News about Rambus and the PS3

[jvd]

The whole point of a console coming at a higher price is to increase the power. If your suggesting that Microsoft would match the PS 3 hardware specs with a lower priced console, I say no way. More powerful specs costs money, they would be losing even more money then they are now.

A $399 PS 3 is a nightmare situation for MS in my opinion. Sony has the hype and momentum to justify a machine at this price to the public. Microsoft would have a hard time convining people to plunk down $399 for its console after the inital release. What does MS do? Stick with $299 and a lower speced machine? Go higher with a $349 or $399 and try to match the PS 3?

You seem to miss my point . Price is all that matters. Yes i think ms can match the sony specs at the same price point or under. You don't know what ms has up its sleeve in terms of cpu , gpu , ram , sound . We know sony is going to use a cell based chip . We know they are most likely going with rdram again in the form of this xdram. But the topic is sony. Sony does not want to loose alot of money on this console because unlike ms this is thier bread winner. So sonys buisness plan is to make no money on the system and mabye loose a little money. But they want to get it quickly to where the system is breaking even or making money. This is so they can drop the system price . Remember nintendo doesn't go for top of the line systems but rather like to stay in the running and have a cheap selling console . Ms likes to spend money and wants the best specs for the machine. So sony knows they have to fall in the middle. They can't afford to have the weakest system since they don't have a built in fan base that would buy the system regardless. They also can't afford to loose alot of money on the system at the start because they don't have huge pockets and other sections of the company that will make enough money to hide the fact or at least make it look like its not loosing as much money as it really is .

So basicly sony has to balance the system. So yea a 400$ ps3 might be hard for ms to fight. But if 150$ of that system cost is a blue ray drive then no it wont be. Because ms can throw in a ten dollar dvd drive and throw in faster chips with that 140 bucks . So you see it all comes down to cost.

 

[Brimstone]

I agree blu-ray at a high cost is a total waste.

Microsoft has yet to prove they can sell systems. Why give MS a chance by going head to head with them at $299? Raise the stakes to $399 and see if MS can hang on. Sony won't have any problem selling every unit they make for the first 18 months at $399. Microsoft at $399 I'm not so sure. If their sales start to slow down and have to lower their price to spur sales it will start a vicous cycle for MS.

 

[jvd]

I agree blu-ray at a high cost is a total waste.

Microsoft has yet to prove they can sell systems. Why give MS a chance by going head to head with them at $299? Raise the stakes to $399 and see if MS can hang on. Sony won't have any problem selling every unit they make for the first 18 months at $399. Microsoft at $399 I'm not so sure. If their sales start to slow down and have to lower their price to spur sales it will start a vicous cycle for MS.

Sorry but you don't sound like you have any clue as to what your talking about. Ms is already out spending sony on every system . Its been that way since the launch of the xbox. The price of retail is not what is being spent on these systems. Sony will have problems above the 300$ mark. Remember the ps3 will go up against the gamecube , ps2 , xbox , xbox2 , gamecube 2. To the biggest extent it will have to go against the huge sales of the ps2. If the system costs to much people will continue to play thier ps2s . Games will most likely come out for the ps2 for at least a year after the ps3 comes out. There is just to huge of a install base to pass it up. Also if the ps3 is plauged with sequals and bad games (it could very well happen. At launch there is normaly 1 or 2 games that are system sellers the rest are just filler. So you see only sony will loose if they spend to much money on the console and raise the price to high.

 

[Fafalada]

To the biggest extent it will have to go against the huge sales of the ps2. If the system costs to much people will continue to play thier ps2s . Games will most likely come out for the ps2 for at least a year after the ps3 comes out. There is just to huge of a install base to pass it up. Also if the ps3 is plauged with sequals and bad games (it could very well happen. At launch there is normaly 1 or 2 games that are system sellers the rest are just filler.

It's been 3 years and PSOne games are still coming out iirc (albeit very seldomly). Not to mention even last year PSOne software shipments worldwide outstripped all other consoles except PS2 - and I don't think anyone would argue PS2 was at any point slowing down because of its predecessor
And PS2 "did" launch at almost 400$ in Japan
Anyway, PS2 would most likely hurt PS3 sales if they don't make it backward compatible, otherwise I doubt it could happen.

 

[Vince]

Hey I like Cell, and im well aware of its advantages.

You don't have to justify your stance, nor did I intend for it to appear that I was *pushing* cell as we're talking in a more generic light. Although, I think I may shift from theory to praxis and use Cell for a thought experiment later on.

However, quantitatively it strikes me that we are hitting a point in graphics where bandwidth is less and less important, and computation power more and more (one of the advantages of Cell IMO). I am not arguing against the fact that more bandwidth is better and that the sea of processors needs to be perpetually filled. But we also have to identify that it assumes defacto that the bottleneck is bandwidth, which need not be the case.

I fully agree.

Ultimately, mathematically lets look at the problem we need to solve. We have a set amount of pixels to fill, a set amount of texels to fetch. A large number of tasks that were once the primary bandwidth hog, have hit a point where we simply do not need all that more to reach a satisfactory solution or at the very least the problem has moved to a different stage (eg framebuffer vs texture bandwidth).

The future trends in gfx seems to be doing a lot more math per pixel (that seems to be growing fast), rather than fetching and writing (which seems to have stayed roughly constant) Eg, the calculations are by and large growing in the shader stage (this irregardless of architecture).

I for one don't even factor in bandwith hits from 'dumb' operations like filtering or framebuffer as it's entirely predicable and basically static as you mentioned.

And it would be inconsistent of myself to state that the burden hasn't shifted to computationally intensive tasks in the current DX9 era as opposed to <DX9 in which bandwith was paramount. I've fought this concerning the NV30 and... it's true.

But, I think we can agree that quantitatively there is a minimalist level of bandwith that is necessary to keep the bottleneck computational. Granted the level is skewed heavily in favor of "the math," but there is a threshold.

My argument comes in that Cell for PS3 will infact be nearing this minimalist threshold due to it's inherient design which is so heavy in FP. On PS2 we saw that for it's theoretical 6.2GFlop, there was what 3.2GB/sec? On PS3 it could very well be 1,000 theoretical GFlops (I said it would be possible back in 2000 when everyone here told me it wasn't - I'm sticking with it) with bandwith of a mere 30GB/sec.

Thus, we'll be seeing the shift to raw calculational ability you spoke of - but I don't believe that added bandwith wouldn't be adventageous in a design that's as open ended as Cell. I allways imagined PS2 programming as playing hot potato with data; I don't think this is going to end - just get wider.

 

[jvd]

To the biggest extent it will have to go against the huge sales of the ps2. If the system costs to much people will continue to play thier ps2s . Games will most likely come out for the ps2 for at least a year after the ps3 comes out. There is just to huge of a install base to pass it up. Also if the ps3 is plauged with sequals and bad games (it could very well happen. At launch there is normaly 1 or 2 games that are system sellers the rest are just filler.

It's been 3 years and PSOne games are still coming out iirc (albeit very seldomly). Not to mention even last year PSOne software shipments worldwide outstripped all other consoles except PS2 - and I don't think anyone would argue PS2 was at any point slowing down because of its predecessor
And PS2 "did" launch at almost 400$ in Japan
Anyway, PS2 would most likely hurt PS3 sales if they don't make it backward compatible, otherwise I doubt it could happen.

Thank you for proving my point , you can add psone to the list of what its competeing against . Btw I'm talking usa price points . In japan it be 500$ usa if over here it launched for 400$.

 

[jvd]

And it would be inconsistent of myself to state that the burden hasn't shifted to computationally intensive tasks in the current DX9 era as opposed to <DX9 in which bandwith was paramount. I've fought this concerning the NV30 and... it's true

To be fair there really is nothing out there that uses dx9 . So we are just seeing dx 9 cards work faster on games that were programed for lots of bandwitdh. In the future we will see the shift. I agree that less bandwitdh is needed. I also agree taht from wht we know of the ps3 design it doesn't need more than a 128 megs system ram.

 

[Fafalada]

Thank you for proving my point , you can add psone to the list of what its competeing against . Btw I'm talking usa price points . In japan it be 500$ usa if over here it launched for 400$.

Fair enough. Although once PSTwo is at <100$ PSOne will become obsolete. And you'd again have two price classes which worked well enough to avoid products competing with each other in the past.

But, I think we can agree that quantitatively there is a minimalist level of bandwith that is necessary to keep the bottleneck computational.

Uhm... I don't think you can really talk about 'tresholds' by ignoring the software factor.
For instance - the proposed Cell design working as a Reyes renderer would spend majority of its time with APUs processing in their local memory, generating much bigger output data streams then input from main memory pool - and would hence spend majority of its time in eDram as well.
On the other hand a traditional polygon pipeline throwing around similar amount of geometry on the far end would be pulling all that data from main memory, making it a decisively more bandwith hungry situation.
All while you have comparable calculation load for both.

 

[Panajev2001a]

Bandwidth is still a concern... in IBM papers talking about Cellular architecture they were specifically mentioning that the growth in applications' data set has been massive and so has been the increase in computational power trying to meet those needs... they were saying that the problem was becoming more of a "data management" problem than a purely computational one...

Guess why we have relatively large amounts of e-DRAM in Cell... they provide lower latency ( compared to off chip DRAM ) and much higher bandwidth and they help killing the bottleneck that was related tgo the slow main memory unable to keep up with the CPU...

System RAM has not grown as fast as CPU's computational power they sustained and that was becoming a big problem hence the decision of moving to a kind of architecture like Cell and to finally be able to move main memory on the chip...

Technically PlayStation 3 will have external RAM and not off-chip main RAM... if we define main RAM as where the code needs to be to be executed we shifted from external RAM to Local Storage ( SRAM ) on chip from which instruction fetch takes place and execution of said instruction commence...

Sorry, I digressed...

With that said, I am less scared about the external RAM bandwidth as we do not need 50 GB/s, the CPU is not feeding 1 TFLOPS of computational power straight from that memory pool... we drive the data we need from the Local Storages and the e-DRAM... the external memory is there to keep the on chip storage full by streaming in data fast enough not to let that on-chip storage empty of data... and I believe that 25.6 GB/s are fast enough for the job.

 

[Panajev2001a]

For instance - the proposed Cell design working as a Reyes renderer would spend majority of its time with APUs processing in their local memory, generating much bigger output data streams then input from main memory pool - and would hence spend majority of its time in eDram as well.
On the other hand a traditional polygon pipeline throwing around similar amount of geometry on the far end would be pulling all that data from main memory, making it a decisively more bandwith hungry situation.
All while you have comparable calculation load for both.

Even more reasons why going by the current PlayStation 3 architectural approach they should "advice" developers to implement a REYES-style renderer in their engine to maximize performance, maybe up to the point of delivering to developers a VERY optimized ( for the PlayStation 3 architecture and real-time rendering ) implementation of such a renderer ( it would come with their SDKs )...

Developers could still use traditional approaches to rendering and it woul still be very fast, but very expert ones like Square and Konami would go for the most optimal route and probably even create their own REYES-like renderer...

 

[Panajev2001a]

I would never expect this comment... out of you.

I think his comment is fair. With sufficient caching, which is what the PS3 should have with all the embedded RAM, more memory (in this case the lowest level in the storage hieracrchy before we hit disks) is a good thing.

Yes and this can link to the longer post I just made in this thread about Cell...

Same, it is not like you and Vince are disagreeing though... so I can agree with both of you as you both recognize the bandwidth as being an issue and how Cell has dealt with it so far ( as I tried to write briefly about in my post )...

 

[wazoo]

Remember the ps3 will go up against the gamecube , ps2 , xbox , xbox2 , gamecube 2. To the biggest extent it will have to go against the huge sales of the ps2.

The only real oponent of the ps3 will be the ps2, others will be either already dead (killed by the ps2) or in their infancy (with much more to prove than the ps3). and the ps1/ps2 transition already has shown that it is not a problem for Sony, much more the opposite (ps1 sales went from 50 to 100M after the ps2 announcement and we have not seen a sign of something slowing down the ps2 sales except Sony production capacity).

 

[marconelly!]

On a console, we are fixed at low resolution, and already with todays minimum spec bandwidth (like xbox) we can draw a scene 5 or 6 times over at 30hz.

It seems to me we are more or less set for framebuffer bandwidth, even with antialiasing. Texture BW might increase, but probably not at the same rate as computation requirements do.

Is 1080p with 2-4X antialiasing a 'fixed low resolution'?

Besides, surely you don't think the quality of graphics that xbox produces is the pinnacle of what is possible and all we have to go for now is more memory?

Polygon counts in today's games are pathetic comparing what they need to be to reach anything resembling the CGI footage. Same for the amount of particles, same for the shader complexity, same for the overall scene complexity (number of moving objects) and AI.

Do you really think that the ability to re-render the scene 5-6 times that today's console have is enough to make anything resembling the complex look of today's CGI?

Speed is a huge issue with todays graphics, much more than memory constraints, IMO.



Btw. can Xbox really re-render the typical game scene 5-6 times in 1/30 second? I've thought the developers tend to do as much as possible in one pass because the hardware doesn't behave well when it comes to re-drawing.

 

[Panajev2001a]

Marconelly, a food for thought...

For instance - the proposed Cell design working as a Reyes renderer would spend majority of its time with APUs processing in their local memory, generating much bigger output data streams then input from main memory pool - and would hence spend majority of its time in eDram as well.
On the other hand a traditional polygon pipeline throwing around similar amount of geometry on the far end would be pulling all that data from main memory, making it a decisively more bandwith hungry situation.
All while you have comparable calculation load for both.

Even more reasons why going by the current PlayStation 3 architectural approach they should "advice" developers to implement a REYES-style renderer in their engine to maximize performance, maybe up to the point of delivering to developers a VERY optimized ( for the PlayStation 3 architecture and real-time rendering ) implementation of such a renderer ( it would come with their SDKs )...

Developers could still use traditional approaches to rendering and it woul still be very fast, but very expert ones like Square and Konami would go for the most optimal route and probably even create their own REYES-like renderer...

 

[jvd]

Is 1080p with 2-4X antialiasing a 'fixed low resolution'?

considering when the system comes out mabye 90% of the target audicance wont be able to play at that res i doubt it matters

 

[marconelly!]

considering when the system comes out mabye 90% of the target audicance wont be able to play at that res i doubt it matters

Well, you could say the same for progressive scan support or surround audio, yet many games support it (on Xbox it's almost mandatory) and for many people it matters and a lot.

Next generation is only going to up the ante and I'd be extremely surprised if HDTV resolutions support will not become recommended if not mandatory.

 

[jvd]

considering when the system comes out mabye 90% of the target audicance wont be able to play at that res i doubt it matters

Well, you could say the same for progressive scan support or surround audio, yet many games support it (on Xbox it's almost mandatory) and for many people it matters and a lot.

Next generation is only going to up the ante and I'd be extremely surprised if HDTV resolutions support will not become recommended if not mandatory.

hdtv support will not be mandatory it will just be an option . Whats the point of putting out a new system if no one can play it or they need to buy a new 500$ tv . Its not going to fly well. The gen after this will be the hdtv gen .

 

[Brimstone]

I agree blu-ray at a high cost is a total waste.

Microsoft has yet to prove they can sell systems. Why give MS a chance by going head to head with them at $299? Raise the stakes to $399 and see if MS can hang on. Sony won't have any problem selling every unit they make for the first 18 months at $399. Microsoft at $399 I'm not so sure. If their sales start to slow down and have to lower their price to spur sales it will start a vicous cycle for MS.

Sorry but you don't sound like you have any clue as to what your talking about. Ms is already out spending sony on every system . Its been that way since the launch of the xbox. The price of retail is not what is being spent on these systems. Sony will have problems above the 300$ mark. Remember the ps3 will go up against the gamecube , ps2 , xbox , xbox2 , gamecube 2. To the biggest extent it will have to go against the huge sales of the ps2. If the system costs to much people will continue to play thier ps2s . Games will most likely come out for the ps2 for at least a year after the ps3 comes out. There is just to huge of a install base to pass it up. Also if the ps3 is plauged with sequals and bad games (it could very well happen. At launch there is normaly 1 or 2 games that are system sellers the rest are just filler. So you see only sony will loose if they spend to much money on the console and raise the price to high.

The biggest mistake Sony can make is to let Microsoft get a foothold in the market. Once Microsoft gains traction they will never go away. In my view this next console generation isn't about making a profit, it's about Sony defending its kingdom from Microsoft at all costs. This is the huge mistake Nintendo made against Sony. They didn't defend against Sony strong enough.

Sony has to bring everything its got to the battle. They have to have a lust to spill Microsofts blood.

Why let Microsoft come in equal to Sony? Going head to head against MS is dangerous. Rewrite the rules and launch at a higher price and hype the hell out of what the PS 3 will offer. Seize the iniative and have MS react to the actions of Sony.

 

[marconelly!]

hdtv support will not be mandatory it will just be an option . Whats the point of putting out a new system if no one can play it or they need to buy a new 500$ tv . Its not going to fly well. The gen after this will be the hdtv gen .

As I've said, 480p and surround sound support is already *almost* mandatory in Xbox games made today (almost, because there's probably 1% games that don't support one or the other)

Going into the next generation, I cannot imagine 480p not being a requirement for all the console games, and 720p being a very recommended option. There's just no reason not to do it, as those resolutions are being supported even today.

*Obviously*, standard interlaced output will be there by default, it's just that games more likely than not will need to have support for HDTV resolutions also.

 

[PC-Engine]

hdtv support will not be mandatory it will just be an option . Whats the point of putting out a new system if no one can play it or they need to buy a new 500$ tv . Its not going to fly well. The gen after this will be the hdtv gen .

As I've said, 480p and surround sound support is already *almost* mandatory in Xbox games made today (almost, because there's probably 1% games that don't support one or the other)

Going into the next generation, I cannot imagine 480p not being a requirement for all the console games, and 720p being a very recommended option. There's just no reason not to do it, as those resolutions are being supported even today.

*Obviously*, standard interlaced output will be there by default, it's just that games more likely than not will need to have support for HDTV resolutions also.

Xbox games are 480p because the GPU renders progressive natively NOT because they think a lot of people will have HDTV hooked up to Xbox

DC has native progressive output too, but most people had it hooked up to normal interlaced tvs NOT computer monitors.