Nvidia are making a 45nm CPU

[bootup05]

http://www.theinquirer.net/default.aspx?article=35216
http://www.engadget.com/2006/10/19/nvidia-has-x86-cpu-in-the-works/
http://hardware.slashdot.org/article.pl?sid=06/10/20/157207&from=rss

Didn't truly believe this at first but there seems to be quite a few sources popping up on the same thing.

Thoughts people ?

(BTW, is my sig alright , I know its this 4 line thing but whenever I try to make the thing use a whole line it keeps popping back?)

 

[maddness]

maybe the inquirer is spreading roumers and the rest are picking it up

 

[NathanE]

It doesn't mention X86 anywhere. My guess is they are developing something that integrates into Hypertransport or Intel CSI as a processing accelerator. It may provide some unique instruction sets to optimise certain unspecialised (CPU) functions.

To come from no where with a chip that is on-par with Intel and AMD's offering is, frankly, utterly impossible.

My 2p

 

[Jokester]

Though to be honest this reads as if it's for low end office boxes or otherwise cheap systems where CPU power isn't that big a deal so it wouldn't need to compete with the high end but on heat output and cost.

Jokester

 

[wizardmaxx]

My conspiracy theory confirmed

http://forums.overclockers.co.uk/showthread.php?t=17627864

 

[Defcon5]

To come from no where with a chip that is on-par with Intel and AMD's offering is, frankly, utterly impossible.

My 2p

Right, as usual!

 

[Plasmoid]

Nvidia couldent make an x86 cpu on their own... or with AMD given the competition.

Also an integrate gpu/cpu couldent be any good for simple power reasons... short of a completly new motherboard and new psu's.

It's probably what everyone thinks it is, and its simply a laptop destined single chip celeron package. Though i wonder how a standard intel FSB would hold up with a gpu on it.

 

[Roughneck]

To come from no where with a chip that is on-par with Intel and AMD's offering is, frankly, utterly impossible.

My 2p

they did it with the orginal G-Force card

 

[NathanE]

Geforce isn't X86. Although I see your point in that it instantly obsoleted 3DFX's, ATI's and Matrox's 3D accelerator offerings, but that was under very different market conditions. 3D acceleration was still in its infancy (and probably still is) so the barriers to entry were still very low.

 

[NathanE]

Nvidia couldent make an x86 cpu on their own... or with AMD given the competition.

Also an integrate gpu/cpu couldent be any good for simple power reasons... short of a completly new motherboard and new psu's.

It's probably what everyone thinks it is, and its simply a laptop destined single chip celeron package. Though i wonder how a standard intel FSB would hold up with a gpu on it.

CPU/GPU hybrid chips will no doubt start off in the low range markets... but the potential for extremely high performance is most definately there. I don't think motherboard or PSU redesigns has stopped progress in the past and it certainly won't this time.

The benefits of having the GPU and CPU on the same die are mahoosive. Latency to access textures would be extremely small. The GPU could use the CPU's L2 cache and general purpose pipelines. The possibilities are endless really.

 

[FrankJH]

surely its a conflict of interest to have probably the two highest cost peinces of a pc ( the cpu and gpu) on one die?

Low cost items are one thing - ie budget laptops etc, yes I can see this happening ie a 2.4Ghz single core + a 6300 gpu or something ( ie both a generation behnd leading specs but good for budget offerings)

The problem i have is that with high spec items, even a E6400 Core 2 duo and a 7950 or something, not only do intel or nvidia know exactly how fast they are going to be until after fabrication , ie there is high possibility of you ending up with one or other of the two halves being of a high quality but the other half letting it down - so all parts are basically a middle road

The other main problem I see is the cost of ownership to the public - the fabulous thing abotu a pc at present is that you can just add one part and get a decent speed boost for a "reasonable " cost , this type of combined chip will make it impossible as they will be physically the same part and the customer will be stuck with it for twice as long to justify the increase in cost.

Also think of the liklihood of shared system and graphics memory yuckkkkkkk!!!

 

[alexthecheese]

surely its a conflict of interest to have probably the two highest cost peinces of a pc ( the cpu and gpu) on one die?

Low cost items are one thing - ie budget laptops etc, yes I can see this happening ie a 2.4Ghz single core + a 6300 gpu or something ( ie both a generation behnd leading specs but good for budget offerings)

The problem i have is that with high spec items, even a E6400 Core 2 duo and a 7950 or something, not only do intel or nvidia know exactly how fast they are going to be until after fabrication , ie there is high possibility of you ending up with one or other of the two halves being of a high quality but the other half letting it down - so all parts are basically a middle road

The other main problem I see is the cost of ownership to the public - the fabulous thing abotu a pc at present is that you can just add one part and get a decent speed boost for a "reasonable " cost , this type of combined chip will make it impossible as they will be physically the same part and the customer will be stuck with it for twice as long to justify the increase in cost.

Also think of the liklihood of shared system and graphics memory yuckkkkkkk!!!

Personally I think it'd be a good idea, providing the costs are right. If you could buy a CPU/GPU combo that would immediately double your machine's power in one go, but - here's the caveat I guess - it was cheaper, then it would be worth it right?

I havent got a clue tbh how these things work but if they were intimately paired then surely it would be more efficient and ultimately a quicker pairing?

 

[YooEntSinMeROYT]

With AMD merging with ATI nVidia should be more than a little concerned at what the future may hold.

 

[lucifersam]

With AMD merging with ATI nVidia should be more than a little concerned at what the future may hold.

I think Nvidia need to be slightly worried, with Intel appearing to gear u for a stab at the high end GPU market and AMD-ATI planning to release CPU-GPU combos there going to need to do something

 

[NathanE]

surely its a conflict of interest to have probably the two highest cost peinces of a pc ( the cpu and gpu) on one die?

Low cost items are one thing - ie budget laptops etc, yes I can see this happening ie a 2.4Ghz single core + a 6300 gpu or something ( ie both a generation behnd leading specs but good for budget offerings)

The problem i have is that with high spec items, even a E6400 Core 2 duo and a 7950 or something, not only do intel or nvidia know exactly how fast they are going to be until after fabrication , ie there is high possibility of you ending up with one or other of the two halves being of a high quality but the other half letting it down - so all parts are basically a middle road

The other main problem I see is the cost of ownership to the public - the fabulous thing abotu a pc at present is that you can just add one part and get a decent speed boost for a "reasonable " cost , this type of combined chip will make it impossible as they will be physically the same part and the customer will be stuck with it for twice as long to justify the increase in cost.

Also think of the liklihood of shared system and graphics memory yuckkkkkkk!!!

I don't think AMD or ATI are kidding themselves when they announced a fairly low key "Fusion" chip. There is a lot of work to be done on the both the technical and commercial side. It will probably be a good 7+ years until the market conditions are ready for these chips in the high end anyway.

But I don't believe for a second that AMD spent, what was it - a few billion?, on buying ATI. If they only wanted to make chips with integrated graphics to flog to the el cheapo markets, there are about a thousand 3D chipset startups out there they could have bought to accomplish that. Clearly AMD has much bigger plans and so they went the full monty and bought the behemoth that is ATI.

Intel is feeling the pressure but they aren't in any rush yet because they've had their graphics department for years. It wouldn't take much to tell them to forget about integrating 3D on northbridges but to look at CPU integration instead... That's their short term solution (damn, those 3 words and Intel in the same paragraph again!) right there.

My 2p

 

[FrankJH]

I don't think AMD or ATI are kidding themselves when they announced a fairly low key "Fusion" chip. There is a lot of work to be done on the both the technical and commercial side. It will probably be a good 7+ years until the market conditions are ready for these chips in the high end anyway.

But I don't believe for a second that AMD spent, what was it - a few billion?, on buying ATI. If they only wanted to make chips with integrated graphics to flog to the el cheapo markets, there are about a thousand 3D chipset startups out there they could have bought to accomplish that. Clearly AMD has much bigger plans and so they went the full monty and bought the behemoth that is ATI.

Intel is feeling the pressure but they aren't in any rush yet because they've had their graphics department for years. It wouldn't take much to tell them to forget about integrating 3D on northbridges but to look at CPU integration instead... That's their short term solution (damn, those 3 words and Intel in the same paragraph again!) right there.

My 2p

Just to clarify - I never suggested that their seperate graphics cards would be irradicated - so obviously the investment of $5.5bn or so would be focused on all areas whether its their chipsets, integrated graphics and seperate predominantly PCI E graphics cards

I would also suggest that anything announced on the day the deal has been ratified is anything but low key myself, pretty major news however you look at it.

The article is suggesting 2 years (end of 2008) for the inital batch, by which time AMD and ATI will be completely integrated , or so they hope. 5 years after that in your estimation for high end integrated graphics is a huge amount of time - I would be surprised if they hadnt already done high end within 5 years from now!!!

While there may be lots of companies who do 3d integrated graphics only a very few are actually at the front line - and while some others maybe worth investing in I would suggest not even a handful would be worth buying outright , and at the end of the day its all about market share for AMD - so it was between nVidia and ATI