product that serves a variety of functions, it's his choice. Same for part 2.
Posted on Reply
#29
newtekie1
Semi-Retired Folder
Two propriatary technologies replaced with one...I don't really care...let them fight it out.

And picking the Hydra chip might save the motherboard manufacturers $3 or $5 per board on SLi licencing, but how much does the Hydra chip cost? I'm betting more than $5.

I wish the two companies would just quite their petty crap, and open both Crossfire and SLi for all platforms, without licencing. If your motherboard can support Two video card, you can run either, that is how it should be. This crap with nVidia locking it to only their chipsets, and AMD locking it to only their chipsets(and Intel's on the Intel side) is idiotic.

The Hydra technology is cool because you can mix difference graphics cards, but I'm betting the overhead of doing this will be horrible, and the performance gains won't be anything near true Crossfire and SLi.
FordGT90ConceptI hope someone has the guts (and resources) to sue NVIDIA for their practices in regards to SLI and PhysX. Their selfish behavior has to stop, now.
PhysX I understand where you are coming from, even though I also understand where nVidia is coming from in that it was ATi's fault for not allowing PhysX to run natively on it's hardware. But SLi? You're kidding right?
Posted on Reply
#30
wolf
Performance Enthusiast
btarunrPerhaps fearing a loss of revenue
Perhaps Indeed, there is a lot of perhaps surrounding this debate ;)

Let me guess, this furthers peoples opinion that Nvidia are evil, or employ underhanded tactics? it's business, and for the most part, they do it well, despite a series of blunders over the past 12-18 months.
Posted on Reply
#31
Black Hades
btarunr[...] Perhaps fearing a loss of revenue, NVIDIA is working on its drivers to ensure that its GeForce GPUs don’t work on platforms that use Hydra. Perhaps this also ensures "quality control, and compatibility", since if the customer isn't satisfied with the quality and performance of Hydra, NVIDIA for one, could end up in the bad books. This could then also kick up warranty issues, and product returns.[...]
If this was only about quality control they'd would just set a Disclaimer and Warning notifying the user that: Any problem encountered while using this Hydra Lucid is in no way responsibility of Nvidia, and not covered by warranty or technical service from their part. This would be the wise and sensible approach. What do they do instead? block it altogether.

No, this is about their revenue almost entirely. This is my opinion.

I rest my case.
Posted on Reply
#32
theubersmurf
Black HadesIf this was only about quality control they'd would just set a Disclaimer and Warning notifying the user that: Any problem encountered while using this Hydra Lucid is in no way responsibility of Nvidia, and not covered by warranty or technical service from their part. This would be the wise and sensible approach. What do they do instead? block it altogether.

No, this is about their revenue almost entirely. This is my opinion.

I rest my case.
agreed
Posted on Reply
#33
Benetanegia
FordGT90ConceptHydra sits on the HAL--above drivers. What NVIDIA is going to do is search for the Hydra HAL and, if found, basically brick the NVIDIA GPUs just for the sake of not letting anyone use NVIDIA cards on a Hydra platform.
I know what they are doing. Even if (should I say "because", in fact) they work above driver level, they break up the normal execution of the game into smaller pieces that are then split across all the GPUs available (the fact that any GPU combination works is based on this fact). That breaks any optimizations that are made in driver level for the complex instructions and specially those that are specific for the games.
As far as normal GPU operation is concerned, Hydra never existed. Hydra is transparent being on the HAL.
As I said it's transparent as far as execution goes, but it changes completely the work that the GPUs are doing.
theubersmurf-part one doesn't matter, if the consumer is willing to sacrifice performance (and I doubt that's the issue for invidia) for a product that serves a variety of functions, it's his choice. Same for part 2.
It might not be a concern for an informed user, but Nvidia might be concerned about the uninformed user who would only be able to see that two Nvidia cards are not performing as they should. Uninformed people is 90% of users so it is a big concern for them, in fact.

Extension to that argument: If performance is worse than SLI/Crossfire overall, what's the point of putting a HD5870 and GTX285 together using Hydra (assuming you had a GTX285) when it will perform the same as GTX285 SLI on a normal X58 board? Why not just buy another GTX285 or sell your GTX285 and get two HD5850 cards and enjoy a real improvement?
Posted on Reply
#34
Black Hades
Yes I agree this is business, cutthroat environment. They have to fight to survive. But it's an old saying that it's not the strongest of a kind to survive, nor the smartest. It's the one more capable to adapt the current situation.

Their products are excellent, that's good but their business scheme is hostile and it'll draw unwanted attention given time. They should refocus on what they do best and that's video cards not c*ckblocking everyone and losing some customers in the process.
Posted on Reply
#35
Benetanegia
And yes, this is exactly the same as AMD refusing to let PhysX run on their hardware. Nvidia was probably not willing to share key code information with them so that optimizations could be made in both sides, so AMD refused to support it.

Lucid is not willing to share their key technology so...

Exactly the same. Plus Nvidia is not blocking Hydra technology, they are just not allowing it to run with their hardware. I'm going to use the mantra of AMD and say that until Lucid makes Hydra an open standard, I'm not concerned wether Nvidia blocks or doesn't block them. I have never liked multi-GPU anyway.
Posted on Reply
#36
theubersmurf
BenetanegiaIt might not be a concern for an informed user, but Nvidia might be concerned about the uninformed user who would only be able to see that two Nvidia cards are not performing as they should. Uninformed people is 90% of users so it is a big concern for them, in fact.

Extension to that argument: If performance is worse than SLI/Crossfire overall, what's the point of putting a HD5870 and GTX285 together using Hydra (assuming you had a GTX285) when it will perform the same as GTX285 SLI on a normal X58 board? Why not just buy another GTX285 or sell your GTX285 and get two HD5850 cards and enjoy a real improvement?
I'm not sure what the second point is here, but when I read the first part, it seems like an attempt to rationalize their choices. If driver optimization is a problem, curtailing compatibility, prior to any effort to work around that problem, is a poor choice. The effort put into blocking compatibility could be spent on a workaround or real solution to the problem.
Posted on Reply
#37
theubersmurf
BenetanegiaAnd yes, this is exactly the same as AMD refusing to let PhysX run on their hardware. Nvidia was probably not willing to share key code information with them so that optimizations could be made in both sides, so AMD refused to support it.

Lucid is not willing to share their key technology so...

Exactly the same. Plus Nvidia is not blocking Hydra technology, they are just not allowing it to run with their hardware. I'm going to use the mantra of AMD and say that until Lucid makes Hydra an open standard, I'm not concerned wether Nvidia blocks or doesn't block them. I have never liked multi-GPU anyway.
No it isn't, licensing from their competitors is not the same as blocking a non-party specific piece of hardware. Lucid does own their chip, but neither AMD nor invidia are forced to license directly from one another.
Posted on Reply
#38
WarEagleAU
Bird of Prey
Didn't read all responses, but Nvidia needs to get over it. Of course I understand them wanting to protect their profit, especially if their Chipsets are suffering, but they are selling a bunch of GPUs. Really Nvidia...
Posted on Reply
#40
wolf
Performance Enthusiast
I feel it is the same, AMD said NO, Nvidia say NO, companies say no when they don't want to. We need to get over it.

and can we please refer to Nvidia by their real name? I mean come on.

Or have we descended into name calling now?
Posted on Reply
#41
Black Hades
Benetanegia[...]Plus Nvidia is not blocking Hydra technology, they are just not allowing it to run with their hardware.
My english may fail me but. "Blocking" and "not allowing" is kind of the same in this context since there are currently 2 major players. Lucid Hydra with Nvidia cards barred... may as well be Crossfire

If X is SLI and Y is Crossfire and lucid hydra is XY then X+Y-X=Y so.. :p
Benetanegia[...]I'm going to use the mantra of AMD and say that until Lucid makes Hydra an open standard, I'm not concerned wether Nvidia blocks or doesn't block them. I have never liked multi-GPU anyway.
Yes it's an great view point on the situation. Now wouldn't it be hilarious if AMD did the same thing and bar their hardware on the Lucid Hydra?:laugh:
Posted on Reply
#42
theubersmurf
wolfI feel it is the same, AMD said NO, Nvidia say NO, companies say no when they don't want to. We need to get over it.

and can we please refer to Nvidia by their real name? I mean come on.

Or have we descended into name calling now?
are you referring to my use of invidia in place of nvidia? How would that be name calling? I'm curious.
Posted on Reply
#43
wolf
Performance Enthusiast
Its not their name? hello!
theubersmurfI prefer it the other way.
that's cool but its not their name.
Posted on Reply
#44
pantherx12
And Joe isn't my name and yet people call me it instead of Joseph.

I'd hardly call that name calling.

Name calling implies malicious/hurtful intentions.
Posted on Reply
#45
wolf
Performance Enthusiast
I'll drop it don't worry, call them what you like, it's not worth an argument.

In any case, this move seems typical of their attitude, and I say good on em.

my 2 cents.
Posted on Reply
#46
btarunr
Editor & Senior Moderator
Black HadesIf this was only about quality control they'd would just set a Disclaimer and Warning notifying the user that: Any problem encountered while using this Hydra Lucid is in no way responsibility of Nvidia, and not covered by warranty or technical service from their part. This would be the wise and sensible approach. What do they do instead? block it altogether.

No, this is about their revenue almost entirely. This is my opinion.

I rest my case.
Which is why I used quotes. The Quality Control BS is what we've been fed, spoonful shovelful, with the recent PhysX cards on systems with ATI incident. It's a great coverup for the uninitiated.
Posted on Reply
#48
theubersmurf
wolfIts not their name? hello!



that's cool but its not their name.
I'm not sure where the name comes from if not invidia, I'm not going to talk about this with you if you're so sure it's derrogatory. If you came to that conclusion based on my ill will regarding this issue, you're entirely mistaken.





If you look at the logo, the eye is the I in invidia. That's solely where it comes from.
Posted on Reply
#49
InnocentCriminal
Resident Grammar Amender
It's not an "i" it's an eye, relating to vision et al. You've got good grounds to think it's inspired by invidia and honestly I wouldn't be surprised if it was an influence, however - nVIDIA are called nVIDIA end of.
Posted on Reply
#50
wolf
Performance Enthusiast
theubersmurfthe eye in the nvidia logo is an "I" invidia means envy in latin.
Thats the only reason I saw ill will, I dont quite see what they would be envious of, I read into that - jealousy.

anyway I'm over it and yeah its settled.
Posted on Reply
Add your own comment
May 2nd, 2024 01:48 EDT change timezone

New Forum Posts

Popular Reviews

Controversial News Posts