Monday, August 25th 2008

NVIDIA to Showcase its x86 Plans this NVISION?

Team R21 of FiringSquad studied the credibility of a rumor on NVIDIA materializing its long-term processor plans. They said they would be surprised if NVIDIA didn't have an x86 plan chalked out at least at a very interior level. The Inquirer speculated earlier that NVIDIA could lift the covers from its x86 plans as early as some time this week, during the ongoing NVISION event. The credibility of this rumor is based purely on who's breeding it. Many point it to have been doing rounds during IDF.

Reality bites: NVIDIA lacks a regularized x86 license which has to be issued by Intel to be able to use x86 in their products. Any mass announcement at this point could cost them. NVISION however looks to be an ideal substrate for discussions on CUDA and NVIDIA's SoC (System on a Chip) plans.Source: FiringSquad
Add your own comment

32 Comments on NVIDIA to Showcase its x86 Plans this NVISION?

#1
candle_86
well Nvidia might release x86 without Intel's premission using there own socket. All they have to do is build it in a clean room without using any of intels x86 code in it, if the built it 100% x86 compatabile in a clean room intel can't say shit
Posted on Reply
#2

^^ Then there will be hell to pay
Posted on Edit | Reply
#3
tigger
I'm the only one
So they cut intel off from sli,but expect a x86 license from them :laugh:
Posted on Reply
#4
candle_86
reread what i said, if they make a clean room x86 without using intel code, similar to how the team behind react OS is building a windows without any windows code.
Posted on Reply
#5

by: tigger69
So they cut intel off from sli,but expect a x86 license from them :laugh:
no they didn't , they gave them sli . maybe they got x86 in return . Will know more later this afternoon .
Posted on Edit | Reply
#6
Jansku07
That's gonna take a hell lotta work to make it happen. And even if they do, AMD and Intel will have over 10 year headstart in x86 business. Well competition is always good, so a third (I'm not counting Via) competitor would be good for customers.. Best luck!
So they didn't , they gave them sli . maybe they got x86 in return
Nvidia didn't give SLI to Intel, they gave it as a 30$ chip for the mobomakers. There's a big difference. And besides, x86 license is way more important than SLI-support.
Posted on Reply
#7

by: Jansku07
That's gonna take a hell lotta work to make it happen. And even if they do, AMD and Intel will have over 10 year headstart in x86 business.
If they manage to lure some key Amd and Intel designers, that headstart may be down to almost nothing.

by: Jansku07
Nvidia didn't give SLI to Intel, they gave it as a 30$ chip for the mobomakers. There's a big difference. And besides, x86 license is way more important than SLI-support.
To me and the consumer , it doesn't matter what they gave them, as long SLI is coming to Intel mobo's, which Intel was not allowed earlier. As far as I'm concerned Nvidia gave permission to Intel to put SLI on their mobo's, doesn't matter if they gave them a chip or a transistor.
Posted on Edit | Reply
#8
Jansku07
To me and the consumer , it doesn't matter what they gave them
It does matter. Nvidia could have made SLI a software solution, so that you wouldn't be stuck with a hot 30$ chip.
Posted on Reply
#9
Darkrealms
by: candle_86
well Nvidia might release x86 without Intel's premission using there own socket. All they have to do is build it in a clean room without using any of intels x86 code in it, if the built it 100% x86 compatabile in a clean room intel can't say shit
I won't argue it is possible but that is a lot of work and they better not code anything (even on their own) that looks like an Intel code or Intel will screw them.
by: wolf2009
To me and the consumer , it doesn't matter what they gave them, as long SLI is coming to Intel mobo's, which Intel was not allowed earlier. As far as I'm concerned Nvidia gave permission to Intel to put SLI on their mobo's, doesn't matter if they gave them a chip or a transistor.
I agree you have a point. Technically buy giving (...selling) the MFGs a chip Intel now has SLI on their chipsets.


Looking at the original source I guess we'll see what happens. (Nothing against you BTA, I'm talking about the Inquirer.)
Posted on Reply
#10
candle_86
by: Darkrealms
I won't argue it is possible but that is a lot of work and they better not code anything (even on their own) that looks like an Intel code or Intel will screw them.

I agree you have a point. Technically buy giving (...selling) the MFGs a chip Intel now has SLI on their chipsets.


Looking at the original source I guess we'll see what happens. (Nothing against you BTA, I'm talking about the Inquirer.)
not true, if they record all the work and all the logs intel cant touch them there either. Also a modern GPU is 100% more complex than an x86 chip.
Posted on Reply
#11
evil bill
by: candle_86
Also a modern GPU is 100% more complex than an x86 chip.
87% of all statistics are made up on the spot...........
Posted on Reply
#12
candle_86
a GPU has billions more transistors than a CPU, it also runs over a hundred parelle tasks at once. Now go find a CPU that does that?
Posted on Reply
#13
a111087
If they do release CPU, it will not be soon, but it will add some competition that's for sure
and that should be good for us
Posted on Reply
#14
evil bill
just pulling your leg matey - 100% seemed such a precise figure......;)


Maybe they already have an X86 license lined up, but they have managed to keep it quiet? Just because they are sabre rattling with each other public doesn't mean a backroom deal hasn't been done with Intel and Nvidia. Who knows what kind of confidentiality agreements could have been signed?
Posted on Reply
#15
TheLostSwede
Well, has anyone considered that US$25 million deal with Transmeta at all?
They made x86 compatible CPUs without any x86 tech in them at all.
What if, nVidia use some more of their tech than was in the press release from a few weeks ago?
It's not impossible or improbable, if they really are working on a high(er)-end CPU, as Tesla can hardly be called high-end.
Posted on Reply
#16
blueskynis
by: candle_86
a GPU has billions more transistors than a CPU, it also runs over a hundred parelle tasks at once. Now go find a CPU that does that?
That only means that GPU has many, tiny parts all doing the same job. Design of a CPU is far more complex than GPU. On other hand GPU design in simpler and it only has more components.

EDIT: Is x64 also licensed by Intel?
Posted on Reply
#17
candle_86
not true, a modern GPU is by far more complex, they do parrelle computing. They also do traditional CPU work as well, there is an FPU and an Interger inside a GPU. Add to that the shaders ect, and you get a complex chip. X86 can be implemented on a modern GPU if you where to program an x86 backend except a GPU would it alot faster.
Posted on Reply
#18
paul_phoenix
hahaha, "GPU is by far more complex because i say so...."

FPU and Integer unit are not GPU exclusive, CPU architectures include them also, and other things more complex like a advanced branch predictor, memory disambiguation, variable instructions to fixed instructions decoders, out of order execution and a lot of other things that GPU doesnt have beacause they dont need to.

GPUs aren't general purposes units, in fact, they are very specific, only with a software layer that pass across CPU (CUDA) is general purposes tasks possible on the GPUs.

sorry my bad english :D
Posted on Reply
#19
Darkrealms
by: candle_86
not true, if they record all the work and all the logs intel cant touch them there either. Also a modern GPU is 100% more complex than an x86 chip.
Intel has a lot of money, if nothing else they would probably try to delay it by making them prove everything before they could release it.
by: evil bill
87% of all statistics are made up on the spot...........
Thats like saying This statement is false. If what you say is true I have an 87% chance of it being false, but then it wouldn't be very true would it. LoL
by: TheLostSwede
Well, has anyone considered that US$25 million deal with Transmeta at all?
They made x86 compatible CPUs without any x86 tech in them at all.
What if, nVidia use some more of their tech than was in the press release from a few weeks ago?
It's not impossible or improbable, if they really are working on a high(er)-end CPU, as Tesla can hardly be called high-end.
Thats funny I kinda asked that question when that news post came up.
by: blueskynis
That only means that GPU has many, tiny parts all doing the same job. Design of a CPU is far more complex than GPU. On other hand GPU design in simpler and it only has more components.

EDIT: Is x64 also licensed by Intel?
I believe x64 was developed around 2000 by AMD and Intel/AMD traded techs. So technically would that mean that Nvidia would have to license x64 from AMD ??
Posted on Reply
#20
Morgoth
x86-64 is a 64-bit superset of the x86 instruction set architecture. Because the x86-64 instruction set is a superset of the x86 instruction set, all instructions in the x86 instruction set can be executed by central processing units (CPUs) that implement the x86-64 instruction set; therefore these CPUs can natively run programs that run on x86 processors from Intel, Advanced Micro Devices (AMD), and other vendors.

Historically, AMD has developed and produced processors patterned after Intel's original designs, but with x86-64, roles were reversed: Intel found itself in the position of adopting the architecture which AMD had created as an extension to Intel's own x86 processor line.

??

there was also a rumor that nvidia want to buy amd to get x86 but if the did it amd would't own the x86 license anny more nor nvidia can via amd
Posted on Reply
#21
v1ch0
by: Morgoth
x86-64 is a 64-bit superset of the x86 instruction set architecture. Because the x86-64 instruction set is a superset of the x86 instruction set, all instructions in the x86 instruction set can be executed by central processing units (CPUs) that implement the x86-64 instruction set; therefore these CPUs can natively run programs that run on x86 processors from Intel, Advanced Micro Devices (AMD), and other vendors.

Historically, AMD has developed and produced processors patterned after Intel's original designs, but with x86-64, roles were reversed: Intel found itself in the position of adopting the architecture which AMD had created as an extension to Intel's own x86 processor line.

??

there was also a rumor that nvidia want to buy amd to get x86 but if the did it amd would't own the x86 license anny more nor nvidia can via amd
NVidia buying AMD?!?! ABSOLUTELY impossible... At least EU won't allow it :D
Posted on Reply
#22
WarEagleAU
Bird of Prey
that was a while back that rumor was going on and Intel, I think, threatened to pull the x86 license for AMD if happened. It didnt and I think we are all the better for it. Its possible Nvidia could make an x86 compatible pc chip, just like IBM compatible PCs, Chips and OS's back in the day. Id be interested in seeing performance numbers on one, to be sure.
Posted on Reply
#23
brian.ca
Does the complexity of CPU vs. GPU even matter? If someone develops a rocket engine 100x more complex and powerful than your normal car engine while keeping it small enough to fit under the hood it will still mean jack if to the general population if they don't have ready access to rocket fuel.

GPUs might be more complex and have x times more processing power but if it can't run operating systems and programs written for x86 does it really have any chance of being accepted when the latter is so firmly entrenched and has some seriously big money behind the effort to keep it as such and even spread it out further?
Posted on Reply
#25
kaneda
by: brian.ca

GPUs might be more complex and have x times more processing power but if it can't run operating systems and programs written for x86 does it really have any chance of being accepted when the latter is so firmly entrenched and has some seriously big money behind the effort to keep it as such and even spread it out further?
Woot for open source developers! XD
Posted on Reply
Add your own comment