Infrabasse

I think the point where my reflexion is flawed is that I'm picturing heatpipes as tube-like TECs that would stay at ambient on one side and get really hot on the other.

BTW, we always talk about water temperature, but it's the actual copper block that cools the CPU. What kind of temps are we seeing on the actual copper blocks?

__________________
Work in progress

Slider46

Well designed air-cooling setups are compairable to most average water-cooling setups.

I understand your concept but I doubt it's going to perform any better.

If the heat from the CPU has already traveled to the heatpipes and up to the fins of the heatsink - why add another interface through which the heat has to transfer before being dissapated? You're adding another barrier through which the heat has to transfer when it would be simpler and probably more efficient to just slap a fan on some fins attached to the heatpipes, OR go with watercooling from the get-go.

No problem with thinking outside the box though

Direct-Die liquid cooling is already implimented for some high-end systems (pretty sure I read that somewhere recently) so maybe it's possible to bring it down to a level where it would be feasible for consumer use

__________________

Slider46

Are you referring to TEC coolers? On a standard ambient water-cooling setup, the copper block just transfers the heat from the chip to the cooling liquid.

TEC coolers produce sub-ambient temps on one side and produce heat on the other side and thus the need for liquid cooling (for the hot side)

__________________

Infrabasse

No for that question I was on about a normal watercooling setup with a normal block.
The block certainly isn't at water temperature, so my question was what kind of temp is the block base compared to the water?

Now I don't know the exact science behing heatpipes but if they are able to displace all the heat from one side to the other and we can cool the hot end adequately, then the cool side of the pipe (cpu side in direct IHS contact) could drop to ambient....
But that's probably not the case and that's where my mistake would be.




edit: IBM is working on a within-die watercooling system, is that what you read recently? Because what I read about direct IHS and direct die watercooling (within a "half waterblock") is a total FAIL.
edit: and true, the second waterblock on the bottom end of the pipe was just stupid thinking on my side

__________________
Work in progress

c5pilot

I see this as more of a chip manufacturers problem than ours. Take Intel for instance, they can barely cool their CPU's with their own HSF at stock speeds. This has to be a huge concern for them. Forget about overclocking, they need to worry about getting stock temps down so a simple HSF can cool their chips, or they have to decide to go with a more elaborate cooling system which means more headaches after the sale.

Cooling a CPU from only one surface is not helping matters. I "thought" they realized this when they came out with the PII. But for whatever reason, CPU's are surface mounted again. Who knows, they may be forced to mount multiple CPUs around the MB just to find room to cool them separately.

It's no wonder CPU speed is not compounding nearly as quickly as the "old" days. They simply can't cool them with simple technology anymore. I think chip makers will inevitably have to embrace another method other than air.

Our problem isn't so much the design of the blocks, it's using ambient temp coolant. To get to the next step you need to chill below ambient.

__________________

zipdogso

Using heatpipes the CPU could be cooled adequately with proper sintered heatpipes i.e. not the ones they usually use in after-market coolers generally these are the cheaper mesh type. With a properly constructed block on the CPU for the HOT ends of the heatpipes (were the evaporation of the coolant takes place.) and the cold ends could even be extended out of the case to to be cooled by external fans. One of the bad points is that most people these days have vertically mounted mobo's and ALL heatpipes work best wilth a straight vertical run.
A single 12mm dia. sintered heatpipe at room temp. can shift a maximum of about 150w and that was measured in the horizontal position, vertical will obviously be a tad better and the warmer confines of a PC would make the maximum wattage go up a bit as well but like I said we are talking about a proper industrial heatpipe not something you can rescue one weekend from a redundant cooler.

mini1337s

It surprises me that we do not see people selling IHF heatsinks/blocks that are one unit. Direct-die cooling probably wouldn't be cost-effective for any company at a consumer level, wheras seperating the gap between heatspreader and block seems like it wouldn't add much difference to the manufacturing process, but would lower temperatures by several degrees even when used at stock levels. While I may be wrong, it only seems logical that this is true, as you can observe temperature differences between thermal pastes, whether through chemical property or proper application. To be honest, I'm surprised I haven't seen a custom waterblock designed this way, but I may just have missed the thread

Anyways, at the end of the day, Intel and AMD probably don't worry too much about creating extremely effective cooling options as a majority of users will run the processors at stock speeds for a fairly short life-span. I mean, all effective cooling adds is the ability to clock higher-speeds and longer lifespan, both not huge deals to a majority of consumers. Really though, how long does the average joe keep his computer? 3-5 years max? And how important is it to the average joe to overclock his components? Max performance is more an enthusiast thing as average consumer and companies are willing to dish out more money for technology anyways, something Intel and AMD take advantage of. It would be awesome to see high performancing cooling in every computer, but if the masses don't demand, Intel and AMD won't bother supplying.
My

Digital Artist

I approve of this post

zipdogso

Which is exactly why intel and AMD wont provide anymore than an HSF combo it is too expensive and making it idiot proof is too hard....after all even PC noobs sometimes have problems installing a HSF !!!

Decent CPU cooling will always be an after market exercise and as the the TDP's rise the onus will be on us to provide sufficient cooling.

Tommie

You only add more layers between the CPU and the water.