The rumour is that Nvidia really really
doesn't like Lucid's Hydra chip - you remember, it's load balancing chip which, when present on a motherboard, allows you to mix and match graphics cards in a multi-GPU system.
The reason is obvious: Lucid is making something that could clearly damage Nvidia's SLI business model, so it's within reason that it would be miffed and it's not unlike Nvidia to protect itself with software DriverIDs and whatnot as has been evident from the recent kerfuffle
over Batman Arkham Asylum
Of course there's more than one side to every story, and we've recently spent time trying to find out what's happening with the Hydra chip and whether there's any truth to the rumours that Nvidia is playing a part by causing a stir with motherboard manufacturers.
We spoke to several Taiwanese motherboard manufacturers recently about whether they are still considering/making a motherboard using the Hydra chip. We got some interesting answers, but only on the condition of anonymity.
Company number 1 said it wasn't, alluding to the fact it had investigated the performance and found it to be poor, and that the design still required a lot of work. Another company it would, but not until next year at the earliest - although this response was more positive than when we had first asked several weeks before.
Finally, the only company that has actively shown off a Hydra board so far: MSI. It's still working on its board but wouldn't give us a date for launch, stating "[MSI] is serious about the quality and making drivers which can mix & match is difficult and will take time."
Having told us just this, we then saw this article
from the Tech Report that details an early look of Lucid Hydra performance. We promptly then received a reply from MSI, essentially washing their hands of it:
"Last week, we confirmed with Lucid that it will run a B2B event introducing the latest Hydra Chip specifically without mentioning MSI Big Bang-Fuzion [motherboard]. However, the recent article shows the company didn’t follow what was promised. At the moment, we are seeking an explanation."
MSI went on to explain the Big Bang Fusion motherboard is currently under a performance NDA between Lucid and MSI;
"The benching result is not tested using an MSI Big Bang-Fuzion motherboard. Lucid used its own engineering sample motherboard to show its [Lucid Hyrdra 200] performance.
However, we consider the picture displaying MSI's Big Bang-Fuzion in the same article as performance figures, is a violation of NDA between MSI and Lucid.
The pictures showing the MSI Big Bang-Fuzion are not the final version and will be different when the board becomes retail. The motherboard shown from MSI has been lent to Lucid for internal testing only."
MSI also stated that because of this delay with its Big Bang-Fusion motherboard, it decided to push forward its Intel P55/nForce 200 Big Bang motherboard to fill the gap. When questioned by bit-tech
, both MSI and Nvidia separately
stressed that neither had heard from or was pushed by the other with regards to this Nvidia board.
Despite the obvious tentativeness in commitment to products from everyone we asked some of the manufacturers we spoke to had claimed that (an) Nvidia rep had called and ranted at them in regards to future Lucid products.
We were told by multiple sources that someone in Nvidia was certainly not happy, although none would go into specifics: whether it was Nvidia having a go at the motherboard manufacturers for making Lucid Hydra products, or the fact they were offering Lucid engineering support.
When we talked to the UK Nvidia office, it claimed it was obviously going to protect its own investments and business model, which includes SLI certification, but it stated outright that it has/will never actually dissuading anyone from using Lucid's products. Nvidia claimed Lucid would never have a product that will threaten its product line, given the amount of cross-platform debugging it would need.
Given that our friends in Taiwan have no reason to lie about the calls from Nvidia, we suspect the UK branch is unaware of what other parts of Nvidia might be angry about. Without further evidence either way, bit-tech
cannot further comment on this.
This isn't to say we think Nvidia is going to actively block the Hydra - in fact, it's doubtful whether it even needs to, given the difficult nature of the task Lucid is attempting. Still, it's fairly obvious why Nvidia isn't going to go out of its way to help the Hydra ship. Lucid's vision of multi-GPU is very different to Nvidia's, and as such, it's got an uphill struggle ahead of it.