Intel FTC Suit Settled - A Peek At The Future

The Intel FTC suit was announced a while back to be in its final steps towards an agreement(please see Intel Settles Lawsuit, Again) and that deal was announced yesterday. The settlement involves no monetary compensation for anyone but does bring a good amount of restrictions on business practices for years to come.

Intel

After some reading(PDFs here), I found the following points more interesting, mainly because some of them are also mostly redundant to the lay person:

• Must provide an interface to it's processors based on the PCI bus for the next 6 years.
• May not provide discounts based on an target share from the OEM's product portfolio.
• May not provide discounts to OEMs if they only buy from Intel(including chipsets).
• Can't provide discounts for more than a year.
• Can set a minimum number of parts for a given lower price offer.
• Can't offer a customer or end user more than one free part per ten purchased to an OEM(can't offer 10k in exchange for a deal of 100k parts). This will then only apply to the end user directly and not OEMs.
• Intel may not design Intel compilers to optimize code for SSE instruction sets exclusively for Intel CPUs:

• If compatible CPUs also support the same instruction sets.
• Must notify the programmer using the compiler if it does not optimize for those CPUs.

Not entirely unexpected points and the compiler portion of the settlement was a nice addition to the whole settlement. There's also some orders mentioning Intel being allowed to change the model number of a product based on what chipsets the OEM chooses to use other than Intel's ones. This makes no sense but the "model number" reference may only mean something like the Centrino branding.

VIA

Had it's cross licensing agreement extended but doesn't mention explicitly whether or not Intel must grant access to DMI and QPI bus licenses - the agreement only covered FSB connect chipset licenses. Also, Intel has retained a right to disallow VIA to build x86 processors that are pin compatible with Intel's platforms.

NVIDIA

I don't expect Nvidia to benefit much because there's nothing I can read into that suggests they might start building chipsets again. Nvidia had a patent cross license agreement with Intel back in 2004 and that has been extended in the settlement. The problem with that particular agreement is that Intel had only granted access to processors that use a front side bus(please see this link from Nvidia), not any who uses DMI or QPI - I don't expect Nvidia's dire situation regarding chipsets to change just from this settlement. Looking back into the issue:

There are reasons for this, other than chipset licences Intel might or not want to concede.
First, there's the LGA 1156 interconnect to the PCH, the new name that Intel gave to the chipset, which stands for Platform Controller Hub and is connected to the CPU via DMI, which is essentially how all Intel ICH southbridges had connected to the northbridge until the LGA 1366 came along and brought a single chipset and QPI. The new P55 chipset has two connections to the CPU: DMI and FDI:


FDI stands for Flexible Display Interface, which is needed for "Clarkdale" based processors that feature an integrated GPU, called iGFX.
The DMI bus is something rather narrow, which apparently delivers only 2GB/s, or 1GB/s in each way. Some diagrams have 4 lanes drawn with the DMI word on it, which translates to 2GB/s as DMI is reportedly PCIe v1.1 derived. These four lanes, at 500MB/s (upstream+downstream) that the PCIe v1.1 calls four, delivers the aggregated 2GB/s.

The old FSB architecture is better well known: the northbridge connected through a wide enough FSB to the processor and had the memory controller attached to it. Typical FSBs for Core 2 range from 1066MT/s to 1333MT/s, on a 64bit wide bus. This translates to a one way bandwidth of 8.5GB/s or 10.6GB/s in case of the 1333MT/s(or 1333MHz) of the FSB. But the memory controller has moved to the CPU and the FSB just doesn't cut it for multi processor platforms, hence QPI and HyperTransport.

The old architecture also had one plus going for it, when the chipset had an integrated graphics core, it could have access not to just 10.6GB/s but to 21.3GB/s, when paired with two DIMMs of DDR3-1333. That was plenty for Nvidia to strap a 9400 core on a chipset and call it the GeForce 9400M. It is slower than the 9400GT but not by much, which only has access to 12.8GB/s in reference designs. The chipset has to share bandwidth with the CPU, so performance ends up similar.

Now, if you're Nvidia, where do you put your integrated graphics core on the LGA 1156 platform? You can't.
If you put the graphics core on the PCH, you only have 1GB/s access to main memory, shared with six 3Gbps SATA ports, 14 USB 2.0 ports and a PCIe x4 interface. I don't remember the last time graphics processors had so little bandwidth to their local framebuffer. Don't believe me? Dial your AMD's HT bus down to 200MHz(which would deliver 800MB/s) and see the slideshow for yourself.
AMD also has to go over the HT bus, but since it's way wider right now - 8GB/s in current processors - there's still some bandwidth there.

There's still an ongoing Intel vs Nvidia(please see Nvidia going to court with Intel) suit regarding the chipset licenses, which will be the one to shed light on that situation.
With that in mind, see:

This could imply that the usage of the DMI bus in LGA 1156 and the latest Atom processors was a way to engineer Nvidia out from the chipset game(mainly the integrated GPU chipset game), but I don't read it that way. Intel played the DMI card well and I sure hope I'm wrong about what I'm reading from this particular point. Either way, I don't see Nvidia bringing a chipset before "Sandy Bridge" is out, probably even only in the "tick" cycle of that architecture.

Nice, but having product roadmaps available, aside from PCI Express interfaces, I only see a benefit to this if DMI and QPI licenses start becoming available to third parties.

At the surface, this deal hardly improved Nvidia's conditions on the marketplace. It could have made a difference back when Intel was bargaining against the ION platform but the ION 2 is just too disconnected from the CPU to be an actual benefit compared to the previous version - you better just go into other low power platforms from both AMD and Intel.

Finally, there are references to an acquiring entity being able to purchase AMD, VIA or Nvidia and retain the licenses granted to the company being bought, which could allow an AMD or VIA buyout from someone. The VIA merger with Nvidia could be interesting but VIA would still be quite expensive for a company with little cash right now and both parties would have to be interested in a merger.

AMD

Pretty much the same that was outlined in the settlement with AMD, the previous settlement was the groundbreaking change for AMD. There's also the buyout clause mentioned previously but I don't see an interesting deal for anyone involving AMD right now.

Conclusion

This settlement places many restrictions on Intel's business practices more than any previous lawsuit. It was actually better to place unambiguous restrictions that fining the company and see no results from months of court hearings.
The main company to benefit from this settlement is AMD but there's now an open door for Nvidia and VIA to merge and become a leading contender on the x86 market, while leaving the chipset license issues without a meaningful resolution for the time being.