Revolution GPU and CPU STILL in development.

[randycat99]

Xbox had a rather poor thermal design, wasn't it pretty much built like a sideways PC case? I'd say gamecube has much better airflow than xbox, gamecube is very wide open inside compared to xbox.

Perhaps [about the merit of Xbox's layout], but my point in that regard was that just the nature of the components alone dictated considerably greater heat dissipation to address than in the GC.

I thank you for the extra information you supplied on the GC, though.

 

[BlackNMild2k1]

Been reading the Beyond3D forums for about 5 years now but I haven't registered till now

But I noticed that you guys haven't discussed the supposed leaked Rev specs that are floating around

posted by Han_Solo on G4 forums (he supposedly works for Factor 5)
http://forums.g4tv.com/messageview....403677&FTVAR_MSGDBTABLE=&STARTPAGE=13

-----------------------------------------------------------

Alright, here are some specs from my sources about Revolution. Its still not the complete or "Satisfied" specs, but they are the numbers that would rival xbox 360 and PS3, like i have been saying all this while.....

Revolution Specs

CPU: IBM Custom PowerPC 3.5 GHz + 4 internal Power PC G5 cores running at 2.5 GHz each. Each core will have 128 KB or 256 Kb L1 cache. The whole CPU will share 512 KB - 1 MB of L2 cache. As you can see they are deciding wether to go up one notch. But never the less, its a very powerful CPU with 4 custom G5 cores.

There will also be two hardware threads per core, 8 hardware threads total.

12 billion dot product operations per second

Theoritical of 10 GHz total + 3 GHz CPU speed

Xbox 360 CPU:
Three symmetrical cores running at 3.2 GHz each = theoritical of 9.6 GHz + 3.2 GHz CPU speed
Two hardware threads per core; six hardware threads total
VMX-128 vector unit per core; three total
128 VMX-128 registers per hardware thread
1 MB L2 cache

Revolution GPU

ATI Custom based RN520 core. the "N" stands for nintendo, and is because the ArtX team is with them, thats why its an "N". There will be 2 GPU cores (just like the nVidia SLI motherboard with two Graphics Chips), this will use ATI's alternative, and will be the first in any console.

Both GPU's will have 256 MB's of GDDR4 memory, with an addional 16 MB of eDRAM total. eDRAM is an onboard flash memory, just like the 3 MB on the gamecubes flipper.

The cores will run at 600 MHz each, (rumours are that its possible 500 MHz each), but i doubt that.

24-to-48 way parallel floating-point dynamically scheduled shader pipelines for each core with unified shader architecture.

Polygon Performance: 500 million triangles per second theoritical, average ingame would be around <100 Million/sec

Shader Performance: 48 billion shader operations per second

Revolution memory

512 MB of 700 MHz Updated 1T-SRAM (its a totally redone design of something new, remember the nitrous i talked about)

thats it for now, i may addd some more later

-------------------------------------------------------------------

ATI and IBM have the coolest running with copper inductors. ATI also got their prototype of an aluminium based solution by Sapphire, thats cooler than water based cooling systems.

So all this will be with it to make it cheap

P.P.S about the aluminium cooling, i meant this

link : http://www.sapphiretech.com/vga/blizzard.asp

-------------------------------------------------------------------
Also, as I said I’m going to add more info, the Revolution will support a PPU chip (Physical Processing Chip). This chip is very new to computer architecture, and it will mainly help in the physics area. There will be 32 MB’s of its own RAM, which will link to the CPU and GPU. To compare to the usage of physics used in current games, Half Life 2 only used 5 MB’s, Rebel strike used around 1.3 MB’s, and RE4 used 3 MB’s, but this is off main memory, which made performance issues.

There will also be a separate sound card that will support only DD 5.1 – DTS 7.1, rumours has it will have 16 MB’s, like the Cube DSP.

--------------------------------------------------------------------

This is some of my info on the PPU chip.
The PPU was built by the company Ageia, and is named the PhysX chip
PhysX is the world's first Physics Processing Unit (PPU), an entirely new category of processor that promises to revolutionize gaming in the physics area. By offloading software physics processing from the CPU and GPU, the PPU completes the triangle of gameplay, graphics and physics, balancing the load of these processing tasks and enabling pervasive interactive reality in games.
PhysX Processor Architecture has been designed to enable radical acceleration of:
· Rigid body dynamics
· Universal collision detection
· Finite element analysis
· Soft body dynamics
· Fluid dynamics
· Hair simulation
· Clothing simulation

There is an SDK program that runs with PhysX and is as follows:
NovodeX Physics SDK
This program called the NovodeX Physics SDK, is a stable, high-performance solution for game developers to enable physics-based gameplay and effects in PC and console titles. This has been used in Half Life 2 extensively and Far Cry. But CPU power handling this and other things isn’t good compared with a CPU doing what it was mean to do, like data flow, and a PPU with this program handling just physics (and can also add physics to the sounds produced), makes the games kick ass. A powerful API for the PhysX PPU, NovodeX enables game developers to inject both software-only and hardware-accelerated features into their games. The NovodeX Physics SDK is also the first and only asynchronous (multithreaded) physics API capable of unleashing the power of multiprocessor gaming systems, just like Xbox 360, PS3 and the Revolution.

-----------------------------------------------------------------

thats everything he posted about Rev so far. Just wanna see what you tech heads @ B3D forums think of this stuff and if you call BS

and here is a second source for the specs (not as much detail)
http://www.techtree.com/techtree/jsp/article.jsp?article_id=3956

P.S. in another post he says the Factor 5 still has a really good relationship w/ Nintendo

edit: fixed spelling error

 

[JasonLD]

That revolution spec looks like Nintendo fan's wish list unless Revolution comes out at least a year later than PS3..

 

[randycat99]

I'm guessing that a rig like that running below 50 W might be just a tad optimistic?

 

[PC-Engine]

This analogy assumes that the output can go to infinity (i.e. no constrictions), which is not the case (haha, no pun intended) of what we are talking about.

For instance, take the water pipe example and attach it to a case where the case has a single outlet.

Case A with wide pipe fills up and water begins to flow out through the outlet.

Case B with small pipe fills up and water begins to flow out through the outlet.

If the cases are the same and the GPM of the pipes are the same, then the water pressure out of the outlet (the turbulence and therefore "noise") is likely the same because the average ambient pressure within the case are about equal. Note that I am NOT a fluid dynamics guy so I could be make incorrect assumptions.

But the flaw in this analogy is leaving the air resistance and hence pressure up to flipping a coin. Yes if you have a faucet, the faucet will be the limiting factor in terms of resistance and therefore airflow pressure, but this is not how smart cooling designs work. Designers do not simply leave the airflow path to fipping a coin.

 

[Teasy]

I don't get that. Why would it have a 3.5Ghz CPU with four custom 2.5Ghz cores on the same chip?, sounds wierd to me. Not that it matters since the specs are obviously a fantasy. I mean, a 160 GFLOP CPU, dual 48 way shader GPU cores with 256MB ram per core and 16MB eDRAM, plus 512MB main ram and a ageia PPU with 32MB ram.. That's absolutely crazy.. That system would whipe the floor with XBox 360 and PS3 combined.

 

[randycat99]

...but this is not how smart cooling designs work. Designers do not simply leave the airflow path to fipping a coin.

Often times the desired size of the channel reaches an upper physical limit, after which it becomes an exercise of how much air velocity can you tolerate at the max. channel size. To assume that the upper physical limit is never reached or needed for even moderate heat dissipations is a bit fanciful.

...or it comes down to you simply cannot fit any more heatsink fins within the available space, so your only open option is to boost air velocity to boost heat dissipation (to the target capacity). Again, it comes down to how much air velocity you can tolerate from a noise standpoint.

 

[PC-Engine]

...for a limited power output. That's the key. Now if you are alluding to the idea that Revolution will consume no more power than GC, then by all means, your idea has merit (anybody know what the power consumption is on a GC, anyway?). If it will be dealing with 200/300/400+ watts, that little "quiet-blow" tunnel is not going to work. At the least, it will become a "loud-blow" tunnel, and more likely will need to be a larger tunnel altogether (including larger heatsinks within, naturally).

Huh? I'm not comparing power output of GCN vs Revolution Sherlock. I'm comparing the PROCESSING power output of GCN vs Xbox hence Revolution vs Xbox 360. Are you intentionally dodging or is there an honest reading comprehension problem as jvd alluded to?

You need to quantify how much power it is dissipating. Are we talking about 40 W, 60 W, 200 W??? In one case, it is completely understandable how it does the cooling in a quiet little package. In the other, your bringing it up becomes utterly irrelevant wrt the requirements that Revolution will present.

Uh no I don't, read above. Are you aware that Xbox 360 is smaller than Xbox? Are you aware that it will also be more quiet? How is that possible?

PPC chips typically use far less power than their x86 counterparts (back in "the day", at least). So your "timing of the hardware" argument is questionable with respect to the "nature of the hardware". Also consider the GPU is a (relatively) tiny little ArtX job, not quite on the same scale as whatever behemoths ATI was building at the time for the PC space (this referring to die size and sheer power consumption, not feature set). So there you have it right there- the build-up of GC included a very conscious effort to use thermal power-miserly type of components. The windtunnel cooling feature is quiet because the thermal load (which will dictate the design) isn't particularly tremendous in anyway. The Xbox is seemingly on the opposite extreme from this (not to say it's power consumption was out of line, however). So you see, the thermal power considerations between an Xbox and a GC are not exactly equivalent.

LMAO, riiiight. Read above. Comparable PROCESSING POWER yet smaller and less heat output. Get it yet?

...unless the heat dissipation demands rise considerably, then confined space becomes a crucial factor. Now if we could get some feedback on what sort of power consumption will be associated with Rev, the logistics of the thermal management will be more clear. Lacking this info, are we to assume that you imply Rev will use no more power than a GC (within 10-20 watts, to be open about it)?

Depends on what the definition of "considerable" is. Fox5 has already shot down your argument. A laptop is outputing near 200W running on AC in an extremelely low profile design without the need for a friggen 50dB snowblower.

That is building my point that the heat dissipation requirements where not particularly tremendous in the first place. Naturally, it can be a quiet affair.

And that's building my point where comparable PROCESSING power doesn't equal comparable heat output.

You suggest there is room for more fins. I don't doubt that, at all. Guess what happens to "drag" as fin surface area goes up? drag ==> pressure loss ==> need more "fan blowage" ==> highly likely more noise

You don't know much about aerodynamics do you? More fins means only slightly more drag in the case of GCN since the length of the fins are parallel to the airflow. You can even make the fins thinner (less drag) or taller (more surface area) or both instead of adding more in total numbers to lesson the increase in drag even further. Thinking stupid isn't a solution.

Alternately, you could just keep the same fin surface area and increase flow rate/air velocity. Very much the same outcome will result- more drag ==> pressure loss ==> more fan blowage ==> highly likely more noise.

Uh not quite. Taken to extremes if you have a big enough heatsink, you will not need a fan. Heck that is the basis for designs such has the Hush PCs. No fans means NO NOISE..Got it??? Not only that but it's flawed to think faster air flow means better cooling. The limit is the rate at which the heat from heatsink is taken away. This is not entirely up to the fan. Taken to extremes if your heatsink is is tiny, no amount of air would be sufficient enough to overcome the limitation of the heatsink. That's why HSF cooling efficiency need to be calculated as a whole not individually. Wouldn't it be funny if you hooked up a motorcycle engine to a bicycle trainy?

Liquid metal cooling is beside the point when discussing pure heat dissipation at the endpoint. You can't even argue that much that it keeps the processor particularly "cool" if it has to literally reach a temperature to melt metal (albeit, a low-melting point metal, but certainly not "cool" by any stroke of imagination) for the whole thing to work. This is still a heat relocation measure, not a heat dissipation measure. It doesn't much matter if you have to dissipate 100 W locally right off the processor or 4" away on a remote heatsink. You still have to dissipate 100 W. Ideally, you have good air flow management implemented in either case, so there is no great disadvantage or advantage whether it is local or remote dissipation. It's still 100 W you have to worry about.

You don't get it do you? I'm very aware of the difference between heat dissipation vs heat transport contrary to what you would like to believe LMAO. You can't just talk about the endpoint to suit your own argument and ignore the heat being dumped back into a case in a DUMB design like Xbox and standard PC designs. Go back to page 3 or 4 where I first mention the liquid metal cooling solution and the flexibility it allows in terms of routing flexibility. It's kinda amusing watching you shift back and forth with little room you have left. If the liquid metal system needs to be at a high temperture to flow/work then all you need to do is enclose the apparatus in an air chamber that's connected to the exhaust port fan so the air can come in at the inlet port close to the processors and out through the exhaust port close to the outside of the case similar to the GCN. This will prevent the dumping of heat back into the case unlike the standard "HSF over processor" cooling design. With the standard design you need massive airflows just to get the heat out of the case because it has to fight all sorts stagnant hot air inside the case.

This is a bobo premise, altogether, because you ignore the nature of the hardware (PPC750 vs. Celeron at almost twice the clockrate, ArtX GPU vs. nVidia's finest, HD, etc...), . Now if Nintendo had managed to put all the Xbox hardware in a GC case, and made it work...that would indeed be an enviable feat. Beyond that, "timing" is utterly meritless.

Yes it's bobo because you don't have an answer. The whole point of this discussion is getting a comparable amount of processing power into a smaller case. This has been done with GCN regardless whether it's using a PPC or not. Sh*t if everbody had to use the same architectures and the same cooling systems we wouldn't be discussing this. Unfortunately for you that's not reality.

You may not need a lot of airflow going throughout the case, but you will at the primary heatsink, regardless of its location. 100 W of dissipation remains as 100 W, 200 W is still 200 W, 300 W is still 300 W. Thermal heat doesn't magically disappear just because you relocate it through a conduit. To top it off, you still have one very hot processor inside your case, if you expect it to melt metal to make the liquid metal component work in the first place. So you end up needing a case fan to evacuate incidental heat inside the case, anyway, in addition to the one blowing on the remote heatsink.

And who said you don't need a heatsink??? I think it is you who are arguing with Mr. Scarecrow.

Ref'ing your own disputes is comical, to say the least. You've shot down nothing other than revealing to us your complete inability to discern between heat distribution and heat dissipation. Potentially jvd has a similar lack of understanding, hence his peculiarly cryptic manner of backstepping out of this "argument".

Heh what's comical is the fact you keep shifting back and forth ignoring all the facts presented before you and desparately hanging on to your strawman argument.

What is the "unsmart" way you refer to? ...to haphazardly blow air indiscriminately inside a case? You have revealed your strawman argument then, because no one was advocating that here in the first place. I believe I remarked specifically on that in one of my posts that a good airflow management should be "standard design" (whether it be in a windtunnel or point-to-point-to-point inside a case or whatever the configuration may be). Any way you go, you want cool "fresh" air to hit your heatsinks (whether local or remote), exhaust air to be expelled asap, and as little mixing of the 2 as possible). This is NOT exclusive to windtunnels, at all, and doesn't change the fact that if you have x watts to dissipate from the processor, you can count on needing to dissipate x watts on the heatsink (remote or not).

An unsmart design is where you flip a coin and accept whatever measely airflow passage is formed including the associated *drag* that comes with it and simple slapping a huge HSF over the processor. It's funny that your memory has suddenly pickup on other people's ideas instead of your flawed one to try and shift the balance. Boy talk about goal posts, your's seems to have quantum states.

Now if it was your (you and jvd) intent to make some comparison between the "worst intake/exhaust fan slapped onto a case" design you can think of to the best discrete windtunnel or remote heatsink design you can think of (completely ignoring ultimate thermal targets, as well), then that was a pretty bloody pointless point to make in the first place. Compare "best of" to "best of", if you have an ounce of sportsmanship in you- naturally, that means an appropriate and effective airflow management system will be in place for either example, period.

You seem to have forgotten the whole point of this discussion, but since I'm such a nice guy exhibiting good sportsmanship, I'll reiterate it main issue for you. Nintendo has shown that it can design a small console with comparable processing power to the Xbox. That is the point. Artificially restricting the comparison down to slapping the innards of an Xbox into a GCN to make your point was never the issue. However if you wish, you can continue to argue that point with Mr. Scarecrow in your spare time.

How many watts of dissipation do you honestly think a GC-sized case can accomodate and still be "quiet"? 60 W? 100 W? 250 W? This will be interesting to get you on record here... If you want to get silly with a range, consider the continuum of a 1500 W hair dryer (fairly skimpy on net case volume), but damn noisy vs. a 100 W Xbox (my guess, feel free to substitute a more appropriate value) which is indisputedly larger and quieter. Where do you think a GC lies? How about Rev? (Yeah, the hair dryer is a pretty extreme example for the upper bounds, but gets the point across that it's pretty easy to get noisy pretty quick when case size goes down and heat dissipation goes up- get this...it's also a pretty darn good example of a "windtunnel", just not a quiet one)

I'm not even gonna comment on the hair dryer because the fan in a hair dryer has a different purpose than a fan in a HSF. The only relevent thing is that a powerful laptop produces a lot of heat, yet it has a very small volume to work with.

Once again, you have confused air flow management with final heat dissipation. This is apparent to "anybody" who knows anything about computers. One day, you may join them, but for now...back to the books for you!

Nice attempt at trying to dodge the main issue. You would make a great politician. Air flow management and heat dissipation are interconnected contrary to what you think, but I'll let you figure that out since by your own words you've already joined the ranks of people who know.

You really don't see the association, do you? This severely demonstrates your inadequacy on the matter- seriously.

I'll start worrying about Helmholtz, when it actually starts to make a difference.

Ah...no. The size of the case is irrelevant. You are moving air at the rate of CFM. By the wording you chose, it is the same. Now air velocities inside a small case vs. a large case may be different...

That was my point ie you'd be cycling the case volume many more times in a smaller case than a larger one with a certain CFM rating. Funny how you try to make it sound like that wasn't the point in an attempt to lesson the damage.

Is this one of those "sky is blue" comments? Big deal! You also have the option to use a much bigger pump on the 100 gallon tank w/o turning it into a geyser, too. Horses for courses...

Of course it's a big deal Sherlock. Let me know when Nintendo puts in a 120mm fan with the associated gaping hole in Revolution.

...but this is not how smart cooling designs work. Designers do not simply leave the airflow path to fipping a coin.

Often times the desired size of the channel reaches an upper physical limit, after which it becomes an exercise of how much air velocity can you tolerate at the max. channel size. To assume that the upper physical limit is never reached or needed for even moderate heat dissipations is a bit fanciful.

...or it comes down to you simply cannot fit any more heatsink fins within the available space, so your only open option is to boost air velocity to boost heat dissipation (to the target capacity). Again, it comes down to how much air velocity you can tolerate from a noise standpoint.

And of course that limit is unkown. Using Xbox or Xbox 360 as an idication of that limit is not wise.

 

[jvd]

this was just emailed to me

Been reading the Beyond3D forums for about 2 years now but still havn't registered.

But I noticed that you guys haven't discussed the supposed leaked Rev specs

posted by Han_Solo on G4 forums (he supposedly works for Factor 5)
http://forums.g4tv.com/messageview....403677&FTVAR_MSGDBTABLE=&STARTPAGE=13

-----------------------------------------------------------
Alright, here are some specs from my sources about Revolution. Its still not the complete or "Satisfied" specs, but they are the numbers that would rival xbox 360 and PS3, like i have been saying all this while.....


Revolution Specs

CPU: IBM Custom PowerPC 3.5 GHz + 4 internal Power PC G5 cores running at 2.5 GHz each. Each core will have 128 KB or 256 Kb L1 cache. The whole CPU will share 512 KB - 1 MB of L2 cache. As you can see they are deciding wether to go up one notch. But never the less, its a very powerful CPU with 4 custom G5 cores.

There will also be two hardware threads per core, 8 hardware threads total.

12 billion dot product operations per second

Theoritical of 10 GHz total + 3 GHz CPU speed

Xbox 360 CPU:

Three symmetrical cores running at 3.2 GHz each = theoritical of 9.6 GHz + 3.2 GHz CPU speed
Two hardware threads per core; six hardware threads total
VMX-128 vector unit per core; three total
128 VMX-128 registers per hardware thread
1 MB L2 cache

Revolution GPU

ATI Custom based RN520 core. the "N" stands for nintendo, and is because the ArtX team is with them, thats why its an "N". There will be 2 GPU cores (just like the nVidia SLI motherboard with two Graphics Chips), this will use ATI's alternative, and will be the first in any console.

Both GPU's will have 256 MB's of GDDR4 memory, with an addional 16 MB of eDRAM total. eDRAM is an onboard flash memory, just like the 3 MB on the gamecubes flipper.

The cores will run at 600 MHz each, (rumours are that its possible 500 MHz each), but i doubt that.

24-to-48 way parallel floating-point dynamically scheduled shader pipelines for each core with unified shader architecture.

Polygon Performance: 500 million triangles per second theoritical, average ingame would be around <100 Million/sec

Shader Performance: 48 billion shader operations per second

Revolution memory

512 MB of 700 MHz Updated 1T-SRAM (its a totally redone design of something new, remember the nitrous i talked about)

thats it for now, i may addd some more later

-------------------------------------------------------------------

Also, as I said I’m going to add more info, the Revolution will support a PPU chip (Physical Processing Chip). This chip is very new to computer architecture, and it will mainly help in the physics area. There will be 32 MB’s of its own RAM, which will link to the CPU and GPU. To compare to the usage of physics used in current games, Half Life 2 only used 5 MB’s, Rebel strike used around 1.3 MB’s, and RE4 used 3 MB’s, but this is off main memory, which made performance issues.

There will also be a separate sound card that will support only DD 5.1 – DTS 7.1, rumours has it will have 16 MB’s, like the Cube DSP.

--------------------------------------------------------------------

This is some of my info on the PPU chip.
The PPU was built by the company Ageia, and is named the PhysX chip
PhysX is the world's first Physics Processing Unit (PPU), an entirely new category of processor that promises to revolutionize gaming in the physics area. By offloading software physics processing from the CPU and GPU, the PPU completes the triangle of gameplay, graphics and physics, balancing the load of these processing tasks and enabling pervasive interactive reality in games.
PhysX Processor Architecture has been designed to enable radical acceleration of:
· Rigid body dynamics
· Universal collision detection
· Finite element analysis
· Soft body dynamics
· Fluid dynamics
· Hair simulation
· Clothing simulation

There is an SDK program that runs with PhysX and is as follows:
NovodeX Physics SDK
This program called the NovodeX Physics SDK, is a stable, high-performance solution for game developers to enable physics-based gameplay and effects in PC and console titles. This has been used in Half Life 2 extensively and Far Cry. But CPU power handling this and other things isn’t good compared with a CPU doing what it was mean to do, like data flow, and a PPU with this program handling just physics (and can also add physics to the sounds produced), makes the games kick ass. A powerful API for the PhysX PPU, NovodeX enables game developers to inject both software-only and hardware-accelerated features into their games. The NovodeX Physics SDK is also the first and only asynchronous (multithreaded) physics API capable of unleashing the power of multiprocessor gaming systems, just like Xbox 360, PS3 and the Revolution.

----------------------------------------------------------------------------------------

I think he also mentioned something about liquid & or metal cooling but I don't see it right now. but anyway thats everything he posted about Rev so far. Just wanna see what you tech heads @ B3D forums think of this stuff and

 

[PC-Engine]

I find those specs hard to believe because it's just too many chips, more than needed to compete with SONY and MS. OTOH if Nintendo wanted to blow away the competition and launch Revolution at $300 then it's possible.

Looking at the cpu doesn't make any sense at all. Why does it say PowerPC 3.5GHz then it says 2.5GHz? :?

eDRAM is an onboard flash memory

 

[jvd]

yea i don't believe it , i believe these were talked about before .

I would believe 2 gpus with 256 megs on both of them . I would believe the 4 processers at 2.5 ghz each . I would believe most of that . Just not combined haha


Mabye we will see a dual core cpu at 2.5 ghz and a ppu ? with a good graphics card that should put it in league with the other systems

 

[Li Mu Bai]

The 2-3x more powerful than the GC quote has been proven erroneous btw, carry on gentlemen.

 

[randycat99]

Huh? I'm not comparing power output of GCN vs Revolution Sherlock. I'm comparing the PROCESSING power output of GCN vs Xbox hence Revolution vs Xbox 360.

Yes, I'm well aware you refer to processing power. Whether or not it really is equivalent in processing power is a whole other discussion, that which I'm sure nintendo fanbois and xbox fanbois have argued ad nauseum many times over. Perhaps the only point you need to take away from that issue is that, maybe they aren't equivalent at all? There certainly would be no shortage of persons willing to argue which was more "powerful" or were they "exactly equal" of some sort...

There is still the thermal power issue, and that is explicitly what my reply was intended to address. The fact remains, the GC had a "quiet windtunnel" largely because it was in a considerably different heat output realm than an XB, not because windtunnels have some super magical cooling property. Whether or not performance parity was actually reached GFLOP for GFLOP, that is up for long debate, to which no conclusive answer may be possible, anyway.

You need to quantify how much power it is dissipating. Are we talking about 40 W, 60 W, 200 W??? In one case, it is completely understandable how it does the cooling in a quiet little package. In the other, your bringing it up becomes utterly irrelevant wrt the requirements that Revolution will present.

Uh no I don't read above.

So where do you place the Rev in power consumption? 40 W? 60 W? 200 W? I think it is clear you intend to imply that Nintendo will come up with a Rev design that manages to match PS3 and XB2 in performance while somehow low-balling power consumption by a great amount. Question is...what do you think it will be, smarty? Are you afraid to go on record on this, despite your strong, seemingly authoritive stance in all peripheral areas to the issue?

LMAO, riiiight. Read above. Comparable PROCESSING POWER yet smaller and less heat output. Get it yet?

...comparable processing power, maybe. Smaller heat output, definitely, and it also puts the "quiet cool" windtunnel in a totally different perspective than you were willing to admit earlier. If it was a mere 5 W, would that still be impressive how quiet it is? Of course, not! It's quiet simply because it's job isn't particularly taxing given the device in question. If it must then handle 200/250/300 W, it may not be so quiet anymore, and that is the rude awakening you have set yourself up for.

Depends on what the definition of "considerable" is. Fox5 has already shot down your argument. A laptop is outputing near 200W running on AC in an extremelely low profile design without the need for a friggen 50dB snowblower.

You seem to have jumped too quickly w/o researching. That a laptop has a 200 W power brick, doesn't necessarily mean the laptop will shed off 200 W continuously. How do you know it may pull 200 W for 5 min, temperatures permitting, and then go into a low-power mode after 5 minutes to keep a lid on things? How do you know the processor/RAM/HD isn't resorting to extreme duty cycling to make brief draws of 200 W possible, but no where near on a continuous basis? Are you aware of the fan noise present under peak loads? A laptop has to go through ridiculous measures to keep from blowing its stack, regardless of any "200 W" rating. It's not impossible to find it drawing peaks of 200 W for a moment, but then classifying it as a small case that can dissipate a continuous 200 W may have no bearing in truth, whatsoever. For all you know, an effective continuous rating may not be any more than 60 W (pervasive, extensive duty cycling can really do wonders). Could you honestly be surprised if it was? No doubt it is an extremely specialized and tuned device to do what it does. Comparing "watts" to "watts" from a laptop to a console can be a stretch.

And that's building my point where comparable PROCESSING power doesn't equal comparable heat output.

Just how far do you think that can be stretched??? This time you won't even have the different ISA's to fudge. Everybody will be using PPC parts of some sort, and both XB2 and Rev will have some sort of ATI "God-card". Chances are processing power-to-thermal power indices will be pretty close, unless you are expecting a miracle.

You don't know much about aerodynamics do you? More fins means only slightly more drag in the case of GCN since the length of the fins are parallel to the airflow. You can even make the fins thinner (less drag) or taller (more surface area) or both instead of adding more in total numbers to lesson the increase in drag even further. Thinking stupid isn't a solution.

Thinner fins become less efficient in exposing T-core to surface area. You can only go so far... Naturally, you will be shoving more fins closer together, if you bothered to make them thinner- so up goes your drag. Taller fins are less efficient in exposing T-core to surface area, unless you make them thicker. Again, you can only go so far... So now your taller, thicker fins will be "closer" as a result of them being thicker. Now your air passage has become occupied by more metal and less air passage==>higher velocity==>higher drag==>which requires stronger fans to truly realize the increased heat transfer you were after in the first place. Adding more fins in total also decreases your air channel, and with all of your suggestions in place, you end up with an air channel that looks a lot more restrictive than free flowing... Now you say you would then increase the size of the air passage, but that is only assuming you had the space to grow your cooler in the first place. Maybe you didn't. Now what? It's not so easy, is it? Does that mean you picked the "stupid way" then? Maybe you are simply encountering "reality" just like anyone else?

Uh not quite. Taken to extremes if you have a big enough heatsink, you will not need a fan. Heck that is the basis for designs such has the Hush PCs. No fans means NO NOISE..Got it???

So then you end up with the sorts of heatsinks you find on 60/100/200 W car amps (and I have seen evidence that even those can be undersized to pull off the claimed power outputs for sustained periods. Do you honestly envision these kinds of heatsinks fitted within a GC case for your premise? That's going to be a bit hard since the heatsinks are easily bigger than the GC, itself. Suffice to say, you were more credible keeping your configurations fan-based.

You don't get it do you? I'm very aware of the difference between heat dissipation vs heat transport contrary to what you would like to believe LMAO.

I guess you have yet to display that from the points you have written so far.

You can't just talk about the endpoint to suit your own argument...

Precisely, it is the argument. If you have created 100 W of heat, your endpoint will be called upon to dissipate [gasp] 100 W of heat.

...and ignore the heat being dumped back into a case in a DUMB design like Xbox and standard PC designs.

True, they could be better, but it doesn't magically erase the plain logistical implications of dissipating 100/200/300 watts at whatever endpoint of your choosing.

Go back to page 3 or 4 where I first mention the liquid metal cooling solution and the flexibility it allows in terms of routing flexibility.

Yeah, sounds great, except "routing flexibility" was not exactly some terrible obstacle in the first place (you got smart engineers working on this, right?). Your bringing up "flexibility" all of a sudden reaks of "a solution looking for a problem", rather than the more important "a problem has found a solution".

It's kinda amusing watching you shift back and forth with little room you have left.

...revisionist commentary aside...

If the liquid metal system needs to be at a high temperture to flow/work then all you need to do is enclose the apparatus in an air chamber that's connected to the exhaust port fan so the air can come in at the inlet port close to the processors and out through the exhaust port close to the outside of the case similar to the GCN.

Wow, genius! ...also more space invested into the cooling componentry. Maybe you are better off just sticking with a case fan to evacuate the internal airspace? It's not such a taboo solution, unless you have some terrible hatred for case fans. It's simply a design element, pce, not a religious idealogy...

This will prevent the dumping of heat back into the case unlike the standard "HSF over processor" cooling design. With the standard design you need massive airflows just to get the heat out of the case because it has to fight all sorts stagnant hot air inside the case.

Wow! Sounds positively awful when you describe it. Maybe you only need massive airflow on the heatsink, and moderate airflow at the case to ensure adequate and regular air refresh? It may not be ideal, but it does work. Also, nothing stops you from placing the local heatsink in its own circulated plenum that intakes and exhausts to outer ambient, as well. No biggy. There's plenty of options available out there, beyond forcing a black or white, stupid way/not stupid way interpretation on things.

This is a bobo premise, altogether, because you ignore the nature of the hardware (PPC750 vs. Celeron at almost twice the clockrate, ArtX GPU vs. nVidia's finest, HD, etc...), . Now if Nintendo had managed to put all the Xbox hardware in a GC case, and made it work...that would indeed be an enviable feat. Beyond that, "timing" is utterly meritless.

Yes it's bobo because you don't have an answer.

No, it's bobo because you gloss over the details to the point of irrelevance.

The whole point of this discussion is getting a comparable amount of processing power into a smaller case. This has been done with GCN regardless whether it's using a PPC or not.

Right there, you may find PPC did have a good amount to do with it. Now that all 3 of the usual suspects will be onboard with PPC, you may not find the processing-to-heat prospect to be of adequate variance to your liking. Whether or not GCN even met your processing claim is still further conjecture. For all you know GCN may have even been behind PS2 in processing (but surely not feature set). I'm sure you are quite amenable to claiming XB was 2x PS2, so if GCN came out to less than half of XB, that certainly blows a rather big hole in your processing-to-heat assertion. Like I said, that's all rather deep conjecture at that point.

Sh*t if everbody had to use the same architectures and the same cooling systems we wouldn't be discussing this. Unfortunately for you that's not reality.

ridiculous...what that has to do with me is even further irrelevant.

You may not need a lot of airflow going throughout the case, but you will at the primary heatsink, regardless of its location. 100 W of dissipation remains as 100 W, 200 W is still 200 W, 300 W is still 300 W. Thermal heat doesn't magically disappear just because you relocate it through a conduit. To top it off, you still have one very hot processor inside your case, if you expect it to melt metal to make the liquid metal component work in the first place. So you end up needing a case fan to evacuate incidental heat inside the case, anyway, in addition to the one blowing on the remote heatsink.

And who said you don't need a heatsink??? I think it is you who are arguing with Mr. Scarecrow.

You read that and thought I was taking issue with a remote heatsink??? You've gone utterly coherent, at this point...

Heh what's comical is the fact you keep shifting back and forth ignoring all the facts presented before you and desparately hanging on to your strawman argument.

I mention "strawman argument", and now you can't wait to work it into your own material. Geezus, that's classic pce!

An unsmart design is where you flip a coin and accept whatever measely airflow passage is formed including the associated *drag* that comes with it and simple slapping a huge HSF over the processor. It's funny that your memory has suddenly pickup on other people's ideas instead of your flawed one to try and shift the balance. Boy talk about goal posts, your's seems to have quantum states.

No one suggested the "unsmart design" you've described, other than your twisted recollection of what people have said. Whether or not you have the option to enlarge airflow passage at will depends entirely if you had the room to spare in the first place. Maybe you do, maybe you don't. Maybe you should be more explicit about your solutions, so everyone can be on the same page ("pce's page", at least)?

You seem to have forgotten the whole point of this discussion, but since I'm such a nice guy exhibiting good sportsmanship,...

You already have forgone that status when you failed to compare best-of to best-of, "ref" your own arguments, provide unnecessary revisionist commentary on past events, and routinely make petty remarks against your opponent's "intelligence". Shall we also revisit your shameful past bout of inappropriate and false-cue smilies used in your posts to others? You are neither "nice" or sportsmanly.

I'll reiterate it main issue for you. Nintendo has shown that it can design a small console with comparable processing power to the Xbox.

...maybe, but it's ultimately conjecture.

That is the point. Artificially restricting the comparison down to slapping the innards of an Xbox into a GCN to make your point was never the issue.

It would certainly illustrate the merit of this magical windtunnel you are so hard-on about.

However if you wish, you can continue to argue that point with Mr. Scarecrow in your spare time.

...always the lack of original material from you... Do you really think you appear saavy, doing that?

I'm not even gonna comment on the hair dryer because the fan in a hair dryer has a different purpose than a fan in a HSF.

Blows heat off the "heat element(s)" onto the exit of the chamber, doesn't it? It does so extremely effectively, I might add. Maybe Nintendo has the intellectual goods to come up with a "stealth blowdryer", right?

The only relevent thing is that a powerful laptop produces a lot of heat, yet it has a very small volume to work with.

Just not the amount of heat you think, perhaps? I know you cannot resist to take it as the gospel, now, but your failure to consider the impacts of duty-cycle and peak vs. continuous loads in that scenario precludes you from realizing the ambiguity of the number given.

Nice attempt at trying to dodge the main issue. You would make a great politician. Air flow management and heat dissipation are interconnected...

Sure they are interconnected, but still not interchangeable. So that makes you wrong on 2 counts. Too bad for you...

Ah...no. The size of the case is irrelevant. You are moving air at the rate of CFM. By the wording you chose, it is the same. Now air velocities inside a small case vs. a large case may be different...

That was my point ie you'd be cycling the case volume many more times in a smaller case than a larger one with a certain CFM rating. Funny how you try to make it sound like that wasn't the point in an attempt to lesson the damage.

It's "lessen". "Lesson" is what I'm giving you. Ooooooh! Whether or not "recirculation" takes place or not will greatly rely, again, on the airflow management. It may be recirculating a lot or very little (or somewhere in between), but I'm guessing this is another case of you giving a comparison between "worst of" to "best of". You seem to absolutely dismiss the scenario where the affected part in the large case is always exposed to the path of incoming fresh, cool air, while warmed air is always closer on its path to the exhaust (not recirculating). The flowrate is exactly the same, but the velocity may be lower. A large case need not have utterly indeterminant flow within. Horses for courses...

Is this one of those "sky is blue" comments? Big deal! You also have the option to use a much bigger pump on the 100 gallon tank w/o turning it into a geyser, too. Horses for courses...

Of course it's a big deal Sherlock. Let me know when Nintendo puts in a 120mm fan with the associated gaping hole in Revolution.

It may need to if it has to shed 200 or 300 W continuous in a GC-sized case w/o shutdown or ceasing of gaming activity...but in your mind they would never have to cuz dey got da "magic tunnel", eh?

And of course that limit is unkown. Using Xbox or Xbox 360 as an idication of that limit is not wise.

Unknown to you, perhaps. Perhaps you feel it is unwise, because even you can see the trend does not help out your point. All we've seen from you is an insistence that a low-velocity windtunnel and liquid metal cooling in a near GC-sized case will suffice to cool a perceived embodiment of Rev that manages processing parity with PS3/XB2, implements equivalent or greater PPC/ATI processing elements, yet uses considerably less electrical power, ultimately. No one at all aims to prevent you from holding to that scenario as truth. I won't even resort to calling you names like "stupid" and such, insult you, accuse you of flipflopping, etc. for believing it. It makes no difference to me if you do believe or not. Now if you can do the same for me, perhaps we can truly share the same playing field when it comes to "sportsmanly conduct"?