The future of integrated graphics processors lies somewhere on the dies of future processors, that much is a certainty. However, this creates a big problem for NVIDIA, whose chipset business will be out, of well, business. Beating everybody to the punch, the company announced yesterday that it is ceasing all development on future chipsets, citing unfair business practices from Intel.
It is often rumoured that NVIDIA has a skunkworks project going on in which they are developing their own x86 processor, combined with an on-die graphics processor. Intel and AMD are working on similar chips, whose arrival would mean the beginning of the end of NVIDIA’s chipset business.
For now we don’t know if this project will ever reach the market, but NVIDIA decided that it wouldn’t wait for the inevitable and just quit with the chipset business before they make any major investments in research and development. They are ceasing all development on chipsets; there will be no Nehalem or Core i5/i7 chipsets coming out of NVIDIA. Questions about chipsets for AMD were carefully avoided, so it is likely they’ll back out of that market too. NVIDIA will continue to make the Ion and 9400M chipsets, but future designs are all off the map now.
The big problem is Intel’s DMI bus. Intel claims that NVIDIA does not have a license to this new bus, effectively making it impossible for NVIDIA to develop chipsets for Nehalem and newer Intel processors. In the pre-Nehalem era, such a license wasn’t necessary, but with the newer processor architectures, NVIDIA does need a license. Without the DMI bus, NVIDIA’s PCH replacement wouldn’t be able to talk to the processor. NVIDIA claims it does have a license, and the matter is currently in court.
“Because of Intel’s improper claims to customers and the market that we aren’t licensed to the new DMI bus and its unfair business tactics, it is effectively impossible for us to market chipsets for future CPUs,” NVIDIA writes, “So, until we resolve this matter in court next year, we’ll postpone further chipset investments for Intel DMI CPUs.”
As Jon Stokes over at Ars points out, while it is of course convenient to place the blame on Intel (and it probably does belong there at least partially) a bigger issue is probably the future of IGPs, as mentioned above. IGPs are moving to the processor die, which means that even if NVIDIA does gain a license, its chipsets would have a lifespan of only a few years, because of the arrival of CPU/GPU combo chips. Add to that the uncertainty of NVIDIA gaining a license at all, and it only makes sense to back out now, before any investments are made into research and development.
Apple is in an interesting position here. The Cupertino company has more or less standardised on using Intel processors with NVIDIA chipsets (the 9400M), so it will be interesting to see what Apple will use for future Macs. I’m not familiar with Intel’s roadmaps, but I find it hard to believe that Intel’s IGP offerings will be as competitive performance-wise as NVIDIA’s offerings would be, and on top of that, does Intel’s roadmap IGPs say anything about OpenCL support?
I do hope that these tensions between Intel and NVIDIA subside, because I would really welcome NVIDIA entering the x86 market with their own processors with integrated graphics chips. The consumer processor market is dominated by Intel and AMD, and more competition is always welcome in this space. VIA is having troubles getting any serious foothold, so maybe NVIDIA’s brand recognition can do the trick here.
For Apple it’s more convenient to move to AMD CPUs now, to continue using nVidia or ATI (AMD) GPUs, which are way more powerful than Intel’s.
This nVidia move actually helps ATI more than it damages Intel.