Cooling Technologies & Ideas (Thread Split)

[Tesselator]

For me, the hardest things to deal with, are the small flex PCB connectors used for the keyboards and whatnot, as they can break the lock tabs if I'm not careful.

Yeah the tech would need a bump if we were to base an entire system on FCB technology. Molded connectors, or finger flanges would be a great addition. Still with multi-layer FCB a CPU would look entirely different. They could if they wanted to, place the silicon die almost directly on the FCB which would allow for partial submersion techniques to be employed to a much greater benefit. There are some designs like this already available which include older 8 and 16 bit processors as well as RAM and other various components so while a bit "out there" when discussing Nehalem level 64 bit processors, it's not entirely impossible either.

The fantastic aside, the CPU "package" and some sort of socketing system could fairly easily be designed for a component-level submersion system IMO. This would be ideal and I think if Intel went for something like this we would be seeing 5 GHz ~ 8 GHz just in the second or third generation products.

Righteous! An 8-core Corei7 or i9 at 8 GHz. That would be a nice bump!

 

[DoFoT9]

Oh really? Hehe. Well as we can see in this thread and others, it depends on your point of view. If you use the machine for earning money, then maybe no. If you use it as an desktop PC replacement and don´t earn money by using it, then yes.

i cant see anything from this thread anymore its been forked!! but yes the general concensus is that if you use it for making money then it will pay for itself - it makes you look more 'pro' and 'stylish', people these days think Macs are cool and hip if your a general user and can afford it then why not!?!? ill be buying one when i can afford it (once i get out of stinking uni!), im a pretty heavy user though

Thinking of the sometimes bad performance of Cider Games, you kinda need an MP for games.
Sure sure, the MP isn´t for games...That´s maybe true, but then no Mac is either. And I recall a certain Apple keynote with someone...who was it again...Ah yes! John Carmack? Right?! What does he do? Programming...games? And then there was EA Games. But they are just too lazy to port games to OpenGL, so Cider is fine?
Oh and there was another guy, I think he´s called Steve Jobs and uh...said sth about games on Macs? *JK around*

I know I know, that id Software and EA thing was in the end only a joke and now they do iPhone games insteand of Mac optimized games. Though, the iPhone is running Mac OS also.

yes completely agree, EA = LAME. its basically an insult to think that we will put up with their crappy Cider games... tsktsktsk

Well, I guess many compaints about the MP comes from people which like to play games, too. They want Mac OS for the usual things you do all day but also games and for that, no MP is a good deal. But that was and is already discussed over and over anyway.

the fact that its a workstation doesn't really help either

Well, still no one did explain why Apple uses $300 CPUs, $60 graphics card and optional WiFi card in a $2500/3300 machine and I guess we will never get an answer. So, let´s stay OT.

they want a profit??

OT it is!

(i have no idea about all this cooling stuff so ill keep quiet )

 

[TheSilencer]

Righteous! An 8-core Corei7 or i9 at 8 GHz. That would be a nice bump!

Well, there are 8GHz overclocks at the Overclocking World Records data base.

However, we talking about CPUs and these things currently (speaking of usual INTEL, AMD chips) achieve maybe 70GFlops. The REAL speed bump is offered by using GPUs, then we talk about 1TFlop (GTX285) and more! Another reason why Apple should include high-end GPUs in the future for their OpenCL support, IMO.

And besides of that, the Software is an issue here. Currently except for 3D rendering, some video and audio Software and a few games nothing takes advantages of multi-threading. While we are beginning to talk and want GPU based apps they try to catch up on old CPU tech - multi-threading is basically done since the 1960s and now consumer get the flavour, what a shame.

Besides of the optimization for multi-threading - and there are is not much Software which scalars up from two to xx threads, meaning they often optimized only for 2 or 4 cores, not for 8, 16, 32 and so on - Software also can and has to be optimized for certain CPU features, like SSE.

So, before jumping on high GHz, start optimize first.

The best examples are GPUs there, too. They sure get faster clock speeds but currently we are stuck at 700MHz to 1000MHz clocks for nearly an year while the overall performance (and I mean not just faster FPS, also better eye-candy quality) became better because of more simultaneous processing power and optimized code (like Shader 4.0), which we also see for consumer CPUs now but again, these are slow, very slow, compared to GPUs.

It looks like the CPU manufacturers of the future are actually GPU manufacturers.


Hm, the basic question of cooling a computer is in the end why they can´t cool them to ambient temps. And the answer is simple anyway. As you run the computer, the ambient temp will increase (you have to cool chipsets, HDDs, even SSDs, GPUs, PSUs, everything which works on electricity also) over time, so you have to cool that down, which leads to air conditioners. My computers run the whole day and at summer days with up to 100°F (and inside rooms even more) I like to use an air conditioner to get at least down to 76°F.
So, while the computers may get more "green" currently, all the energy saving is obsolete at least for my work and home place.

 

[nanofrog]

Yeah the tech would need a bump if we were to base an entire system on FCB technology. Molded connectors, or finger flanges would be a great addition. Still with multi-layer FCB a CPU would look entirely different. They could if they wanted to, place the silicon die almost directly on the FCB which would allow for partial submersion techniques to be employed to a much greater benefit. There are some designs like this already available which include older 8 and 16 bit processors as well as RAM and other various components so while a bit "out there" when discussing Nehalem level 64 bit processors, it's not entirely impossible either.

Putting such a large die on flex PCB may have some significant issues with thermal cycling though, even in immersion. By no means as bad as if it weren't though. Now that would be horrible IMO.

It's meant to flex afterall... (BTW, I'm thinking in terms of temp differential from the backside compared to the die surface). With less structual integrity of the laminate, I'd be concerned cycling would break the solder joints (direct or socket mount, though the socket may be less so) over time, and a notable shorter functional lifespan compared to FR4 of proper thickness for the application. Not true DOA, but effectively the same, as electronic repairs aren't common anymore. So no one locally available to resolder (likely), assuming there's no damaged traces.

The fantastic aside, the CPU "package" and some sort of socketing system could fairly easily be designed for a component-level submersion system IMO. This would be ideal and I think if Intel went for something like this we would be seeing 5 GHz ~ 8 GHz just in the second or third generation products.

Righteous! An 8-core Corei7 or i9 at 8 GHz. That would be a nice bump!

Hmm... I'm sensing you're drooling all over yourself!

From a practical perspective, I do wonder how many non enthusiast users would like to have to deal with immersion. Even techs/IT,... in a professional environment. So I'd prefer to keep it to a closed loop system, for practicality sake, and ease of maintenance.

I do think either is quite possible though, and with temps getting to the point of necessitation, something along these lines will exist. Thus economy of scale kicks in, and we can afford it.

I can imagine cases where one of the side panels becomes a passive radiator (for reduced noise, lower complexity). The surface area should be sufficient to work, and low noise has it's market. Personally, the lowered complexity may make it more attractive, but I'm not sure about the cost on it, as aluminum extrusion of an size/weight isn't exactly cheap. Forged may be a possibility as well, especially for internal facing (pressed fins in a grooved plate). Simple, easy, and should be inexpensive.

But I still like the idea of the cold plate example I linked previously. I'm thinking it has the right balance of compromise of adequate performance (enough overhead to last a good while), uses existing technology and manufacturing techniques (time to availability could be weeks, and no manufacturing equipent to invent) , and cheap enough to make the business side happy.

Sort of thinking out loud, so I hope it makes sense.

 

[Tesselator]

woohoo, thread fork.

since we're still speaking some cray, I had this in the picture thread, but this was me and a cray 2 (and a x/mp in the background) that I worked on "back in the day".

Looks cool!

There's one or two PC companies that make fully emersed case systems, and I think Armari's even runs with flourinert:

http://www.armari.co.uk/xcp-gallery.asp

Though it was only prototype as the coolant itself made the machine too expensive (!!)

Hardcore computing has a machine that is retail available that is fully immersed:
http://www.hardcorecomputer.com/

The thermal properties of the cheaper mineral oil coolant in Hardcore's must be waaay lower though because they do immersion AND convective, which seems odd. "Core Coolant is a mineral-oil-like substance that's approximately 750 times denser than air and can store twice as much heat per unit of mass"

"On Reactor, the thermal interface between the CPU and its heat sink is made of indium, which is simple to install and transfers heat extremely well"

Anyonne know the thermal properties of Indium vs copper? I don't have the time to sort past the copper-indium-gallium stuff to dig it out.

Very cool links! I'm not sure I follow the convection remark though. And I dunno too much about indium other than it's atomic number is 49.

But all my ramblings so far have been on the assumption that only certain parts CPU, ChipSet, GPU, Maybe RAM, should be submerged. For the rest air-cooling is good enough.

Well, there are 8GHz overclocks at the Overclocking World Records data base.

However, we talking about CPUs and these things currently (speaking of usual INTEL, AMD chips) achieve maybe 70GFlops. The REAL speed bump is offered by using GPUs, then we talk about 1TFlop (GTX285) and more! Another reason why Apple should include high-end GPUs in the future for their OpenCL support, IMO.

And besides of that, the Software is an issue here. Currently except for 3D rendering, some video and audio Software and a few games nothing takes advantages of multi-threading. While we are beginning to talk and want GPU based apps they try to catch up on old CPU tech - multi-threading is basically done since the 1960s and now consumer get the flavour, what a shame.

Besides of the optimization for multi-threading - and there are is not much Software which scalars up from two to xx threads, meaning they often optimized only for 2 or 4 cores, not for 8, 16, 32 and so on - Software also can and has to be optimized for certain CPU features, like SSE.

So, before jumping on high GHz, start optimize first.

I think vendors have already done most of the "optimizing" for MT that they're going to do. As you say it's already old - so where we are is likely to be about where we'll stay for a good while longer in that regard. But almost all software will benefit immediately from higher clocks. We can get higher clocks by addressing the cooling issues. I think a hybrid system with some components submerged or encapsulated within a flow chamber and the rest of the system air-cooled would be the way to go.

Hm, the basic question of cooling a computer is in the end why they can´t cool them to ambient temps. And the answer is simple anyway. As you run the computer, the ambient temp will increase (you have to cool chipsets, HDDs, even SSDs, GPUs, PSUs, everything which works on electricity also) over time, so you have to cool that down, which leads to air conditioners. My computers run the whole day and at summer days with up to 100°F (and inside rooms even more) I like to use an air conditioner to get at least down to 76°F.
So, while the computers may get more "green" currently, all the energy saving is obsolete at least for my work and home place.

Thus the hybrid idea. Fans will take care of ambient temperatures for SOHO systems as they currently do.

Putting such a large die on flex PCB may have some significant issues with thermal cycling though, even in immersion. By no means as bad as if it weren't though. Now that would be horrible IMO.

It's meant to flex afterall... (BTW, I'm thinking in terms of temp differential from the backside compared to the die surface). With less structual integrity of the laminate, I'd be concerned cycling would break the solder joints (direct or socket mount, though the socket may be less so) over time, and a notable shorter functional lifespan compared to FR4 of proper thickness for the application. Not true DOA, but effectively the same, as electronic repairs aren't common anymore. So no one locally available to resolder (likely), assuming there's no damaged traces.

Right. The die itself only has connections on the perimeters (kinda like shown in the images below added for the sake of example: charge encoder, DDR2 DAC, AMD QuadCore) so a FCB or even just a different kind of packaging could be employed to allow both sides to be submerged. This is the entire reasoning I'm using here as well. I want the CPU up off the fiberglass board so we can submerge just it. Of course this part of the discussion is only a dream and came up just as an example of now negligent CPU chip manufacturers are about cooling, and some simple remedies.

It's a dream not because it can't be done or even that it's not already being done. COG and OLB mounting is commonplace in LCD designs. It's a dream simply because Intel and AMD are not currently doing that with their processor products. WB (Wire Bondning) or FCB (Flip Chip Bonding) on FPC (Flexible Printed Circuit) SMT (Surface-Mount Tech) are better alternatives to PoP (Package on Package) mounting when it comes to submersion cooling but certainly not the only alternatives available. I think many of the AMD graphic cards are already using WB --> BGA (Ball Grid Array) PoP surface mounting and giving us a closer access. They're just still on thick-ass fiber glass ML boards. And it's still PoP right? Or at least it looks like that [WB board level to a BGA SMT package] to me, I'm not sure:

I can just see something like this on a thinner substrate base like maybe that used in FPC SMT. Ever been to a prototronica convention like Semicon West or other? AP (Advanced Packaging) is fun stuff man! There's a few annual shows in Nagoya the manufacturing capitol of Japan, where I live, that I get invites to. I dig it anyway... I know, I'm such a nerd...

Hmm... I'm sensing you're drooling all over yourself!

From a practical perspective, I do wonder how many non enthusiast users would like to have to deal with immersion. Even techs/IT,... in a professional environment. So I'd prefer to keep it to a closed loop system, for practicality sake, and ease of maintenance.

Not really drooling as much as bitching. I thought of this in the early 80's. It's being done in other areas. Why not for PC's? Grr... If it were, then 6 and 8 GHz would already be the target point instead of Intel's now very tired looking 3 GHz. So my bitching really has little to do with the overclocking enthusiasts. All this should be just as transparent to the user as air cooling is now - or even more so.

I do think either is quite possible though, and with temps getting to the point of necessitation, something along these lines will exist. Thus economy of scale kicks in, and we can afford it.

I can imagine cases where one of the side panels becomes a passive radiator (for reduced noise, lower complexity). The surface area should be sufficient to work, and low noise has it's market. Personally, the lowered complexity may make it more attractive, but I'm not sure about the cost on it, as aluminum extrusion of an size/weight isn't exactly cheap. Forged may be a possibility as well, especially for internal facing (pressed fins in a grooved plate). Simple, easy, and should be inexpensive.

But I still like the idea of the cold plate example I linked previously. I'm thinking it has the right balance of compromise of adequate performance (enough overhead to last a good while), uses existing technology and manufacturing techniques (time to availability could be weeks, and no manufacturing equipent to invent) , and cheap enough to make the business side happy.

Sort of thinking out loud, so I hope it makes sense.

Yeah, both are better ideas than we're currently forced to deal with. Neither requires a very specialized process really. http://www.asymtek.com/ We can currently chip-shoot (print) or mount onto just about any surface material. how the heat is carried away is not as important as it actually being carried away. Currently manufacturers would have us believe that we're stuck at around 3GHz because of thermal dynamic coefficients but I say they're doing a bit of wankery. As TheSilencer keeps pointing out; even sunday garage hacks can do better and to a great benefit! Imagine if that were mainline instead of hack-line.

The way it is now is retarded. Basically what they're telling us is: "If it's too hot, blow on it.". My gawd man... LOL


BTW, has anyone in the public eye ever admitted trying "flourinert" to cool their PC? I haven't looked but I will from now. If you guys find anything or know of anyone, post it up. It would be interesting to see the results.


.

 

[TheSilencer]

Well, these Cray cooling reactors are just real hard ware!

I don´t know. As far as I understood flourinert it´s very important to keep that at a certain temp or else it will lose it´s thermic value.

I still think that TEC is the most interesting way to go. AseTek vapochill/waterchill towers are really nice for this. The technology is safe when used the right way and very efficient.

Besides of this, usual water-cooling is still the most popular way. There are water-coolers for CPU, GPU, RAM, HDDs, water-cooled PSUs and chipsets. The challenge here is to make these things so consumer friendly (modular) that you can exchange every piece on the fly and include emergency procedures in case the pump fails and so on, i.e. the computer shuts itself down then. It also would be expensive then, even the tech is old and simple.

Personally I think when looking back in computer history that the most reasonable development is indeed the GPU based calculation and therefore CPU speed is not the goal of the future.
Whatever CPU development will deliver the next year(s), GPUs are and will be fare superior. Even if CPUs will reach, let´s just say, 200GFlops, the next generation of GPUs is supposed to deliver 3TFlops (GTX385).

The idea with better accessable packaging of CPUs/GPUs is cool!

Another thing that I think is technical viewed very interesting, Tesla Power Supply. Wireless electricity. So you can get rid of some VRMs, which produce heat also. The components only chew what they really need. It´s complete insane IMO how i.e. these CPU VRMs work today and the last decades. You fire 12V into them but the CPU runs at 1.25V or even lower. I know, it´s because of Ampere, but still, what a waste.

AND I demand intelligent computers! To be honest, computers are as dumb as they were in the early days but that´s OT.

 

[thepawn]

Very cool links! I'm not sure I follow the convection remark though. And I dunno too much about indium other than it's atomic number is 49.

I meant immersion and conduction, meaning that even though the machine was submerged, they also still had pumped heat sync contacts on the cpu/etc.

 

[Tesselator]

Well, these Cray cooling reactors are just real hard ware!

I don´t know. As far as I understood flourinert it´s very important to keep that at a certain temp or else it will lose it´s thermic value.

I still think that TEC is the most interesting way to go. AseTek vapochill/waterchill towers are really nice for this. The technology is safe when used the right way and very efficient.

Oh, this...

http://www.techzine.nl/exreviews/7951/Asetek-VapoChill-LS-Cooler-at-DriverHeaven.html

and

Yeah, I've linked to those a few times here when talking about OC'ing here and there. I think someone ripped apart their refrigerator and fit it to their PC. LOL. I like'em too from what I can tell by looking and reading. And not too terribly priced either. The CEO keeps an interesting blog.

http://www.asetek.com/content/blogcategory/38/216/
The "What Happened" articles are interesting and relevant to this discussion a bit.

Besides of this, usual water-cooling is still the most popular way. There are water-coolers for CPU, GPU, RAM, HDDs, water-cooled PSUs and chipsets. The challenge here is to make these things so consumer friendly (modular) that you can exchange every piece on the fly and include emergency procedures in case the pump fails and so on, i.e. the computer shuts itself down then. It also would be expensive then, even the tech is old and simple.

Personally I think when looking back in computer history that the most reasonable development is indeed the GPU based calculation and therefore CPU speed is not the goal of the future.
Whatever CPU development will deliver the next year(s), GPUs are and will be fare superior. Even if CPUs will reach, let´s just say, 200GFlops, the next generation of GPUs is supposed to deliver 3TFlops (GTX385).

The idea with better accessable packaging of CPUs/GPUs is cool!

Another thing that I think is technical viewed very interesting, Tesla Power Supply. Wireless electricity. So you can get rid of some VRMs, which produce heat also. The components only chew what they really need. It´s complete insane IMO how i.e. these CPU VRMs work today and the last decades. You fire 12V into them but the CPU runs at 1.25V or even lower. I know, it´s because of Ampere, but still, what a waste.

AND I demand intelligent computers! To be honest, computers are as dumb as they were in the early days but that´s OT.

It's not OT. Can't talk about the future without examining the past. Tesla is definitely a hero of mine! Too bad so much of his work was stolen or suppressed. Especially the world wide free wireless power stuff. It's neat and fun to wrap your head around too. Probably why he's my hero.

And also a big problem with this discussion is that we're doing it on an Apple site and Apple doesn't provide the means to overclock neither do officially support OS X on non-Apple hardware that I know of. Yet more retardation to overcome.

I meant immersion and conduction, meaning that even though the machine was submerged, they also still had pumped heat sync contacts on the cpu/etc.

Ah, OK. Thanks for clarifying man!

 

[TheSilencer]

Yeah, or like these towers.

This AseTek stuff is reasonable but sure nothing what Apple might have in mind for computer cooling.

Well, if you consider "Rhapsody" as part of the OS X development which did run at usual PCs... OK, it was more meant for NeXT computers.

Oh yeah, Tesla was great! The world wide wireless power would be so cool...
Didn´t INTEL showed last year (or this year?) that neat wireless electricity concept?
Well, since Apple and INTEL having a deal (or have to deal with eachother ) why not dream of an Tesla Mac!

An alternate idea for getting rid of PSUs and VRMs could be induced power like this wireless charger for the iPhone.
Imagine you get just an inducer plate and dump a mac at it and it will magically work. Witchcraft!
There would be still the problem of providing enough AMPs but optimized hardware for this technology should run way more efficient.

 

[nanofrog]

But almost all software will benefit immediately from higher clocks. We can get higher clocks by addressing the cooling issues.
Thus the hybrid idea. Fans will take care of ambient temperatures for SOHO systems as they currently do.

Increasing clocks is definitely the way to go for improving single threaded applications.

As for fans, and partial immersion, I think this would be the type of ideas to consider, not only for cost, but ease of internal access. I can't help but think having to drain a sealed case before being able to swap out anything would be both time consuming, and a mess.

Right. The die itself only has connections on the perimeters (kinda like shown in the images below added for the sake of example: charge encoder, DDR2 DAC, AMD QuadCore) so a FCB or even just a different kind of packaging could be employed to allow both sides to be submerged. This is the entire reasoning I'm using here as well. I want the CPU up off the fiberglass board so we can submerge just it. Of course this part of the discussion is only a dream and came up just as an example of now negligent CPU chip manufacturers are about cooling, and some simple remedies.

I think I'm understanding what you want to do (both flat sides exposed for coolant across the combined surface areas, with the contacts to follow the perimeter), but given the pin counts on CPUs, ... I'm not sure it's practical. Not that it couldn't be done, but the perimeter would have to be expanded on each side to accomodate so many pins (over 340 pins per side, just for the current 1366 pin count on Nehalem), giving a rather large package. For board space reasons, it would likely make more sense to edge mount it, and the immersion done via a clamshell arrangement of 2 separate blocks (can't bore holes through the die, or the perimiter board due to trace routing without further increasing the size to make room for holes). Rather complicated.

It's a dream not because it can't be done or even that it's not already being done. COG and OLB mounting is commonplace in LCD designs. It's a dream simply because Intel and AMD are not currently doing that with their processor products. WB (Wire Bondning) or FCB (Flip Chip Bonding) on FPC (Flexible Printed Circuit) SMT (Surface-Mount Tech) are better alternatives to PoP (Package on Package) mounting when it comes to submersion cooling but certainly not the only alternatives available. I think many of the AMD graphic cards are already using WB --> BGA (Ball Grid Array) PoP surface mounting and giving us a closer access. They're just still on thick-ass fiber glass ML boards. And it's still PoP right? Or at least it looks like that [WB board level to a BGA SMT package] to me, I'm not sure:

I think it's TSV (Through-hole Silicon Via).

I can just see something like this on a thinner substrate base like maybe that used in FPC SMT. Ever been to a prototronica convention like Semicon West or other? AP (Advanced Packaging) is fun stuff man! There's a few annual shows in Nagoya the manufacturing capitol of Japan, where I live, that I get invites to. I dig it anyway... I know, I'm such a nerd...

Ceramics would allow for thinner substrates, but are delicate (brittle), so if the stress exceeds the design, it will shatter. It's also an insulator, so on hotter chips, isn't a good idea. Polymers on their own, would end up too thick, but perhaps a ceramide composite would do.

As for attending an AP convention in person, No, I've never had the chance. (Never had the budget available for everything on the wish list, so it ended up on the chopping block).

Yeah, both are better ideas than we're currently forced to deal with. Neither requires a very specialized process really. http://www.asymtek.com/ We can currently chip-shoot (print) or mount onto just about any surface material. how the heat is carried away is not as important as it actually being carried away. Currently manufacturers would have us believe that we're stuck at around 3GHz because of thermal dynamic coefficients but I say they're doing a bit of wankery. As TheSilencer keeps pointing out; even sunday garage hacks can do better and to a great benefit! Imagine if that were mainline instead of hack-line.

The way it is now is retarded. Basically what they're telling us is: "If it's too hot, blow on it.". My gawd man... LOL

Conductive polymer traces are going to be big IMO, but in systems where the thermal cycling isn't severe. Otherwise, the polymer will stretch (ellongate), and the silver may either break contact (grains loose physical contact when poly chains stretch), or change in it's resistive value. This would need to be studied, and there are other polymers that are truly conductive (not just metal filled).

As for clocks, it's not exactly wankery, as the layers do create limits. Add in the lowering of clocks for stability/reliability, and a safety margin (lower warranty returns due to failure), it's rather abysmal we're stuck at under 4GHz. It can be pushed higher, and better cooling at low cost is the key IMO. Another solution is to increase the surface area (fewer layers), but IIRC, has been written off due to cost, as it results in fewer parts/wafer. The wafers are still expensive to produce.

 

[designgeek]

Ran across this in my email, don't know how exactly it could contribute to PC cooling but it sure sounds cool. New Material Could Cool Electronics 100 Times More Efficiently

 

[DoFoT9]

Ran across this in my email, don't know how exactly it could contribute to PC cooling but it sure sounds cool. New Material Could Cool Electronics 100 Times More Efficiently

very appropriate!! however the link takes you to the home page..

 

[designgeek]

very appropriate!! however the link takes you to the home page..

It's working for me. I've recently become obsessed with cooling since my MBP gets near boiling (seriously) while encoding movies. My solution so far is to let it sit outside on a cool night, that keeps my temps down to about 160 F.

 

[DoFoT9]

It's working for me. I've recently become obsessed with cooling since my MBP gets near boiling (seriously) while encoding movies. My solution so far is to let it sit outside on a cool night, that keeps my temps down to about 160 F.

160F?!? is that all????

my CoreDuo MBP has been IDLING at 80°C lately (176F). if i do anything taxing it can easily hit 100°C (212F). the hardware cut off for it is around 110°C -> 120°C.

however i am going to replace the thermal paste (i think i discussed that back a few posts with nano).

 

[Tesselator]

I think I remember something about Apple getting sued for laptops that burned a few people and caught on fire and stuff.

Sounds pretty crazy to me: Mobile parts running at 100c. It seems to me that Apple has heat problems across the board on all of their computing products. I wonder when they'll catch-on and make the proper personnel changes?!?

 

[DoFoT9]

I think I remember something about Apple getting sued for laptops that burned a few people and caught on fire and stuff.

Sounds pretty crazy to me: Mobile parts running at 100c. It seems to me that Apple has heat problems across the board on all of their computing products. I wonder when they'll catch-on and make the proper personnel changes?!?

well seeing as though my problem isnt fully apples fault (the problem being the CoreDuo in my case).. you know...

 

[nanofrog]

160F?!? is that all????

my CoreDuo MBP has been IDLING at 80°C lately (176F). if i do anything taxing it can easily hit 100°C (212F). the hardware cut off for it is around 110°C -> 120°C.

however i am going to replace the thermal paste (i think i discussed that back a few posts with nano).

Yeah, you really need to take care of that ASAP, or you may end up with a dead CPU far faster than you expect. Those temps are way too high for C2D (even mobile).

 

[Dr.Pants]

Yeah, you really need to take care of that ASAP, or you may end up with a dead CPU far faster than you expect. Those temps are way too high for C2D (even mobile).

Yep - clean out some dust and reapply thermal paste! If you have it on, turn it off NOW and read manuals on how to do it on another computer.

Ran across this in my email, don't know how exactly it could contribute to PC cooling but it sure sounds cool. New Material Could Cool Electronics 100 Times More Efficiently

Anybody ELSE worried about how that material looks heterogeneous?! The cooling won't be uniform with that, IMO; probably doesn't matter NOW, but maybe sometime later, etc.

What I like is how its hard to mix Cu and C. Of course it has to be fused at high temps and cooled correctly, or else we get a mix of copper and graphite as the "diamond"'s structure breaks down

Maybe to prove my point here....
Also, I like how they used the term "bond" in the article. Not exactly the right word! One mixes a mixture, not bonds a mixture.

 

[DoFoT9]

Yeah, you really need to take care of that ASAP, or you may end up with a dead CPU far faster than you expect. Those temps are way too high for C2D (even mobile).

i might order it this arvo, i need a MicroSD card aswell for my new phone, and the computer store is right near my uni im thinking some OCZ thermal paste, are they ok?? they have gotten some good reviews.

 

[Tesselator]

well seeing as though my problem isnt fully apples fault (the problem being the CoreDuo in my case).. you know...

Did you changes the processor or something? If not then Apple is responsible - legally speaking as well as commonsensically.

Yep - clean out some dust and reapply thermal paste! If you have it on, turn it off NOW and read manuals on how to do it on another computer.

Or he could just submerge it in a vat of salad oil. Typing might be a little funky... But I dunno, some people like that sorta thing.

 

[nanofrog]

i might order it this arvo, i need a MicroSD card aswell for my new phone, and the computer store is right near my uni im thinking some OCZ thermal paste, are they ok?? they have gotten some good reviews.

It's stock, so any decent past (TIM) will be fine. Most of the different "brands" are just different labels on the same stuff anyway.

Did you changes the processor or something? If not then Apple is responsible - legally speaking as well as commonsensically.

I'm assuming it's out of warranty, so he's on his own.

Or he could just submerge it in a vat of salad oil. Typing might be a little funky... But I dunno, some people like that sorta thing.

Hmm... Lets not go there...

 

[Tesselator]

I'm assuming it's out of warranty, so he's on his own.

Yeah, but if it burns his lap or starts a fire and he sues, then Apple will be paying a lot more than just the price of a replacement.

 

[DoFoT9]

Yep - clean out some dust and reapply thermal paste! If you have it on, turn it off NOW and read manuals on how to do it on another computer.

its my personal laptop! im not going to turn it off!! its been going great 24/7 for the last 3 years, im sure it will last a few more days

Did you changes the processor or something? If not then Apple is responsible - legally speaking as well as commonsensically.

i cant, its soldered on!! but good point, legally its apples issue to resolve.

Or he could just submerge it in a vat of salad oil. Typing might be a little funky... But I dunno, some people like that sorta thing.

HAHAHAHA you need to get that idea out of your head

It's stock, so any decent past (TIM) will be fine. Most of the different "brands" are just different labels on the same stuff anyway.

not this one, the C2D was completely changed in many ways. my CD is pathetic compared to it in heat, performance, longevity etcetc (maybe not price though haha)

I'm assuming it's out of warranty, so he's on his own.

went out of warranty 2 years ago!! but its only a thermal replacement.. its easy..

Hmm... Lets not go there...

totally agree

 

[nanofrog]

Anybody ELSE worried about how that material looks heterogeneous?! The cooling won't be uniform with that, IMO; probably doesn't matter NOW, but maybe sometime later, etc.

What I like is how its hard to mix Cu and C. Of course it has to be fused at high temps and cooled correctly, or else we get a mix of copper and graphite as the "diamond"'s structure breaks down

Maybe to prove my point here....
Also, I like how they used the term "bond" in the article. Not exactly the right word! One mixes a mixture, not bonds a mixture.

Let's skip the crystalline structure for a second.

There's a few variants as well, such as copper-titaniumintermetallide, aluminum, and silver. Perhaps a few others.

Advantages of Cu diamond composites:

• Excellent thermal conductivity (470 W/mK at 20°C)

• Low coefficient of thermal expansion (6.7 at 20°C)

• Well adapted to semiconductors and ceramics

I've also stumbled across a version that's thermal conductivity is 650 W/m-K.

Take a look at this, particularly at page 4 (microscopy image of AgCD composite). Maybe this will help with the first part I skipped (it's homogeneous).

 

[nanofrog]

Yeah, but if it burns his lap or starts a fire and he sues, then Apple will be paying a lot more than just the price of a replacement.

I certainly understand your point, but I don't know if they think that way. Presumably not, unless the laws are in favor of the consumer on this, despite waranty period.

its my personal laptop! im not going to turn it off!! its been going great 24/7 for the last 3 years, im sure it will last a few more days

You should in the interim, while you get some paste in hand and get it applied.

i cant, its soldered on!! but good point, legally its apples issue to resolve.

How are the laws for consumer rights where you are?
That is, do they favor corporations (like they do in the US), or the consumer?
(I'm under the impression the UK has a better outlook for consumers as an example). Unless I totally misread a few posts in other threads.

not this one, the C2D was completely changed in many ways. my CD is pathetic compared to it in heat, performance, longevity etcetc (maybe not price though haha)

In what sense?

Isn't the chip (since soldered to the board), and the cooler stock?