Got any more neat specs for that Cray 2? Like how many CPUs or clockrate? 2 GB of memory in 1985?! That must have been a HUUUUUGE array!
All I can find is:
1 foreground processor, 4 background processors
244 MHz clock speed
2GBytes RAM
64 or 128-way memory interleave
128KB Cache
8 16-word prefetch buffers
Foreground processor controls I/O with up to four 4-Gb/S channels
Check out this cool link. This guy took apart some cray 2 modules and shows how the circuit boards were stacked one atop the other to form a very dense three-dimensional circuits.
As these chips were all ECL (very VERY hot running chips), no wonder they needed liquid cooling--there was no way they could have put heat sinks on all of those devices. And the cooling systems described in this thread wouldn't work, as every one of the thousands of chips would need attention--this thing *had* to be immersed!
Air can only carry away so much heat; the more heat you want to dissipate, the more air flow and surface area you need for a heatsink. Diminishing returns starts to set in fairly rapidly once you get beyond the Alpha heatsink stage. No matter how large the heatsink, its contact point is still the CPU's slug. As heat moves beyond this contact area, it encounters something called "spread resistance". The reason you see copper bases in some heatsinks is that copper is more efficient in spreading heat over its surface than aluminum, but only up to a point.
So if the task at hand is to get as much cooling concentrated in a given area, all things being equal the more efficient "dissipating medium" (i.e. air or water) will cool better than its inferior. In the case of air vs. water, there is no contest - water is way better.
So what you have documented is that liquid cooling is applicable where heat spread is inadequate to keep the CPU below max operating temperatures. Being that a given CPU seems to do just fine with a conventional air-cooled heatsink (in this case, a G5), that is pretty much fair indication that heat spread is not yet an issue. I think what you have finally caught onto is that you don't just go and add water-cooling just for the hell of it. You add it where simple air cooling cannot be done practically. Contrary to your little excerpt, you don't even need to add it just because it is "better" or because it is "teh cool" setup. You use it because that is what is required to make your system work.
From Intel's thermal design guide for the Pentium 4:
Quote:
In addition to passive heatsinks, fan heatsinks and system fans, other solutions exist for cooling integrated circuit devices. For example, ducted blowers, heat pipes and liquid cooling are all capable of dissipating additional heat. Due to their varying attributes, each of these solutions may be appropriate for a particular system implementation.
In 60 pages of thermal management guidelines, this is the only mention of liquid cooling. You'll get no argument from any engineer--myself included--that pumping cooled fluid through a heat sink on a processor will draw heat away faster than moving air. However, to make the blanket statement that this is "better", as in that overclockers' article, is miss the point. As has been mentioned many times before, you are still left with the problem of getting the heat out of the fluid once it has drawn it away from the hot device. Also, if simpler, cheaper methods work, there is little incentive to use anything else.
Engineers who keep their jobs subscribe to the KISS method of design: "Keep it Simple, Stupid". Given that, until it is an engineering necessity, you simply won't see liquid cooling in widespread use in mainstream PCs. Someday, more exotic cooling methods likely will be necessary (note the Cray 2 above), but they currently are not. If liquid cooling were so obviously the superior solution, we would see far more examples of it in the marketplace--not just in ads in overclockers' magazines. Note the major difference between "superior solution" and "superior method"--liquid cooling is clearly a superior method, but not currently the superior solution.
Anyway, iSegway, do you honestly feel that the engineers who design PCs (Apple or otherwise) are so misguided as to not see the truth about liquid cooling? My current design has very hot components that run at 2.488 GHz; Careful chassis design and thermal analysis mean that we still only need passive heat sinks. Good engineering is about doing more with less--throwing complexity and money at a problem is a cop out when simpler, cheaper and perfectly valid solutions exist.
So what you have documented is that liquid cooling is applicable where heat spread is inadequate to keep the CPU below max operating temperatures.
What you have documented is that you will never admit you were wrong.
When I read your most recent post of obsurdity and proof of your mental problems I was so shocked I literally, though inadvertantly did a "Spit-take"(as they call it in show business). A spit-take is when you blow a liquid out of your mouth or nose, or both, because you are so flabbergasted.
The reason I bring this up is because if I had the G5 sitting on my desk at this moment I would have blown Coca-cola all over my 5,000 + dollar computers internals... and the 9 fans would have seen to it that this soda was properly applied to the entire innards of said beast.
It's funny... I wonder how much we as customers will pay for the G5's "case" which isn't really a case at all?
"post of obsurdity and proof of your mental problems"?
Seriously, get a grip.
Did you even read about the "heat spread" part of your excerpt, or are you narrow-focused on only the part that says "it's better"?
With regard to spittakes, I have to say that nothing nothing else would suit me more when watching that sexy T'Pau firing off "logical", compelling reasons to have sex with her... (Well, that's what's showing tonight on Enterprise. I almost changed channels to watch the Daily Show, instead, but now I'm sure glad I didn't.)
Anyway, iSegway, do you honestly feel that the engineers who design PCs (Apple or otherwise) are so misguided as to not see the truth about liquid cooling? My current design has very hot components that run at 2.488 GHz; Careful chassis design and thermal analysis mean that we still only need passive heat sinks. Good engineering is about doing more with less--throwing complexity and money at a problem is a cop out when simpler, cheaper and perfectly valid solutions exist.
Lets get one thing straight... I love apples designers, and engineers, innovativeness. I absolutely love it. I love most of their design choices...but this is one that is really disappointing.
This wind tunnel solution won't last. It doesn't have enough life in it and even as is it has some horrible problems. As I have pointed out over and over.
So why even bother with this bizarre system? At the very least they should have gone to a liquid system. But there are even some other systems that have a lot of potential, too.
Sure... we can just deal with all all the problems of the G5 case design... but why should we when we have other alternatives?
We can walk to work... or wash our clothes on a rock in a river... OR we can spend some extra money and buy a car and a washing machine to make our lives easier and better. Hell, we don't really need a computer... there are many alternatives to do the things we do on computers... yet we pay extra money for the convenience and luxury a computer provides... so why are we unwilling to pay for a better, smaller, quieter and cleaner method of cooling our computer?
so why are we unwilling to pay for a better, smaller, quieter and cleaner method of cooling our computer?
Simple- because the product that has been offered is already good enough, manageable in size, quiet enough, and clean enough. Maybe not to you, but other people seem to be buying in droves. Apple can't seem to keep them stocked.
What you are arguing is why doesn't Dodge make the Viper "better" by adding nitrous? Well, because the Viper engine is quite "adequate" just the way it comes from the factory.
Simple- because the product that has been offered is already good enough, manageable in size, quiet enough, and clean enough. Maybe not to you, but other people seem to be buying in droves. Apple can't seem to keep them stocked.
This wind tunnel solution won't last. It doesn't have enough life in it and even as is it has some horrible problems. As I have pointed out over and over.
I haven't seen any horrible problems mentioned. It's cheap to build, scales beautifully, runs quietly, and allows Apple to make a tower that is cleanly laid out and incredibly easy to upgrade. Dust? That's been a problem for as long as computers have existed. (Some enterprising spider even built a cobweb in my 8600!) It's what those cans of compressed air are for. And anyway, if you put the radiator fans in your liquid-cooled case inside the case, then you still have the dust problem - or, if they're outside the case, you have the noise problem.
Quote:
so why are we unwilling to pay for a better, smaller, quieter and cleaner method of cooling our computer?
We are, and we have, when it's actually better. I have a passively cooled computer sitting on my desk, right now, and I paid a premium for it because it's passively cooled, small, and quiet. But the Cube is able to be that way because it's barely upgradeable internally. The question is, is it worth the extra cost, weight, clutter, and difficulty upgrading to make a tower with liquid cooling? Apple's engineers felt not, and considering that they have, if anything, led the industry in using various passive cooling technologies - including liquid - I don't think they did this out of ignorance, or out of some loyalty to the fan. Of course there are flaws, but any solution has flaws, not least the liquid solution. Engineering is about tradeoffs.
In fact, I'm recalling a post on Ars (which I'm unwilling to go and find right now) stating that there's a channel in the G5's CPU heatsinks to aid in heat dispersion. So there might be a bit of liquid cooling in there now - where it's needed, doing what it's best at doing.
I am not an engineer so ignore this if I am wrong. But it seems to me that liquid cooling will bring a certain degree of design and engineering freedom to the designers.
Sure, liquid cooling only moves the heat away from the CPU, but by doing that wouldn't you have more flexibility in case design? I feel that Apple should be (and probably already is) looking into liquid cooling as a future possibility.
Someone compared the situation to cars. The way I see it, car engineers are now able to pack a lot more components in a smaller area because of liquid cooling. It may not cool better than air cooling but it makes the engineering simpler by moving the cooling components out of the immediate vicinity of the engine. Otherwise, all modern car engines will have humongous fins all over them and the car hoods will bulge out to accomodate those fins. Liquid cooling also allows you to safely slap stuff on the engine like a turbo.
In the example of the G5, the dual CPU can be placed horizontal to each other rather than vertical, thus reducing the height (not to mention the effect of the smaller heatsinks). That will allow it to fit under my desk and probably a lot of other people's desks too.
NEC's water cooling system put the radiator (is that the right word?) in the back of the case. The water-cooling does not allow NEC to extract more performance from the CPU but it supposedly (according to them) allow the computer to run quieter. Don't know why. Ask them.
EDIT: BTW, Apple claims that the new G5 is about 35dB on average use. NEC claims that their P4 3Gig is about 33dB on average use.
As for the complaint that water cooling will prevent people from tampering with the innards of the computer. That is exactly what a lot of computer companies want. They don't want you mess with insides as much as possible. It becomes a support nightmare when users mess with the insides of their machine. And if the car example is indication, computers will eventually get too complicated for anybody to mess around with. Remember, people complained about fuel injection, dual overhead cams, VTEC, electronic monitoring systems etc. They all said the same thing: I oppose it because it will limit our abilities to soup up the engines! But it all happened anyway. That is the nature of progress.
Sure... we can just deal with all all the problems of the G5 case design... but why should we when we have other alternatives?
What are these problems and deficiencies in the G5 case design that hinder performance???
It's actually quieter than the previous G4, has faster hard drives and the cooling zones might be the first step towards integrating liquid cooling into sections...who knows? all i know is that it would be stupid to hold up releasing the G5 'cuz the liquid-cooled-gets-respect-at-LANs case isn't ready yet...
And, really...why do you assume that Apple isn't even looking at liquid cooling?? Do you really think they're not? Do you really think that if it were ready and practical,cheap enough and ready at the same time as teh G5, they wouldn't release it??
I am not an engineer so ignore this if I am wrong. But it seems to me that liquid cooling will bring a certain degree of design and engineering freedom to the designers.
Absolutely, and you can see that in Apple's portables, in the iMac, and in the Cube.
Quote:
As for the complaint that water cooling will prevent people from tampering with the innards of the computer. That is exactly what a lot of computer companies want.
But the G5 is designed to invite "tampering" in the form of RAM, HDD, PCI and AGP upgrades, even to the extent of making it as easy as it has ever been to add any of those items to a tower (well, OK, you have to pull a fan out to get to the RAM...). Furthermore, they have a lot of headroom with the power supply and the CPUs, because the fans cooling those can handle the hottest current configuration easily by running at idle.
So, as I've said repeatedly, yes liquid and other cooling techs are better for "sealed" boxes, and they allow designers to come up with amazing, compact and unconventional designs. Apple's own lineup has demonstrated that beautifully. But the PowerMac G5 is a tower. It has to easily accept GeForce FX cards, 12" PCI cards, 10,000 RPM SATA drives, 8GB (or 16GB, since the PM will be able to accept 2GB modules when those become available) of the fastest RAM on the market, and one or two workstation-class CPUs, all while running quietly. It's not a stretch to see how the cooling choices that Apple made specifically for the PMG5 are appropriate to that model. Nor is it hard to see how liquid cooling would compromise what was clearly one of the main design goals of the G5, which is to be easily upgradeable.
That was a fantastic post, BeigeUser. I agree with everything you said... I wish I could have said it as eloquently.
I actually believe the whole expandability and modification aspect of personal computers needs to be addressed in some different way. We either need to treat the bulk of our computers as if they were some type of appliance... like your homes water heater OR like your homes stereo system.
One scenario puts the bulk of the computer which is just absolutley functional and has no regard for aesthetics and is built for easy modification in your basment or a closet or attic.
Then you have the modular stackable stereo angle... where each piece could have its own small case that you could buy for it. Almost like varying sizes of christmas boxes.
Lets say you get a new video card... buy a "christmas box" like apple case in the thickness that matches the thickness of the new video card. Simply stack it on top of your existing PC.
This is really difficult to describe. lol I need to draw some pictures or something.
I actually believe the whole expandability and modification aspect of personal computers needs to be adressed in some different way. We either need to treat the bulk of our computers as if they were some type of appliance... like yout homes water heater OR like your homes stereo system.
Or like, I don't know... the Apple //, or the original Mac, or the iMac, or any PowerBook, or...
Quote:
One scenario puts the bulk of the computer which is just absolutley funcional and has no regard for aesthetics and is built for easy modification.
Or, which turns function into a rather stunning form, which is one of the core precepts of industrial design. After all, you could just as easily build "a computer which is absolutely functional and has no regard for aesthetics and is built to be modular and stackable," too. In fact, I'd offer that a design that isn't constrained by the need to be modular or stackable gives the designer more liberty as far as making it attractive.
Quote:
Lets say you get a new video card... buy a "christmas box" like apple case in the thickness that matches the thickness of the new video card. Simply stack it on top of your existing PC.
This is really difficult to describe. lol I need to draw some pictures or something.
No, you just need the cable-friendly version of HyperTransport, which is a reality now.
One scenario puts the bulk of the computer which is just absolutley functional and has no regard for aesthetics and is built for easy modification in your basment or a closet or attic.
Well it won't matter how large or noisy it is down there then, will it?
Quote:
Then you have the modular stackable stereo angle... where each piece could have its own small case that you could buy for it. Almost like varying sizes of christmas boxes.
Lets say you get a new video card... buy a "christmas box" like apple case in the thickness that matches the thickness of the new video card. Simply stack it on top of your existing PC.
That one is just too classic to touch! Aww, what the hell- I suppose each of these boxes will have their own liquid cooling, too?
Or like, I don't know... the Apple //, or the original Mac, or the iMac, or any PowerBook, or...
You don't store iMacs in your basement or in your closet... and you can't stack components modularly. I don't understand what you are saying here.
Quote:
In fact, I'd offer that a design that isn't constrained by the need to be modular or stackable gives the designer more liberty as far as making it attractive.
But in the G5's case you don't have either... this is my point.
Quote:
No, you just need the cable-friendly version of HyperTransport, which is a reality now.
Can you make a modular casing system like a stereo system? Has someone done this recently?
You don't store IMacs in your basment or in your closet... and you can't stack components modularly. I don't understand what you are saying here.
I was riffing off the "computer like an appliance" idea, which has been central to Apple's thinking for over two decades now. And which I happen to think is a great idea.
Quote:
Can you make a modular casing system like a stereo system? Has someone done this recently?
HT over cable is new enough that nobody's deployed it yet, but it's the same technology that wires up the various parts of the G5's motherboard, and it happens to be able to be sent over a cable (up to 3 feet long, I think) at full bandwidth. Meaning that you can have several discrete boxes that act, in a very real sense, as if they were all together on the same big motherboard. As long as the various components were within 3 feet of each other it wouldn't matter what they looked like or what they were shaped like, but stackable components are certainly possible (just be aware of heat issues). The closest current answer to what you want is the good old 19" rack, which has been used to "stack" components together for decades now.
As to having the things in a closet - well, that brings in networking, and wireless networking, both of which are significant bottlenecks. If you want to hook up a display then the whole stack will have to be somewhere in the general vicinity of a desk. Closets are the domain of headless servers, not personal computers.
Comments
Originally posted by Randycat99
Got any more neat specs for that Cray 2? Like how many CPUs or clockrate? 2 GB of memory in 1985?! That must have been a HUUUUUGE array!
All I can find is:
- 1 foreground processor, 4 background processors
- 244 MHz clock speed
- 2GBytes RAM
- 64 or 128-way memory interleave
- 128KB Cache
- 8 16-word prefetch buffers
- Foreground processor controls I/O with up to four 4-Gb/S channels
Check out this cool link. This guy took apart some cray 2 modules and shows how the circuit boards were stacked one atop the other to form a very dense three-dimensional circuits.As these chips were all ECL (very VERY hot running chips), no wonder they needed liquid cooling--there was no way they could have put heat sinks on all of those devices. And the cooling systems described in this thread wouldn't work, as every one of the thousands of chips would need attention--this thing *had* to be immersed!
The Cray 1 sounds really amazing, though. NAND gate chips? Oh, man... I'd love to see that circuit board.
WHY IS WATER COOLING BETTER?
Air can only carry away so much heat; the more heat you want to dissipate, the more air flow and surface area you need for a heatsink. Diminishing returns starts to set in fairly rapidly once you get beyond the Alpha heatsink stage. No matter how large the heatsink, its contact point is still the CPU's slug. As heat moves beyond this contact area, it encounters something called "spread resistance". The reason you see copper bases in some heatsinks is that copper is more efficient in spreading heat over its surface than aluminum, but only up to a point.
So if the task at hand is to get as much cooling concentrated in a given area, all things being equal the more efficient "dissipating medium" (i.e. air or water) will cool better than its inferior. In the case of air vs. water, there is no contest - water is way better.
From Intel's thermal design guide for the Pentium 4:
In addition to passive heatsinks, fan heatsinks and system fans, other solutions exist for cooling integrated circuit devices. For example, ducted blowers, heat pipes and liquid cooling are all capable of dissipating additional heat. Due to their varying attributes, each of these solutions may be appropriate for a particular system implementation.
In 60 pages of thermal management guidelines, this is the only mention of liquid cooling. You'll get no argument from any engineer--myself included--that pumping cooled fluid through a heat sink on a processor will draw heat away faster than moving air. However, to make the blanket statement that this is "better", as in that overclockers' article, is miss the point. As has been mentioned many times before, you are still left with the problem of getting the heat out of the fluid once it has drawn it away from the hot device. Also, if simpler, cheaper methods work, there is little incentive to use anything else.
Engineers who keep their jobs subscribe to the KISS method of design: "Keep it Simple, Stupid". Given that, until it is an engineering necessity, you simply won't see liquid cooling in widespread use in mainstream PCs. Someday, more exotic cooling methods likely will be necessary (note the Cray 2 above), but they currently are not. If liquid cooling were so obviously the superior solution, we would see far more examples of it in the marketplace--not just in ads in overclockers' magazines. Note the major difference between "superior solution" and "superior method"--liquid cooling is clearly a superior method, but not currently the superior solution.
Anyway, iSegway, do you honestly feel that the engineers who design PCs (Apple or otherwise) are so misguided as to not see the truth about liquid cooling? My current design has very hot components that run at 2.488 GHz; Careful chassis design and thermal analysis mean that we still only need passive heat sinks. Good engineering is about doing more with less--throwing complexity and money at a problem is a cop out when simpler, cheaper and perfectly valid solutions exist.
So what you have documented is that liquid cooling is applicable where heat spread is inadequate to keep the CPU below max operating temperatures.
What you have documented is that you will never admit you were wrong.
When I read your most recent post of obsurdity and proof of your mental problems I was so shocked I literally, though inadvertantly did a "Spit-take"(as they call it in show business). A spit-take is when you blow a liquid out of your mouth or nose, or both, because you are so flabbergasted.
The reason I bring this up is because if I had the G5 sitting on my desk at this moment I would have blown Coca-cola all over my 5,000 + dollar computers internals... and the 9 fans would have seen to it that this soda was properly applied to the entire innards of said beast.
It's funny... I wonder how much we as customers will pay for the G5's "case" which isn't really a case at all?
"post of obsurdity and proof of your mental problems"?
Seriously, get a grip.
Did you even read about the "heat spread" part of your excerpt, or are you narrow-focused on only the part that says "it's better"?
With regard to spittakes, I have to say that nothing nothing else would suit me more when watching that sexy T'Pau firing off "logical", compelling reasons to have sex with her... (Well, that's what's showing tonight on Enterprise. I almost changed channels to watch the Daily Show, instead, but now I'm sure glad I didn't.)
Anyway, iSegway, do you honestly feel that the engineers who design PCs (Apple or otherwise) are so misguided as to not see the truth about liquid cooling? My current design has very hot components that run at 2.488 GHz; Careful chassis design and thermal analysis mean that we still only need passive heat sinks. Good engineering is about doing more with less--throwing complexity and money at a problem is a cop out when simpler, cheaper and perfectly valid solutions exist.
Lets get one thing straight... I love apples designers, and engineers, innovativeness. I absolutely love it. I love most of their design choices...but this is one that is really disappointing.
This wind tunnel solution won't last. It doesn't have enough life in it and even as is it has some horrible problems. As I have pointed out over and over.
So why even bother with this bizarre system? At the very least they should have gone to a liquid system. But there are even some other systems that have a lot of potential, too.
Sure... we can just deal with all all the problems of the G5 case design... but why should we when we have other alternatives?
We can walk to work... or wash our clothes on a rock in a river... OR we can spend some extra money and buy a car and a washing machine to make our lives easier and better. Hell, we don't really need a computer... there are many alternatives to do the things we do on computers... yet we pay extra money for the convenience and luxury a computer provides... so why are we unwilling to pay for a better, smaller, quieter and cleaner method of cooling our computer?
Originally posted by iSegway
so why are we unwilling to pay for a better, smaller, quieter and cleaner method of cooling our computer?
Simple- because the product that has been offered is already good enough, manageable in size, quiet enough, and clean enough. Maybe not to you, but other people seem to be buying in droves. Apple can't seem to keep them stocked.
What you are arguing is why doesn't Dodge make the Viper "better" by adding nitrous? Well, because the Viper engine is quite "adequate" just the way it comes from the factory.
Sometimes you make it too easy...
Simple- because the product that has been offered is already good enough, manageable in size, quiet enough, and clean enough. Maybe not to you, but other people seem to be buying in droves. Apple can't seem to keep them stocked.
You sound like Bill Gates. LMAO
Originally posted by iSegway
This wind tunnel solution won't last. It doesn't have enough life in it and even as is it has some horrible problems. As I have pointed out over and over.
I haven't seen any horrible problems mentioned. It's cheap to build, scales beautifully, runs quietly, and allows Apple to make a tower that is cleanly laid out and incredibly easy to upgrade. Dust? That's been a problem for as long as computers have existed. (Some enterprising spider even built a cobweb in my 8600!) It's what those cans of compressed air are for. And anyway, if you put the radiator fans in your liquid-cooled case inside the case, then you still have the dust problem - or, if they're outside the case, you have the noise problem.
so why are we unwilling to pay for a better, smaller, quieter and cleaner method of cooling our computer?
We are, and we have, when it's actually better. I have a passively cooled computer sitting on my desk, right now, and I paid a premium for it because it's passively cooled, small, and quiet. But the Cube is able to be that way because it's barely upgradeable internally. The question is, is it worth the extra cost, weight, clutter, and difficulty upgrading to make a tower with liquid cooling? Apple's engineers felt not, and considering that they have, if anything, led the industry in using various passive cooling technologies - including liquid - I don't think they did this out of ignorance, or out of some loyalty to the fan. Of course there are flaws, but any solution has flaws, not least the liquid solution. Engineering is about tradeoffs.
In fact, I'm recalling a post on Ars (which I'm unwilling to go and find right now) stating that there's a channel in the G5's CPU heatsinks to aid in heat dispersion. So there might be a bit of liquid cooling in there now - where it's needed, doing what it's best at doing.
Sure, liquid cooling only moves the heat away from the CPU, but by doing that wouldn't you have more flexibility in case design? I feel that Apple should be (and probably already is) looking into liquid cooling as a future possibility.
Someone compared the situation to cars. The way I see it, car engineers are now able to pack a lot more components in a smaller area because of liquid cooling. It may not cool better than air cooling but it makes the engineering simpler by moving the cooling components out of the immediate vicinity of the engine. Otherwise, all modern car engines will have humongous fins all over them and the car hoods will bulge out to accomodate those fins. Liquid cooling also allows you to safely slap stuff on the engine like a turbo.
In the example of the G5, the dual CPU can be placed horizontal to each other rather than vertical, thus reducing the height (not to mention the effect of the smaller heatsinks). That will allow it to fit under my desk and probably a lot of other people's desks too.
NEC's water cooling system put the radiator (is that the right word?) in the back of the case. The water-cooling does not allow NEC to extract more performance from the CPU but it supposedly (according to them) allow the computer to run quieter. Don't know why. Ask them.
EDIT: BTW, Apple claims that the new G5 is about 35dB on average use. NEC claims that their P4 3Gig is about 33dB on average use.
As for the complaint that water cooling will prevent people from tampering with the innards of the computer. That is exactly what a lot of computer companies want. They don't want you mess with insides as much as possible. It becomes a support nightmare when users mess with the insides of their machine. And if the car example is indication, computers will eventually get too complicated for anybody to mess around with. Remember, people complained about fuel injection, dual overhead cams, VTEC, electronic monitoring systems etc. They all said the same thing: I oppose it because it will limit our abilities to soup up the engines! But it all happened anyway. That is the nature of progress.
Originally posted by iSegway
Sure... we can just deal with all all the problems of the G5 case design... but why should we when we have other alternatives?
What are these problems and deficiencies in the G5 case design that hinder performance???
It's actually quieter than the previous G4, has faster hard drives and the cooling zones might be the first step towards integrating liquid cooling into sections...who knows? all i know is that it would be stupid to hold up releasing the G5 'cuz the liquid-cooled-gets-respect-at-LANs case isn't ready yet...
And, really...why do you assume that Apple isn't even looking at liquid cooling?? Do you really think they're not? Do you really think that if it were ready and practical,cheap enough and ready at the same time as teh G5, they wouldn't release it??
Originally posted by BeigeUser
I am not an engineer so ignore this if I am wrong. But it seems to me that liquid cooling will bring a certain degree of design and engineering freedom to the designers.
Absolutely, and you can see that in Apple's portables, in the iMac, and in the Cube.
As for the complaint that water cooling will prevent people from tampering with the innards of the computer. That is exactly what a lot of computer companies want.
But the G5 is designed to invite "tampering" in the form of RAM, HDD, PCI and AGP upgrades, even to the extent of making it as easy as it has ever been to add any of those items to a tower (well, OK, you have to pull a fan out to get to the RAM...). Furthermore, they have a lot of headroom with the power supply and the CPUs, because the fans cooling those can handle the hottest current configuration easily by running at idle.
So, as I've said repeatedly, yes liquid and other cooling techs are better for "sealed" boxes, and they allow designers to come up with amazing, compact and unconventional designs. Apple's own lineup has demonstrated that beautifully. But the PowerMac G5 is a tower. It has to easily accept GeForce FX cards, 12" PCI cards, 10,000 RPM SATA drives, 8GB (or 16GB, since the PM will be able to accept 2GB modules when those become available) of the fastest RAM on the market, and one or two workstation-class CPUs, all while running quietly. It's not a stretch to see how the cooling choices that Apple made specifically for the PMG5 are appropriate to that model. Nor is it hard to see how liquid cooling would compromise what was clearly one of the main design goals of the G5, which is to be easily upgradeable.
I actually believe the whole expandability and modification aspect of personal computers needs to be addressed in some different way. We either need to treat the bulk of our computers as if they were some type of appliance... like your homes water heater OR like your homes stereo system.
One scenario puts the bulk of the computer which is just absolutley functional and has no regard for aesthetics and is built for easy modification in your basment or a closet or attic.
Then you have the modular stackable stereo angle... where each piece could have its own small case that you could buy for it. Almost like varying sizes of christmas boxes.
Lets say you get a new video card... buy a "christmas box" like apple case in the thickness that matches the thickness of the new video card. Simply stack it on top of your existing PC.
This is really difficult to describe. lol I need to draw some pictures or something.
Originally posted by iSegway
I actually believe the whole expandability and modification aspect of personal computers needs to be adressed in some different way. We either need to treat the bulk of our computers as if they were some type of appliance... like yout homes water heater OR like your homes stereo system.
Or like, I don't know... the Apple //, or the original Mac, or the iMac, or any PowerBook, or...
One scenario puts the bulk of the computer which is just absolutley funcional and has no regard for aesthetics and is built for easy modification.
Or, which turns function into a rather stunning form, which is one of the core precepts of industrial design. After all, you could just as easily build "a computer which is absolutely functional and has no regard for aesthetics and is built to be modular and stackable," too. In fact, I'd offer that a design that isn't constrained by the need to be modular or stackable gives the designer more liberty as far as making it attractive.
Lets say you get a new video card... buy a "christmas box" like apple case in the thickness that matches the thickness of the new video card. Simply stack it on top of your existing PC.
This is really difficult to describe. lol I need to draw some pictures or something.
No, you just need the cable-friendly version of HyperTransport, which is a reality now.
Originally posted by iSegway
One scenario puts the bulk of the computer which is just absolutley functional and has no regard for aesthetics and is built for easy modification in your basment or a closet or attic.
Well it won't matter how large or noisy it is down there then, will it?
Quote:
Then you have the modular stackable stereo angle... where each piece could have its own small case that you could buy for it. Almost like varying sizes of christmas boxes.
Lets say you get a new video card... buy a "christmas box" like apple case in the thickness that matches the thickness of the new video card. Simply stack it on top of your existing PC.
That one is just too classic to touch! Aww, what the hell- I suppose each of these boxes will have their own liquid cooling, too?
Or like, I don't know... the Apple //, or the original Mac, or the iMac, or any PowerBook, or...
You don't store iMacs in your basement or in your closet... and you can't stack components modularly. I don't understand what you are saying here.
In fact, I'd offer that a design that isn't constrained by the need to be modular or stackable gives the designer more liberty as far as making it attractive.
But in the G5's case you don't have either... this is my point.
No, you just need the cable-friendly version of HyperTransport, which is a reality now.
Can you make a modular casing system like a stereo system? Has someone done this recently?
Well it won't matter how large or noisy it is down there then, will it?
Exactly! Wow... yer smart! That was my point dimwit, or should I call you Bill?
That one is just too classic to touch! Aww, what the hell- I suppose each of these boxes will have their own liquid cooling, too?
Sure. Why couldn't you have a modular cooling system? The flow of electricity is very similar to flowing water....
Originally posted by iSegway
You don't store IMacs in your basment or in your closet... and you can't stack components modularly. I don't understand what you are saying here.
I was riffing off the "computer like an appliance" idea, which has been central to Apple's thinking for over two decades now. And which I happen to think is a great idea.
Can you make a modular casing system like a stereo system? Has someone done this recently?
HT over cable is new enough that nobody's deployed it yet, but it's the same technology that wires up the various parts of the G5's motherboard, and it happens to be able to be sent over a cable (up to 3 feet long, I think) at full bandwidth. Meaning that you can have several discrete boxes that act, in a very real sense, as if they were all together on the same big motherboard. As long as the various components were within 3 feet of each other it wouldn't matter what they looked like or what they were shaped like, but stackable components are certainly possible (just be aware of heat issues). The closest current answer to what you want is the good old 19" rack, which has been used to "stack" components together for decades now.
As to having the things in a closet - well, that brings in networking, and wireless networking, both of which are significant bottlenecks. If you want to hook up a display then the whole stack will have to be somewhere in the general vicinity of a desk. Closets are the domain of headless servers, not personal computers.