why nvdia still not making big monolithic CPU with ARM ?

[Idontcare]

i think you forget one thing AMD currently taking approach to open up their architecture for third party. so maybe (although its unlikely) they will licensing several AMD patented CPU technology to nvdia ??? like 3Dnow etc.

Its not a matter of forgetting, it is a matter of it not being relevant or applicable. 3DNow! went nowhere even in the x86 based ISA for which it was intended, it certainly isn't suddenly going to make ARM all the more a contender of x86 today just because they include it.

Consider this, the best that AMD has to offer ARM is already included in its current Bulldozer, Llano, and Bobcat products. Which of those competes well with Intel?

Now consider the IC design pipeline and timeline. It is about 4yrs long.

So what is ARM (or Nvidia) going to take from AMD today in their designs for an ARM chip that won't be produced for 4yrs but is still going to be competitive against what Intel is going to be selling in 4yrs?

This is not as easy as stating (1) and then waiting for (3):

(1) I challenge you to make a "big" ARM core
(2) Somebody go figure out what that means and how to do it
(3) Profit

That step (2) is the whole reason it hasn't been done already, and until the problems with step (2) are circumvented there is going to continue to be too little reward and too much risk for any company to pursue it.

btw is 3Ghz ARM really difficult to build ?????maybe with just increasing tegra frequencies, and without expanding the ISA, nvdia could in theory make high performance ARM based CPU ???

3GHz ARM is not difficult to build. But the IPC would suck so bad it would not be a product that anyone would want to buy. That would be like making a 3GHz 486 cpu, the stalls in the pipeline would be massive in clock counts, making the performance not all the good but power usage would be high.

To make a 3GHz ARM chip that maintained decent performance would require all the cache tricks, branch predictors, memory bandwidth, etc that modern day multi-GHz x86 CPU's have. The ROI is just no there to try and design that from the ground up.

Again it took AMD and Intel 20yrs to get to where they are today, you can't really expect ARM and Nvidia to make that transition in 4yrs. There is a huge gulf in technical capabilities.

If AMD thought it was doable then they would have done it instead of designing Brazos/Bobcat from the ground up. Same with Intel and Atom. The very people who know exactly how challenging it is to do this stuff in reality looked at it and concluded they were better off doing ground-up new designs with x86 rather than ARM.

That should tell you a lot. Combine that with the fact there are no "big" ARM chips out there and now you have the benefit of using both the logic of deduction and the logic of induction to close in on the answer from above and below.

 

[wahdangun]

Its not a matter of forgetting, it is a matter of it not being relevant or applicable. 3DNow! went nowhere even in the x86 based ISA for which it was intended, it certainly isn't suddenly going to make ARM all the more a contender of x86 today just because they include it.

Consider this, the best that AMD has to offer ARM is already included in its current Bulldozer, Llano, and Bobcat products. Which of those competes well with Intel?

ok, i think you are right, but then what is the reason for AMD to open up their architecture for third party ??? i mean nvdia could possible slap together their GPU and AMD bobcat CPU to make their own fusion processor.

Now consider the IC design pipeline and timeline. It is about 4yrs long.

So what is ARM (or Nvidia) going to take from AMD today in their designs for an ARM chip that won't be produced for 4yrs but is still going to be competitive against what Intel is going to be selling in 4yrs?

This is not as easy as stating (1) and then waiting for (3):

(1) I challenge you to make a "big" ARM core
(2) Somebody go figure out what that means and how to do it
(3) Profit

That step (2) is the whole reason it hasn't been done already, and until the problems with step (2) are circumvented there is going to continue to be too little reward and too much risk for any company to pursue it.


3GHz ARM is not difficult to build. But the IPC would suck so bad it would not be a product that anyone would want to buy. That would be like making a 3GHz 486 cpu, the stalls in the pipeline would be massive in clock counts, making the performance not all the good but power usage would be high.

To make a 3GHz ARM chip that maintained decent performance would require all the cache tricks, branch predictors, memory bandwidth, etc that modern day multi-GHz x86 CPU's have. The ROI is just no there to try and design that from the ground up.

Again it took AMD and Intel 20yrs to get to where they are today, you can't really expect ARM and Nvidia to make that transition in 4yrs. There is a huge gulf in technical capabilities.

If AMD thought it was doable then they would have done it instead of designing Brazos/Bobcat from the ground up. Same with Intel and Atom. The very people who know exactly how challenging it is to do this stuff in reality looked at it and concluded they were better off doing ground-up new designs with x86 rather than ARM.

That should tell you a lot. Combine that with the fact there are no "big" ARM chips out there and now you have the benefit of using both the logic of deduction and the logic of induction to close in on the answer from above and below.

i think it won't need 20 years to develop competitive cpu based on ARM architecture because they could borrow some patent from AMD and after the fact that ARM currently have advanced in blazing speed, i mean we can even buy quadcore phone right now.

and i think ARM currently is being held back with their ultra low power design. nad because there is absolutely no demand for high performance one

 

[Ferzerp]

ok, i think you are right, but then what is the reason for AMD to open up their architecture for third party ??? i mean nvdia could possible slap together their GPU and AMD bobcat CPU to make their own fusion processor.

How exactly would nvidia make an x86 processor without an x86 license again?

 

[wahdangun]

How exactly would nvidia make an x86 processor without an x86 license again?

because AMD can fussing their X86 CPU with let say nvdia GPU to make maybe AMD fusion APU green edition ???

so because the CPU part still manufactured by AMD, they can get away with it.

maybe if VIA is decent alternative for nvdia

but i think its unlikely nvdia will do such a thing.

 

[Ferzerp]

But that's AMD having a part made for nvidia... That's not really going to happen.

Why the keen nvidia focus here?

 

[wahdangun]

But that's AMD having a part made for nvidia... That's not really going to happen.

Why the keen nvidia focus here?

but why not ?? if there is a profit to be made in there, I'm pretty sure they will do it. especially the current AMD CEO is more aggressive toward making profit for his company.

and the side effect is maybe AMD and NVDIA fanboy will unite.

 

[Tuna-Fish]

Just a little OT:

If you understand why PC lost to consoles in terms of number of gamers and game titles,

PC hasn't lost to consoles in number of gamers and game titles. The only metric where consoles did beat pc was total revenue (as in: all consoles vs pc), and pc is rapidly catching up to consoles there. If the present trends continue, pc gaming revenue will surpass consoles again by mid-2014.

The entire "PC GAMING IS DYING!!!11" was mostly bad press exaggerating bad numbers. PC gaming is objectively doing better today than *it ever has before*. Console gaming is objectively doing better today than *it ever has before.* PC gaming has a larger installed base, more gamers, more games, and a steeper growth curve. The only reason consoles (briefly) beat it on revenue is that piracy was (and is) a bigger problem on the PC, and that pc games typically cannot maintain as high ticker prices for as long.

 

[aigomorla]

What would a "big" ARM chip do?

it would let me play skyrim on my texas instruments TI-9000

:whiste:

ahahahahaha....

i think the op forgets what arm is actually suposed to do to ask for a big arm chip.

 

[dma0991]

i think it won't need 20 years to develop competitive cpu based on ARM architecture because they could borrow some patent from AMD and after the fact that ARM currently have advanced in blazing speed, i mean we can even buy quadcore phone right now.

and i think ARM currently is being held back with their ultra low power design. nad because there is absolutely no demand for high performance one

I'm pretty sure that you are aware that the performance gap between a 4 core Intel versus a 4 core ARM processor is HUGE. The addition for more cores in the handheld space (smartphones, tablets) are more of becoming a marketing gimmick rather than offering tangible performance increase.

Just a little OT:



PC hasn't lost to consoles in number of gamers and game titles. The only metric where consoles did beat pc was total revenue (as in: all consoles vs pc), and pc is rapidly catching up to consoles there. If the present trends continue, pc gaming revenue will surpass consoles again by mid-2014.

The entire "PC GAMING IS DYING!!!11" was mostly bad press exaggerating bad numbers. PC gaming is objectively doing better today than *it ever has before*. Console gaming is objectively doing better today than *it ever has before.* PC gaming has a larger installed base, more gamers, more games, and a steeper growth curve. The only reason consoles (briefly) beat it on revenue is that piracy was (and is) a bigger problem on the PC, and that pc games typically cannot maintain as high ticker prices for as long.

That in bold should probably explain what I'm getting through. Piracy has always been an issue with PC and it is definitely not slowing down. I'm 50/50 when it comes to piracy as it can be both bad and good but the rate that pirated games appear in PC is way faster than consoles. I was aware that Skyrim for PC was cracked just weeks or days after its initial release. Piracy exists for consoles too but there are definitely more steps required to pirate them.

Then there is the developers who would create games for consoles first and PC second. That was the issue I had with GTA 4 and its buggy console port and a resource hog. We're getting graphics that are still not a huge improvement as PC games are held back by console graphics.

Don't get me the wrong way, I'm all the way for PC gaming and I do not think that PC gaming will be dead but if you're a developer, wouldn't you want to create a game with a hardware that is predictable(hardly any changes) and is less likely to be pirated and have more paying customers. That is the same problem with Android vs iOS, more are willing to develop for iOS because they are more likely to get paying customers and its hardware is more predictable(restricted).

 

[Arzachel]

Consider this, the best that AMD has to offer ARM is already included in its current Bulldozer, Llano, and Bobcat products. Which of those competes well with Intel?

HSA and GPU tech in general. ARM floating point performance is cripplingly bad, so GPU cores that you can also offload fp operations to make a lot of sense. Just a nitpick, I agree with the rest of your post.

That in bold should probably explain what I'm getting through. Piracy has always been an issue with PC and it is definitely not slowing down. I'm 50/50 when it comes to piracy as it can be both bad and good but the rate that pirated games appear in PC is way faster than consoles. I was aware that Skyrim for PC was cracked just weeks or days after its initial release. Piracy exists for consoles too but there are definitely more steps required to pirate them.

Bullshit, piracy is less of a problem than treating your customers like thieving bastards and clinging to obsolete business models is. Just a few days ago Tim Schafer announced, that Double Fine will fund their next game through Kickstarter. Less than 24 hours later they had gotten more than 1 million dollars, more than twice of what they had asked for. Niche, Indie, f2p and PC-centric games are thriving while console "AAA" developers are using piracy as a scapegoat for their failures.

 

[Ferzerp]

dma0991 falls in to the trap of "all customers are willing to commit copyright infringement if it is easy". This just isn't true. Do people do it? Yes. Would they have spent the money otherwise? Rarely.

You talk about PC cracks, but miss the whole point that the only thing DRM does is kill the resale market, it does nothing to non-customers anyway. Those people will just install a patched binary that has it stripped out.

edit: I also like how he talks about skyrim "cracks". Skyrim shipped on steam with no DRM. It didn't even have steam DRM. They later encrypted the executable though.

 

[Arzachel]

dma0991 falls in to the trap of "all customers are willing to commit copyright infringement if it is easy". This just isn't true. Do people do it? Yes. Would they have spent the money otherwise? Rarely.

That's the thing, there are so many evidence of the contrary: Good Old Games, Steam sales, Humble Indie bundles and now Kickstarter.

 

[Idontcare]

HSA and GPU tech in general. ARM floating point performance is cripplingly bad, so GPU cores that you can also offload fp operations to make a lot of sense. Just a nitpick, I agree with the rest of your post.

Oh I totally agree, but again that's not what ARM is about. ARM is about selling you a license to their small core and then you, be it AMD or Nvidia, bolt on your IP...but you are not bolting it into the ARM core, you bolt it next to it.

That is HSA. That is the irony of AMD claiming they are on top of this market when ARM and its partners have been doing HSA for years. Tegra is every bit HSA as Brazos.

But you would not label Brazos as "big x86" just because it has an OpenCL compute compatible fpu graphics core bolted onto the same die, would you? Nor would you label tegra as "big ARM", but they could accomplish the same.

So I still contend we need to be clear what we are talking about here. Making a large IC that contains an ARM core somewhere within the SOC is not the same thing as making a large IC that is large because the ARM core itself has been ballooned to proportions rivaling Bulldozer and Sandy Bridge.

 

[dma0991]

Bullshit, piracy is less of a problem than treating your customers like thieving bastards and clinging to obsolete business models is. Just a few days ago Tim Schafer announced, that Double Fine will fund their next game through Kickstarter. Less than 24 hours later they had gotten more than 1 million dollars, more than twice of what they had asked for. Niche, Indie, f2p and PC-centric games are thriving while console "AAA" developers are using piracy as a scapegoat for their failures.

Just how many companies are willing to do such a business model whereby profit is not their main goal. I know for one that Minecraft is one of what you're describing. The problem is that not many have that mindset where they would pirate the game and buy the original copy later, that is also assuming that the game has replayability in the first place.

edit: I also like how he talks about skyrim "cracks". Skyrim shipped on steam with no DRM. It didn't even have steam DRM. They later encrypted the executable though.

Obviously you're not in the know to know exactly what is possible on the illegal side of things. I will assure you that Skyrim is not the first game that I have seen being pirated that fast.

 

[Tuna-Fish]

Just how many companies are willing to do such a business model whereby profit is not their main goal. I know for one that Minecraft is one of what you're describing.

In whay is profit not the main goal? Notch *alone* made more than 40M of pure profit on Minecraft last year. EA made a loss. Smaller games selling with innovation instead of expensive content seem to be a very good business proposition. Tim Schafer expects to do a profit on his crowd-funded game.

Obviously you're not in the know to know exactly what is possible on the illegal side of things. I will assure you that Skyrim is not the first game that I have seen being pirated that fast.

The point is that that's not the point. The games I like most are probably the deep historical strategy games made by Paradox. They have never had *any* copy protection. You can literally copy over the game folder and they will run. Yet, they seem to sell enough to cover their losses and release bigger and better ones almost every year now. If you make good content, people will give you money. PC games make up for the losses of piracy by shedding the middlemen in the channel -- Paradox sells it's games on it's own site. When they sell a game for 40$, their pre-tax revenue on that is more than 39$.

 

[dma0991]

In whay is profit not the main goal? Notch *alone* made more than 40M of pure profit on Minecraft last year. EA made a loss. Smaller games selling with innovation instead of expensive content seem to be a very good business proposition. Tim Schafer expects to do a profit on his crowd-funded game.

I mean that games like Minecraft are not as expensive as games like Skyrim, so a smaller profit. I do play Minecraft, legit, but if games on PC is focusing on gameplay and interaction with MMOGs, less emphasis is placed on the need for better graphics. The last few MMOGs that I've looked at is not very different from WoW style graphics. It is kinda sad that we're not focusing on having bleeding edge graphics.

 

[wahdangun]

I'm pretty sure that you are aware that the performance gap between a 4 core Intel versus a 4 core ARM processor is HUGE. The addition for more cores in the handheld space (smartphones, tablets) are more of becoming a marketing gimmick rather than offering tangible performance increase.


That in bold should probably explain what I'm getting through. Piracy has always been an issue with PC and it is definitely not slowing down. I'm 50/50 when it comes to piracy as it can be both bad and good but the rate that pirated games appear in PC is way faster than consoles. I was aware that Skyrim for PC was cracked just weeks or days after its initial release. Piracy exists for consoles too but there are definitely more steps required to pirate them.

Then there is the developers who would create games for consoles first and PC second. That was the issue I had with GTA 4 and its buggy console port and a resource hog. We're getting graphics that are still not a huge improvement as PC games are held back by console graphics.

Don't get me the wrong way, I'm all the way for PC gaming and I do not think that PC gaming will be dead but if you're a developer, wouldn't you want to create a game with a hardware that is predictable(hardly any changes) and is less likely to be pirated and have more paying customers. That is the same problem with Android vs iOS, more are willing to develop for iOS because they are more likely to get paying customers and its hardware is more predictable(restricted).

I'm not saying nvdia tegra 3 have better performance than their latest quadcore, the thing is if nvdia could build ARM cpu that is fast enough to play some latest games.

Oh I totally agree, but again that's not what ARM is about. ARM is about selling you a license to their small core and then you, be it AMD or Nvidia, bolt on your IP...but you are not bolting it into the ARM core, you bolt it next to it.

That is HSA. That is the irony of AMD claiming they are on top of this market when ARM and its partners have been doing HSA for years. Tegra is every bit HSA as Brazos.

But you would not label Brazos as "big x86" just because it has an OpenCL compute compatible fpu graphics core bolted onto the same die, would you? Nor would you label tegra as "big ARM", but they could accomplish the same.

So I still contend we need to be clear what we are talking about here. Making a large IC that contains an ARM core somewhere within the SOC is not the same thing as making a large IC that is large because the ARM core itself has been ballooned to proportions rivaling Bulldozer and Sandy Bridge.

oh I'm understand now, so what do you mean is, you can't change/ modify their core architecture you can only include it in your SOC is that true ?