The way I see it, Intel traded top prospect rookies in exchange for a proven star. Intel gets most of the benefits now while NVIDIA may get bigger rewards in the future depending on how they play their cards (and invest their money).
"Interestingly the agreement also classifies an “Intel Architecture Emulator” as being a proprietary product. At first glance this would seem to disallow NVIDIA from making an x86 emulator for any of their products, be it their GPU holdings or the newly announced Project Denver ARM CPU. . Being officially prohibited from emulating x86 could be a huge deal for Denver down the road depending on where NVIDIA goes with it."
At second glance the agreement would also seem to disallow it:
1.8. “Intel Architecture Emulator” shall mean software, firmware, or hardware that, through emulation, simulation or any other process, allows a computer or other device that does not contain an Intel Compatible Processor, or a processor that is not an Intel Compatible Processor, to execute binary code that is capable of being executed on an Intel Compatible Processor.
1.12. “Intel Compatible Processor” shall mean any Processor that (a) can perform substantially the same functions as an Intel Processor by compatibly executing or otherwise processing (i) a substantial portion of the instruction set of an Intel Processor or (ii) object code versions of applications or other software targeted to run on an Intel Processor, in order to achieve substantially the same result as an Intel Processor; or (b) is substantially compatible with an Intel Processor Bus.
An interesting constraint, but IMHO likely to be at worst a mild speed bump. In 4-5 years we'll know better: maybe Nvidia will hit that constraint (IMHO doubtful as they'll have plenty of new workloads/apps); or the current trajectory will make it irrelevant (ok, we can't run your legacy x86 apps--who cares cuz we got plenty of other new and more important ones that people want).
In short, that constraint appears to be Intel looking to the past and attempting to protect their legacy turf, while Nvidia looks forward and their lawyers yawn and say "yeah, whatever, talk to you in six years."
What does "x86 is still a superior product" mean? That the x86 ISA is superior? That processors using the X86 ISA are superior? That specific vendor implementations of x86 ISA are superior? Or what?
And superior to what? We have an existence proof that x86 is demonstrably inferior in billions of cases. Otherwise why don't all those billions of routers, cell phones, microwave ovens, washers, dryers, refrigerators, cars, etc. use x86 instead of Z80, MIPS, ARM, etc. and their derivatives?
Yes, "x86 is still a superior product" in some ways and for some applications, but the area is shrinking, so be careful what ground you choose and how you choose to defend it.
I can't tell if vol7ron is being ironic, or authentic, but this is the kind of junk I'd expect nVidia to run into for consumers.
With retrospect, I could see the mobile-ARM market as lucrative and still booming, and the opposite of the desktop-market. Seems to me that nVidia is wagering everything, that their GPGPU + ARM architecture would beat Intel's SoC solution in terms of power-usage and parallel computation throughput.
But vol7ron brings up a good point, that the PC-users (and gamers) really only know of x86. And it doesn't matter if there's a better architecture/SoC out there, because it doesn't run Starcraft/LabView/Fishbowl. Even with a library-blessing from Microsoft, an ARM+GPGPU system still might not be licensed to actually run what software people are locked to.
Lord knows the developers are either slow to port across architectures (sans Apple). And nVidia is betting that somebody will take up the call for their ARM+GPGPU powerhouse. Microsoft might do it, but I've seen a lot of promising things Microsoft might do, that get axed. Apple might do it, but they've already been burned by straying from hardware too different than the norm. Linux/Android will definitely do it, but my enthusiasm for Linux distros depend on how well it runs Wine.
The way the article is worded, it seems like NVIDIA didn't have much of a claim, but because Intel had problems with AMD, the FTC just threw them in there. Basically, it makes it seem like the FTC is the bad guy. Seriously, though, the payment to NVIDIA shouldn't be more than AMD's, but the good thing to note, is that it's not a lump sum, so the NPV isn't as high as it would be.
As for the EU, I would say "F-them", even if it came at a cost. I'm getting sick and tired of the EU quibbles. The only reason they're stepping in is because they want a piece of the pie too. The settlement with AMD/NVIDIA should be enough.
It might be worth mentioning that Microsoft said they're going to support non-x86 chips as well. Thinking Denver, or something ARM-based.
The Intel-AMD and Intel-Nvidia disputes are completely different animals. They are completely unrelated, and involve very different laws and jurisdictions.
Intel-AMD dispute involvea monopolistic practices and restraint of trade, which necessarily involved the FTC/EU. Intel-Nvidia dispute is a straight contact/IP dispute AFAIK.
I don't see how that "makes it seem like the FTC is the bad guy." or how "the FTC just threw them in there" since the FTC (or EU) has no involvement in the Intel-Nvidia contretemps. Please enlighten.
Ok, I get that the FTC was a lurker, and if Intel and Nvidia didn't settle, Nvidia might still goad the FTC into action to make life painful for Intel.
What I would dispute is that the the FTC was much of a threat or substantially involved in this case, especially considering that the Nvidia-Intel spat preceded the FTC action (and was not initiated as part of an FTC complaint, although it subsequently contributed to it). The FTC effectively wemt MIA following the AMD settlement; as Ryan said "the FTC didn't get everything they wanted".
In short, I think the FTC's actions had pretty much run their course, and their involvement (real or potential) likely had very little to do with the this Intel-Nvidia settlement. Was the FTC still a club Nvidia could weild? Maybe as an irritant to Intel, but as a substantive threat I doubt it, and this article (among others) does nothing to suggest otherwise. YMMV.
Yes. I want to also say, that I was not serious about the EU comment. I'm not exactly sure what happened there and I think everyone knows EU business is important, both to intel and the global economy.
I just re-read what was written and it seemed more sour then humorous. I hope this clarifies it was not meant to be serious.
Intel has been breaking laws and being a general dick for quite a long time now. Intel is the bad guy and they should be sued another 100billion for wasting everyone's time lying in the courts pretending not to have done anything wrong.
"While Intel’s approval isn’t necessarily essential for Denver like it would be for an x86 CPU, it clearly is easier to build Denver without the risk of NVIDIA suing the pants off of NVIDIA again."
Several other sites are stating that NVIDIA does get the rights to DMI/QPI chispets , but they are still not going to make them , and instead will focus on Denver and Tegra .
"NVIDIA Licensed Chipsets shall not include any Intel Chipsets that are capable of electrically interfacing directly (with or without buffering or pin, pad or bump reassignment) with an Intel Processor that has an integrated (whether on-die or in-package) main memory controller, such as, without limitation, the Intel Processor families that are code named ‘Nehalem’, ‘Westmere’ and ‘Sandy Bridge.’”
It specifically spells out the fact that NVIDIA can't make chipsets for Nehalem or SB generation CPUs.
Funny how an idea gets put out there and taken up by others in the media and spread around, regardless of whether it is correct, isn't it? Makes you wonder how many people who write for the "news" actually go to the source and how much of what we see depends on the first reporter getting it right.
I always roll my proverbial eyes when someone says "everybody knows" something. Right, there's proof for ya, "everybody knows" it.
Thanks Ryan for being someone who goes to the actual source of information, not just another person's report of it.
what we have because of the lack of third party chipsets is the P67/H67/X58/no Z67 fiasco - high priced boards with less on the boards and a total mess.
I tend to agree with this statement. Had there been other chipset makers at the launch of SNB the enthusiasts would have flocked to the Z67-similar mobo as the current offerings are crap.
On-chip x86 memory controller patents are owned by AMD, Intel violated those patents with Nehalem, that was the primary thing driving Intel to the bargaining table - paying off AMD to get access to those licenses.
Intel cannot re-license those AMD patents to Nvidia, so Nvidia remains locked out of the on-chip memory controller x86 market.
Unless I am mistaken, if nVidia got the right to manufacture an x86 chip, any IP owned by AMD would have to be licensed from AMD, and not through Intel.
Yeah. That's what I just said. Intel cannot relicense AMD technology, so any IP owned by AMD would have to be licensed by Nvidia from directly from AMD, something not likely to happen.
Imagine a Sandy Bridge without a graphics engine (No next-gen for you-LOL Larrabee??) - Intel got a bargain; spread out over 6 years, is no-thing, I’m sure they would have paid a kings ransom- it would have been their death…Nvidia missed it
Agreed. Intel paying off AMD for its integrated memory controller patents and Nvidia for integrated graphics patents. That's the big difference between a Sandy Bridge and a Core2duo cpu's. Intel says that they are going to have revenues of 13 bil because of Sandy Bridge this year, more than enough to pay off AMD and Nvidia for their IP's. A real bargain for Intel indeed. AMD and Nvidia should've be smart enough and license their technology to Intel and receive $x per every processor intel sells.
It seems no one mentioned the SLI licensing that Intel has paid to NVIDIA prior to the settlement that 'allows' NVIDIA not to cripple SLI support at GeForce/Quadro drivers. That seems to be the only 'NVIDIA technology' that Intel ever gets out of the settlement, if the settlement is really just extending existing cross licensing deals. After all, the Intel HD Graphics has so far been based on PowerVR SGX cores licensed to Intel, it is unlikely for Intel to use new cores that that have incompatible instruction sets, especially now that Sandy Bridge-based CPUs include graphics cores that will get supported (or already supported?) by Intel compilers.
We’ve updated our terms. By continuing to use the site and/or by logging into your account, you agree to the Site’s updated Terms of Use and Privacy Policy.
Facebook
Twitter
RSS
30 Comments
Back to Article
AmdInside - Monday, January 10, 2011 - link
The way I see it, Intel traded top prospect rookies in exchange for a proven star. Intel gets most of the benefits now while NVIDIA may get bigger rewards in the future depending on how they play their cards (and invest their money).has407 - Monday, January 10, 2011 - link
Thanks Ryan; very good synopsis."Interestingly the agreement also classifies an “Intel Architecture Emulator” as being a proprietary product. At first glance this would seem to disallow NVIDIA from making an x86 emulator for any of their products, be it their GPU holdings or the newly announced Project Denver ARM CPU. . Being officially prohibited from emulating x86 could be a huge deal for Denver down the road depending on where NVIDIA goes with it."
At second glance the agreement would also seem to disallow it:
1.8. “Intel Architecture Emulator” shall mean software, firmware, or hardware that, through emulation, simulation or any other process, allows a computer or other device that does not contain an Intel Compatible Processor, or a processor that is not an Intel Compatible Processor, to execute binary code that is capable of being executed on an Intel Compatible Processor.
1.12. “Intel Compatible Processor” shall mean any Processor that (a) can perform substantially the same functions as an Intel Processor by compatibly executing or otherwise processing (i) a substantial portion of the instruction set of an Intel Processor or (ii) object code versions of applications or other software targeted to run on an Intel Processor, in order to achieve substantially the same result as an Intel Processor; or (b) is substantially compatible with an Intel Processor Bus.
An interesting constraint, but IMHO likely to be at worst a mild speed bump. In 4-5 years we'll know better: maybe Nvidia will hit that constraint (IMHO doubtful as they'll have plenty of new workloads/apps); or the current trajectory will make it irrelevant (ok, we can't run your legacy x86 apps--who cares cuz we got plenty of other new and more important ones that people want).
In short, that constraint appears to be Intel looking to the past and attempting to protect their legacy turf, while Nvidia looks forward and their lawyers yawn and say "yeah, whatever, talk to you in six years."
vol7ron - Monday, January 10, 2011 - link
But x86 is still a superior product.has407 - Tuesday, January 11, 2011 - link
OK, I'll bite...What does "x86 is still a superior product" mean? That the x86 ISA is superior? That processors using the X86 ISA are superior? That specific vendor implementations of x86 ISA are superior? Or what?
And superior to what? We have an existence proof that x86 is demonstrably inferior in billions of cases. Otherwise why don't all those billions of routers, cell phones, microwave ovens, washers, dryers, refrigerators, cars, etc. use x86 instead of Z80, MIPS, ARM, etc. and their derivatives?
Yes, "x86 is still a superior product" in some ways and for some applications, but the area is shrinking, so be careful what ground you choose and how you choose to defend it.
Tros - Tuesday, January 11, 2011 - link
I can't tell if vol7ron is being ironic, or authentic, but this is the kind of junk I'd expect nVidia to run into for consumers.With retrospect, I could see the mobile-ARM market as lucrative and still booming, and the opposite of the desktop-market. Seems to me that nVidia is wagering everything, that their GPGPU + ARM architecture would beat Intel's SoC solution in terms of power-usage and parallel computation throughput.
But vol7ron brings up a good point, that the PC-users (and gamers) really only know of x86. And it doesn't matter if there's a better architecture/SoC out there, because it doesn't run Starcraft/LabView/Fishbowl. Even with a library-blessing from Microsoft, an ARM+GPGPU system still might not be licensed to actually run what software people are locked to.
Lord knows the developers are either slow to port across architectures (sans Apple). And nVidia is betting that somebody will take up the call for their ARM+GPGPU powerhouse. Microsoft might do it, but I've seen a lot of promising things Microsoft might do, that get axed. Apple might do it, but they've already been burned by straying from hardware too different than the norm. Linux/Android will definitely do it, but my enthusiasm for Linux distros depend on how well it runs Wine.
Haawser - Monday, February 13, 2017 - link
Won't change anything. Even if Nvidia get to emulate x86, there's no way AMD will let them use x64. And x86 without x64 is less than useless.vol7ron - Monday, January 10, 2011 - link
The way the article is worded, it seems like NVIDIA didn't have much of a claim, but because Intel had problems with AMD, the FTC just threw them in there. Basically, it makes it seem like the FTC is the bad guy. Seriously, though, the payment to NVIDIA shouldn't be more than AMD's, but the good thing to note, is that it's not a lump sum, so the NPV isn't as high as it would be.As for the EU, I would say "F-them", even if it came at a cost. I'm getting sick and tired of the EU quibbles. The only reason they're stepping in is because they want a piece of the pie too. The settlement with AMD/NVIDIA should be enough.
It might be worth mentioning that Microsoft said they're going to support non-x86 chips as well. Thinking Denver, or something ARM-based.
has407 - Monday, January 10, 2011 - link
The Intel-AMD and Intel-Nvidia disputes are completely different animals. They are completely unrelated, and involve very different laws and jurisdictions.Intel-AMD dispute involvea monopolistic practices and restraint of trade, which necessarily involved the FTC/EU. Intel-Nvidia dispute is a straight contact/IP dispute AFAIK.
I don't see how that "makes it seem like the FTC is the bad guy." or how "the FTC just threw them in there" since the FTC (or EU) has no involvement in the Intel-Nvidia contretemps. Please enlighten.
vol7ron - Monday, January 10, 2011 - link
Re-read the article if you're curious about how the FTC popped in.It's still unclear who was in the wrong. Not saying intel is innocent, but AMD outcome didn't help.
has407 - Tuesday, January 11, 2011 - link
Ok, I get that the FTC was a lurker, and if Intel and Nvidia didn't settle, Nvidia might still goad the FTC into action to make life painful for Intel.What I would dispute is that the the FTC was much of a threat or substantially involved in this case, especially considering that the Nvidia-Intel spat preceded the FTC action (and was not initiated as part of an FTC complaint, although it subsequently contributed to it). The FTC effectively wemt MIA following the AMD settlement; as Ryan said "the FTC didn't get everything they wanted".
In short, I think the FTC's actions had pretty much run their course, and their involvement (real or potential) likely had very little to do with the this Intel-Nvidia settlement. Was the FTC still a club Nvidia could weild? Maybe as an irritant to Intel, but as a substantive threat I doubt it, and this article (among others) does nothing to suggest otherwise. YMMV.
vol7ron - Tuesday, January 11, 2011 - link
Yes. I want to also say, that I was not serious about the EU comment. I'm not exactly sure what happened there and I think everyone knows EU business is important, both to intel and the global economy.I just re-read what was written and it seemed more sour then humorous. I hope this clarifies it was not meant to be serious.
Phoenixlight - Tuesday, January 11, 2011 - link
Intel has been breaking laws and being a general dick for quite a long time now. Intel is the bad guy and they should be sued another 100billion for wasting everyone's time lying in the courts pretending not to have done anything wrong.
anactoraaron - Monday, January 10, 2011 - link
"intel doesn't get" in the last table should be changed from "?" to "keeping 1.5 billion dollars"HA!
smookyolo - Tuesday, January 11, 2011 - link
I second this.neutralizer - Monday, January 10, 2011 - link
"While Intel’s approval isn’t necessarily essential for Denver like it would be for an x86 CPU, it clearly is easier to build Denver without the risk of NVIDIA suing the pants off of NVIDIA again."Page 2.
GeorgeH - Monday, January 10, 2011 - link
Great article, thank you.For the curious, Intel made ~$60 Billion (profit) in the past decade, so ~4.2 Billion really is a tiny faction (~7%.)
Muhammed - Tuesday, January 11, 2011 - link
Several other sites are stating that NVIDIA does get the rights to DMI/QPI chispets , but they are still not going to make them , and instead will focus on Denver and Tegra .The sites are:
1- http://techreport.com/discussions.x/20222
2-http://www.brightsideofnews.com/news/2011/1/10/nvi...
3-http://arstechnica.com/business/news/2011/01/intel...
so who is right ?
Ryan Smith - Tuesday, January 11, 2011 - link
I have every reason to believe we are.From the amended chipset licensing agreement:
"NVIDIA Licensed Chipsets shall not include any Intel Chipsets that are capable of electrically interfacing directly (with or without buffering or pin, pad or bump reassignment) with an Intel Processor that has an integrated (whether on-die or in-package) main memory controller, such as, without limitation, the Intel Processor families that are code named ‘Nehalem’, ‘Westmere’ and ‘Sandy Bridge.’”
It specifically spells out the fact that NVIDIA can't make chipsets for Nehalem or SB generation CPUs.
Muhammed - Tuesday, January 11, 2011 - link
Thanks for confirmation , I guess they misunderstood the statements .Sabresiberian - Tuesday, January 11, 2011 - link
Funny how an idea gets put out there and taken up by others in the media and spread around, regardless of whether it is correct, isn't it? Makes you wonder how many people who write for the "news" actually go to the source and how much of what we see depends on the first reporter getting it right.I always roll my proverbial eyes when someone says "everybody knows" something. Right, there's proof for ya, "everybody knows" it.
Thanks Ryan for being someone who goes to the actual source of information, not just another person's report of it.
;)
james.jwb - Tuesday, January 11, 2011 - link
what we have because of the lack of third party chipsets is the P67/H67/X58/no Z67 fiasco - high priced boards with less on the boards and a total mess.7Enigma - Tuesday, January 11, 2011 - link
I tend to agree with this statement. Had there been other chipset makers at the launch of SNB the enthusiasts would have flocked to the Z67-similar mobo as the current offerings are crap.HangFire - Tuesday, January 11, 2011 - link
On-chip x86 memory controller patents are owned by AMD, Intel violated those patents with Nehalem, that was the primary thing driving Intel to the bargaining table - paying off AMD to get access to those licenses.Intel cannot re-license those AMD patents to Nvidia, so Nvidia remains locked out of the on-chip memory controller x86 market.
Stuka87 - Tuesday, January 11, 2011 - link
Unless I am mistaken, if nVidia got the right to manufacture an x86 chip, any IP owned by AMD would have to be licensed from AMD, and not through Intel.HangFire - Tuesday, January 11, 2011 - link
Yeah. That's what I just said. Intel cannot relicense AMD technology, so any IP owned by AMD would have to be licensed by Nvidia from directly from AMD, something not likely to happen.ash9 - Tuesday, January 11, 2011 - link
Imagine a Sandy Bridge without a graphics engine (No next-gen for you-LOL Larrabee??) - Intel got a bargain; spread out over 6 years, is no-thing, I’m sure they would have paid a kings ransom- it would have been their death…Nvidia missed itasH
pugster - Wednesday, January 12, 2011 - link
Agreed. Intel paying off AMD for its integrated memory controller patents and Nvidia for integrated graphics patents. That's the big difference between a Sandy Bridge and a Core2duo cpu's. Intel says that they are going to have revenues of 13 bil because of Sandy Bridge this year, more than enough to pay off AMD and Nvidia for their IP's. A real bargain for Intel indeed. AMD and Nvidia should've be smart enough and license their technology to Intel and receive $x per every processor intel sells.quanta - Wednesday, January 12, 2011 - link
It seems no one mentioned the SLI licensing that Intel has paid to NVIDIA prior to the settlement that 'allows' NVIDIA not to cripple SLI support at GeForce/Quadro drivers. That seems to be the only 'NVIDIA technology' that Intel ever gets out of the settlement, if the settlement is really just extending existing cross licensing deals. After all, the Intel HD Graphics has so far been based on PowerVR SGX cores licensed to Intel, it is unlikely for Intel to use new cores that that have incompatible instruction sets, especially now that Sandy Bridge-based CPUs include graphics cores that will get supported (or already supported?) by Intel compilers.tafreire2011 - Thursday, January 13, 2011 - link
CrossFire support will end on Intel motherboards? Or not?cesthree - Tuesday, January 18, 2011 - link
"If you ask Intel, NVIDIA’s agreement only covers AGTL+,"This was Intel's only mistake, allowing Nvidia to disgrace the AGTL+ with their presence.
Stick to GPU's Nvidia; nothing past PCIE bus, please.