Nvidia to enter into CPU business?

[AthlonXP1800]

http://www.ngohq.com/home.php?page=Articles&go=read&arc_id=91

Nvidia to make Nvidia CPU? Look like it will happen in the future. :)

[a12ctic]

thatll be awsome, then i can get a computer completly powered by nvidia :)

[monly]

Maybe a hybrid cpu/gpu?:D

[agentkay]

Sounds logical considering that GPUs are quite likely to become general purpose processors in the long term, as Tim Sweeney and others have predicted/mentioned.

[CaptNKILL]

I'd buy one :D

[Tuork]

Maybe a hybrid cpu/gpu?:D

THAT would be cool.

Massive heatsink I would imagine. :p

[Redeemed]

I'm not sure I like the idea. I mean, the CPU does its current job just fine, whereas the GPU does its job just fine. I don't see why the two should become one. How would the two divy up the memory between themselves? What about multi-core CPUs? And also the communication between the CPU(s) and the GPU(s) if they are all combined into one?

As far as a PU that does it all- that USED to be the CPU, and look how that turned out.

Then, lastly, the cooling requirements and power consumption would probably be attrocious. I'm sure the minimal sized fan would be 92mm and the heatsink would have to be pure CU with heatpipes. Granted, we'd have more room in the computer tower since there would technically be no video card to take up space. Which, again, makes you wonder how would SLi, Quad SLi, or CrossFire work?

I just don't like the idea. Too many issues to overcome. Maybe around 2015 this might be plausible, but before then- how?

[ViN86]

I'm not sure I like the idea. I mean, the CPU does its current job just fine, whereas the GPU does its job just fine. I don't see why the two should become one. How would the two divy up the memory between themselves? What about multi-core CPUs? And also the communication between the CPU(s) and the GPU(s) if they are all combined into one?

As far as a PU that does it all- that USED to be the CPU, and look how that turned out.

Then, lastly, the cooling requirements and power consumption would probably be attrocious. I'm sure the minimal sized fan would be 92mm and the heatsink would have to be pure CU with heatpipes. Granted, we'd have more room in the computer tower since there would technically be no video card to take up space. Which, again, makes you wonder how would SLi, Quad SLi, or CrossFire work?

I just don't like the idea. Too many issues to overcome. Maybe around 2015 this might be plausible, but before then- how?
theyre not saying that theyre going to merge the CPU and GPU into one chip. what theyre saying is that nvidia is going to design CPU's like the Athlon64 and Conroe sometime in the future. the chip will run like current CPUs do and it will compete with Intel and AMD.

i believe it. think of how complex and specialized GPU's are. i bet CPU's are easier to design.

cant find a link or anything to support this though, so feel free to say if im wrong, heh :p

[borntosoul]

cpu's are far more harder to design and bring to market at a set price point IMO. if you think NV are just gonna step up and be competitive with Intel and AMD - its not gonna happen. NV and IBM together might be possible but i highly doubt it.

[Redeemed]

Well, from a comparison stand point, I don't see how a CPU could be any harder to produce than GPU. I guess they would need a fab specifically designed for producing CPUs, as the manufacturing process is different. But other than that, once they have the functional fab I'm sure they'd produce a decent competitor.

Only probably would be making that CPU of theirs competitive price wise, and letting everybody know about it at the same time.

theyre not saying that theyre going to merge the CPU and GPU into one chip. what theyre saying is that nvidia is going to design CPU's like the Athlon64 and Conroe sometime in the future. the chip will run like current CPUs do and it will compete with Intel and AMD.

i believe it. think of how complex and specialized GPU's are. i bet CPU's are easier to design.

cant find a link or anything to support this though, so feel free to say if im wrong, heh

Okay, I apologize and stand corrected as I misunderstood what was being said. My bad. ;)

This seems far more feasible. In fact, it even sounds awesome. If true, I wonder if ATI will follow suit.

And after that AMD and Intel will start making high end GPUs. Heh! That would be wierd.

[Soylent]

Well, from a comparison stand point, I don't see how a CPU could be any harder to produce than GPU. I guess they would need a fab specifically designed for producing CPUs, as the manufacturing process is different. But other than that, once they have the functional fab I'm sure they'd produce a decent competitor.

GPUs are made from blocks with known timing characteristics. CPUs often need a whole lot of down and dirty hand tuning(the characteristics of which you need to verify yourself!) and they are very general so they need a large amount of messy kludge like out of order execution, register renaming, out of order load/store, decoding of lots and lots of x86 instructions, branch prediction, low latency cache and so on.

[ViN86]

Link to article.... (http://www.sharkyextreme.com/forums_spotlight/38/4.shtml)

GPU vs CPU - Please explain architectural differences.
By mellojoe September 09, 2001, 11:47 PM

What are the major (and minor, heck everything) differences between a CPU and a Graphics Processing unit. I mean, why is it that there are 1.4 to 2Ghz processors out but GPU core clocks have only reached 500mhz (or approximately). My Radeon, for instances, is only clocked at 200mhz.

Is it different like RISC is differenct from CISC?

It seems like CPU processing power is no longer a bottleneck to major applications anymore. The only component that is actually stressed these days is the GPU. So why is it that the GPU falls so far behind the CPU in clock frequency?

By idris5 September 10, 2001, 04:03 AM

You've touched on an interesting subject here where complexity is usually wrongly judged (by the less technically minded) by the transistor count.

The difference between the attainable clock rates of a GPU and a CPU comes down to the way they are designed - hang on, I know you think this is an obvious comment, but bare with me.

CPU's are designed using an approach called full custom. This means that every single transistor is hand optimised for its location in the circuit. So, for example, if you want to drive a long interconnect you'll need a high driving strength transistor (so the aspect ratio of gate width to length is increased). There are lots of tricks that can be brought to bare to eek out tiny performance advantages in each transistor (like placing transistors next to each other so that they abutt which means no interconnect is required), however it takes a long time to design using full custom methods.

GPU designers, on the other hand, absolubtly can not afford long turn around times - time to market is king - so they use another approach called cell based design.

Cell based design takes a standard set of cells (which are gates like buffers, AND gates, latches etc) that are unique to a foundary process like TI 0.13 or TSMC 0.18. These cells are generic and are instantiated into the design with little (I'm not going to overcomplicate this too much) consideration to the function of the cell in that part of circuit. Because of this the performance achievable is not as high as with full custom, however you can iterate the back end of the design (RTL->Netlist->layout [chip floorplan]->GDSII [what is sent to the foundary for manufacturing]) very quickly (three weeks or so, depending on the speed of the machines available) and fix problems as and when they come up.

Cell based design isn't simple and there are an awful lot of pitfalls (many of which are brought about by the design tools), but the big advantage it has over full custom is time to market, which it does at the expense of speed.

By idris5 September 10, 2001, 05:28 AM

Oh, and here is an example of one of the major limitations of cell based design with current design tools.

Synthesis tools can currently only swallow around 500,000 instances (gates, sort of) in one go. So therefore if you have a GPU of 10 million + instances (not transistors) you have to break it down into lots of sub blocks and synthesise these blocks seperately.

The problem here is that even though you make as much effort as possible to place pins on these macro blocks as close to each other as possible, you almost always haves buses that need to go to at least two blocks and so one has to follow a much longer path (as you can't stack macro blocks on top of each other).

Obviously you can try and make sure that your critical path has macro blocks placed as close to each other as possible, but you will always get some degradation in maximum clock rate.

Edit - Realised that many won't know what synthesis is! Synthesis is a process that occurs around the middle of the design flow. You take a verilog RTL description of your design, which is a reasonably high level description of your design (its a bit C [programming language] like), and this is then synthesised to a gate level description of your design called a netlist.
found something about the design of GPU and CPU. seems the two are quite different. too bad i know nothing about the design of these components or how difficult it is, so i cant comment/assess how easy the transition will be.

[Redeemed]

GPUs are made from blocks with known timing characteristics. CPUs often need a whole lot of down and dirty hand tuning(the characteristics of which you need to verify yourself!) and they are very general so they need a large amount of messy kludge like out of order execution, register renaming, out of order load/store, decoding of lots and lots of x86 instructions, branch prediction, low latency cache and so on.

I never looked at it that way, good point. I just assumed that since both were extremely complex components that they should be equivalently difficult to produce. I now stand corrected. :D

I think it would be good to have another competitor in the CPU business and between ATI and nVidia the latter is in the better position to begin competing. Though I just hope nVidia doesn't bite off more than it can chew.

And I'm sure ATI would try to join in on the parade eventually.

[ViN86]

I never looked at it that way, good point. I just assumed that since both were extremely complex components that they should be equivalently difficult to produce. I now stand corrected. :D

I think it would be good to have another competitor in the CPU business and between ATI and nVidia the latter is in the better position to begin competing. Though I just hope nVidia doesn't bite off more than it can chew.

And I'm sure ATI would try to join in on the parade eventually.
if theyre not purchased by AMD first, heh.

[tieros]

It would probably be crazy for them to try and go up against AMD or Intel right out of the chute, but with their current focus on the mobile device market, being able to offer a basic CPU, along with the GPU and glue as a single package may not be a bad idea.

Or maybe a coprocessor for the HTX socket. I'd love to have a 1K wide vector coprocessor if anyone from nVidia is listening :D

[Soylent]

(The turn around times for CPUs(major revisions like 486->pentium->pentium pro etc.) are about half a decade long. Even if Intel had desperately wanted to dump netburst they were absolutely stuck with it for several years.)

[supra]

id buy one too :D

[Heinz68]

http://www.ngohq.com/home.php?page=Articles&go=read&arc_id=91

Nvidia to make Nvidia CPU? Look like it will happen in the future. :)SURE IT LOOKS LIKE.
"Somebody meets one off the highest Nvidia executive (he/she) in a restaurant and according to Him/Her, Nvidia will invade the processor market in the long future."

I did meet somebody in restaurant from Switzerland and (he/she) told me that Switzerland will invade China way before Nvidia invades CPU market.
Not sure but I think that somebody from Switzerland he/she might have had few too many.

[retsam]

i hope everyone knows that nvidia cant just make an x86 cpu they would have to lisence tech from amd and intel.intel would have to give out an license jsut for x86 8088 op code if im not mistaken...not an easy thing todo, imho

[msxyz]

I don't think they have the necessary experitse. As someone else pointed out, state of the art CPUs don't rely on standard "Building blocks" used by ASICS manufacturers. Intel and AMD put considerable efforts into tuning not only the manufacturing process but also constantly improving the mask layout (in the past this was done by hand!) for the sake of speed and efficiency.

Just look at Via, for example. With the Cyrix team gone (much of them left after Via didn't invest in the true 6x86 successor supporting Centaur cores instead) all they can build are simple in order CPUs which can barely manage to surpass 1 GHz with current technologies. If we look at other synthetizable cores, they're in the same ballpark. MIPS, SH, Arm... all of them are still struggling to go beyond 1 GHz despite some of them do have complex and very efficient architectures (ie MIPS64)

[Toss3]

SURE IT LOOKS LIKE.
"Somebody meets one off the highest Nvidia executive (he/she) in a restaurant and according to Him/Her, Nvidia will invade the processor market in the long future."

I did meet somebody in restaurant from Switzerland and (he/she) told me that Switzerland will invade China way before Nvidia invades CPU market.
Not sure but I think that somebody from Switzerland he/she might have had few too many.
:D :D

Heh think about having an NVIDIA cpu and gpu, now we only need MBs(we 've got nforce, but green nvidia sli mainboards would be so cool :headbang:) and RAM and we'd be set. ;)

[yooyo]

Maybe NVidia want to produce add-on coprocessors for upcoming AMD technology...or... they can licence hypertransport from AMD and produce chipset/mobo with two sockets AMD + GPU connected by hypertransport.

(pirate)

[Nv40]

Well if you look at the latest TIm Sweney by jackup , he clearly told that he expect Gpus to be seen as temporary solution when looking back 10 years in the future. Games will have bigger outdoors and more massive battles with more complex AI ,in others words more realism.those things cannot be accelerated eficiently with Gpus.

We are close to hit a wall in games design where games could not be better graphically ,until we have a way more powerfull cpus or multicore and ,multithreading is fully supported in the gaming industry. The future in graphics is raytracing and every other indirect ilumination technique,and Gpus arent very efficient with that kind of work loads. they are very usefull now that games are low in geometry and high on texturing. So Gpus still will have a place for many years . but sooner or later everything will return again to the "old ways" of Software renderings.. with games recreating very complex and alive worlds that the FIlm industry have been using alot for CG movies but this time in real time.