Will case size limit Revolution's processing capability?

Status
Not open for further replies.
It doesn't spontaneously become 80/60/40 W for one system, just because you think one way is "more efficient" (...just a hint, but the "more efficient" forced-convection heatsink won't be the one with the slow, quiet flow, anyway...).

A water cooled system could cool just as well as one with a loud fan while being much quieter, simply because it can move the hot air to the outside of the case before using a fan to dissapate it.
 
...and then what happens once on the outside of the case? You still need to dissipate the heat. So it's either it's going to be a gargantuan heat sink with slow flow or a the heatsink that fits in proportion to the case with moderate flow, at the least. 100 W generated ==> 100 W to be dissipated.

We need to ditch this "worst of" compared to "best of" scenario. The "worst of" only leads to more misconception. Once you compare "best of" to "best of", you realize that there can be a quiet "internal-cooling" setup just as easily as there is a quiet "external-cooling" setup. Similarly, if the heat dissipation requirements are extreme enough, an "external-cooling" setup can end up just as noisy as an "internal-cooling" setup. There's no real difference here, except one uses an "extra part" in the process, which most assuredly introduces an efficiency loss to the overall system, rather than creating a net boost in system efficiency, as is often assumed. The amount of heat transfer is not a concrete indicator to "efficiency", anyway, though they are often confused as interchangeable.
 
randycat99 said:
...and then what happens once on the outside of the case? You still need to dissipate the heat. So it's either it's going to be a gargantuan heat sink with slow flow or a the heatsink that fits in proportion to the case with moderate flow, at the least. 100 W generated ==> 100 W to be dissipated.

The simple fact that you can move the heat to the outside of the case or the rear vents means you need less cooling .

If you were to release the heat from the heatsink on the chip the fans to cool it will use that hot air meaning you need more and more cooling to keep the chip cool .

If you can release the heat at teh rear of the case you need less cooling as none of the hot air will be recycled into the cooling
 
randycat99 said:
...and then what happens once on the outside of the case? You still need to dissipate the heat. So it's either it's going to be a gargantuan heat sink with slow flow or a the heatsink that fits in proportion to the case with moderate flow, at the least. 100 W generated ==> 100 W to be dissipated.

Yeah, but it matters where you dissipate it. Obviously it's better to try and funnel it outside the case, then just letting a cpu fan dissipate it right into the case and hoping it finds its way out. You have to dissipate 100W no matter what, but you don't have to let it sit right over where it's generated. Water cooling + a rather quiet fan should achieve results comparable to an ear deafening fan and no water cooling. Well, if the design is like the gamecube's case, then the ear deafening fan would be much better, but for a typical pc case I'd say comparable results.
 
Once again, you are comparing a "worst of" case to a "best of" case, which is not going to give you meaningful results. It should be fairly obvious that a comparison that means anything useful will imply good airflow management for either choice. Otherwise we could argue, that the remote heatsink scenario would be not much better off, if you are venting the warmed internal air over it. Naturally, the question becomes, "why would you do that, if you bothered to make the heatsink remote, in the first place?". Indeed, that would be another example of pce's "dumb designs". Hence, the logic should follow that either configuration will benefit from good airflow management, and either configuration can be foiled by bad airflow management.
 
randycat99 said:
Once again, you are comparing a "worst of" case to a "best of" case, which is not going to give you meaningful results. It should be fairly obvious that a comparison that means anything useful will imply good airflow management for either choice. Otherwise we could argue, that the remote heatsink scenario would be not much better off, if you are venting the warmed internal air over it. Naturally, the question becomes, "why would you do that, if you bothered to make the heatsink remote, in the first place?". Indeed, that would be another example of pce's "dumb designs". Hence, the logic should follow that either configuration will benefit from good airflow management, and either configuration can be foiled by bad airflow management.

Well in that case, there is probably still a slight advantage to using water cooling.(though I've seen results with proper ducting in PCs that gives results equal or close to water)
 
randycat99 said:
Once again, you are comparing a "worst of" case to a "best of" case, which is not going to give you meaningful results. It should be fairly obvious that a comparison that means anything useful will imply good airflow management for either choice. Otherwise we could argue, that the remote heatsink scenario would be not much better off, if you are venting the warmed internal air over it. Naturally, the question becomes, "why would you do that, if you bothered to make the heatsink remote, in the first place?". Indeed, that would be another example of pce's "dumb designs". Hence, the logic should follow that either configuration will benefit from good airflow management, and either configuration can be foiled by bad airflow management.

You wouldn't as the only thing that would still get hot enough to cause a drastic change in temps would be the ram .

Other than that the two main heat generators (cpu , gpu ) would have the heat piped to the exit vents and the cooling air would go through the front of the case over the other components right into the radiator on the rear inside of the case and then sucked through it buy the fan on the other side of the radiator pulling the air directly out of the case .


In a normal set up you have the two main heat generators pushing the air onto everything else . Which means not only does the cpu / gpu get very hot but that hot air then passes over all the other components making them run hotter which means you need even more case cooling than you wuold need with the liquid metal part
 
jvd said:
In a normal set up you have the two main heat generators pushing the air onto everything else .

We'll just label this as jvd's "dumb" design scenario. Fortunately, there is no cosmic rule that says an internal design must be laid-out as such. If optimal heat dissipation and noise are "of concern", good airflow management will be implemented...

As for the remote heatsink scenario, thermal isolation of the heat conduit components will not be perfect, anyway. So you'll still have some heat build-up in the case, anyway. This is not to say it is an insurmountable obstacle. Nevertheless, it is an additional effect that may or may not be ignored, just as you premise that good airflow management may or may not be ignored in an internal-cooling design.
 
We'll just label this as jvd's "dumb" design scenario. Fortunately, there is no cosmic rule that says an internal design must be laid-out as such. If optimal heat dissipation and noise are "of concern", good airflow management will be implemented...

Lay out is dependent on the design of the board and the design is dependent on the requirements of the power routing and trace routing .

Sorry to say but in a small case there is only so many places you can put the cpu and gpu and the disregarded heat from each of those will end up being recycled by the other and blown over other parts . There is no way around it . Not only that but you would need twice as many fans . You would need 1 on the gpu , 1 on the cpu , 1 intake and one out take .

In my set up you would need one intake and one out take .


But regardless there is no point in continuing this as you just will keep beliving what you want .

So i'm done in this thread . Have a good night
 
You took the word "layout" too literally. The airflow management will be a part of the "layout" (as much as the physical positioning of devices on a board), as well, where you can employ various baffles and isolators to keep hot flows and cold flows "discrete" and eliminate stagnating air bodies. I imagine your next argument will be that doing this will then be too difficult and too expensive... :rolleyes: (ultimately, no more of a pain than implementing a fancy "liquid-cooling" or quiet windtunnel solution)
 
I guess what Randy is trying to say (and correct me if I am wrong) is that it's a false comparison to assume that the Xbox360 and/or PS3 are going to have poor heat dissipation techniques whereas the Revolution is going to have a smart one.

In other words, what is preventing either MS or Sony from using liquid coolers as well (and perhaps they've always intended too)? And then the gap that the Revolution has to close is that much more because we can no longer assume that a "smart" cooling design for the Revolution is going to help it catch up.
 
it depends on how you start off with the design. IF you start off wanting power with cooling as an after thought then you wont get the best cooling. But if you want a small case with a small design form and power cooling will be the first thing you look at and then you design your chips to what you believe is the maximum amount of heat you can cool in the size of a case you want .


From whats shown of the ms case they have a heat pipe that simply is a normal heatsink. IT seems like they are going to use the size of the case and alot of air movement to keep it cool (At least from waht they showed in the coloney video of the insides )

I do not know what the ps3 is going to look like inside .
 
The mere fact that you need fan(s) that will need finite electrical input to drive the desired flowrate, would seem to disagree that "drag" is negligible. Otherwise, you could power the fans to speed once, and then cut the power, as the "perpetual motion flow" continues forever and ever...which we most certainly know will not be the case, "smart" design or not.

Grasping for straws...

"Drag" or pressure loss is an inherent effect from very low air velocities to high velocities, as well as simply being exponentially related. Fans aren't exactly "positive displacement" kind of devices, either. So they, indeed, do need to work to overcome even mild pressure losses in the overal flowpath, "smart" design or not. It's not like these are superchargers that could blow the intake manifold off an engine block.

If drag is say less than 5% for the total system then it won't matter relative to the other 95%.

Indeed, you are implying breaking the 1st law, when you suggest that the cooling demands will be fundamentally less than the heat input, just because the flow management is "smarter". Inherently, 100 W of heat created will need to be 100 W of heat dissipated. Dissipating 100 W would most certainly be more difficult in a poor flow management style setup, but you shouldn't be comparing "worst of" to "best of" in the first place. Once you've normalized the "best of" for 2 setups, the only thing that is left is 100 W of dissipation for either system. It doesn't spontaneously become 80/60/40 W for one system, just because you think one way is "more efficient" (...just a hint, but the "more efficient" forced-convection heatsink won't be the one with the slow, quiet flow, anyway...).

Uh no I'm not implying that, but thanks for telling me what I'm implying. Go back and reread jvd's post about HSF vs water or liquid metal cooling if you have a memory problem.

So there is the "smartly" designed system, the "dumb" system, and this just in...the "fantasy/mythical" designed system.

It's a hypothetical system just like the other systems you brought forth. Difference is it's not using some Xbox as the ultimate example of cooling efficiency. Maybe you should look at the size of Xbox's motherboard if you don't have a clue. Best of best.....RIIIIIIGGGHHHHTTT!!!! I guess we should artificially restrict the computing architectures to the same design too to futily prove our point. Hey why not just pretend the only option is to transplant the Xbox's innards into a GCN? :LOL:

I'm sorry but last time I checked this is the realworld where everything is NOT created equal. ;)
 
In 2005, you'd think we'd be living in perfect conditions. Vacuum and -273C. Where everything created is equal and everyone is equal. Instead we're stuck with this medieval atmosphere which makes it impossible to advance our technology. Funny enough, the closest to that is Russia!

Russia - the most advanced place on Earth.
 
london-boy said:
In 2005, you'd think we'd be living in perfect conditions. Vacuum and -273C. Where everything created is equal and everyone is equal. Instead we're stuck with this medieval atmosphere which makes it impossible to advance our technology. Funny enough, the closest to that is Russia!

Russia - the most advanced place on Earth.

I'm just glad the Yugo wasn't THE benchmark for automotive design since everyone would have a pretty hard time beating its uber smart design. :LOL:

Anyway I think I've found the answer to what can be achieved with a Revolution CPU. Are you guys ready for this?

A dual core PPC with 4 Clearspeed CSX600 coprocessors onboard.

As you may or may not know a single CSX600 math coprocessor only consumes 5W at 250MHz while having 50 GFLOPS of single OR double precison performance peak. Four of these coprocessors will net you 200 GFLOPS of single OR double precision performance peak and consuming only 20W and only running at 250MHz. This would put both CELL and XeCPU to shame. ;) :LOL:

Is this what Nintendo means by lower power? :devilish:

I'm waiting for Randycat99 to come back and say, "No it can't be, impossible, not fair! Nintendo can only use the same technology as the competition!!". :LOL: :rolleyes:
 
Communication between those processors?

Are they as general purpose as SPE's while still having their math prowess?

Local mem pools for the processors?

If no mem pools or similar level of general purpose programmability wouldn't that leave the Dual PPC core still pretty tasked? Compared to Cell and the X360-CPU in being tasked how good is this?

Dual core dual issue, quad issue what? VMX? Clock?

Can these co-processors act independently and asynchronously? What would that take in order for that to be true?

Still seems a complex solution no? More like Cell but also like X360's CPU no?

Cost? Still jive with Nintendo wanting a cheap console with this configuration being so complex or should I say comparably robust?

If so is there reason to understand why MS would not have done the same?

I'm am very skeptical of Nintendo being able to get comparably robust and powerful parts in the space of 3DVD cases AND still being cheap and especially cheaper than the competition in doing so.
 
PC-Engine said:
Difference is it's not using some Xbox as the ultimate example of cooling efficiency.

Do you have an XBox? I do and it doesn't seem to be any louder than my GCN. It hasn't ever failed under any sort of load I've thrown at it so whatever cooling their using works.

Could the efficiency of the cooling be improved upon? Probably like everything else.

PC-Engine said:
Maybe you should look at the size of Xbox's motherboard if you don't have a clue.

But this time around it seems quite possible that the primary heat sources will be nearly the same. In other words, we all recognize that the Xbox used 'off the shelf parts' to cobble together a console and Nintendo had a nice little package created for them courtesy of ArtX.

This time around though, both MS and Nintendo went to the same companies to produce their consoles. Will they be the same? I dunno and neither does anyone else here. Perhaps there are two separate design teams or perhaps not.
 
Status
Not open for further replies.
Back
Top