cooling performance

[monohouse]

I recently bought a nice swiftech water cooling system, but the performance is disappointing, the parts are very good and impressive but something just doesn't add up.

using the old water cooling system, a thermaltake kit, I had a cpu temperature of 60 C, at room temperature of 35 C, and video card temperature of 60 C (with a thermaltake W2 block).

with the new parts, mcp 355 pump, gtz, mcw60-r2 and mcr 320, I have a 55 C cpu temperature, at room temperature of 33 C, and video card temperature of 55 C (fans = 7V, at 12V there is maybe 1 or 2 degres difference).

I don't know what it is or why, but im sure that 500$ should do better, much better.

what could be the problem ?

[genec57]

What are you cooling? That has a lot to do with it. Check your mounts. A poorly mounted block can make a huge difference in temps.

[monohouse]

nothing special, just a dual conroe and a 3870 all at stock voltage
what should be the temperatures at ?

[Metric]

nothing special, just a dual conroe and a 3870 all at stock voltage
what should be the temperatures at ?
Temps should be much lower, somewhere around mid 40's to lower 50s under load depending on your ambients. In your case, is the 33c room your ambient, because that would explain the high temperatures...

I'm running a Q6600 @ 3.6, X38 chipset and a 8800GTX in a single triple radiator (PA120.3) and I load in the low 50s with a load delta a little under 20c.

[monohouse]

so 33 + 20 = 53
but you run quad core, and overclocked with overvoltage. i run dual and without overvoltage and the GTX is probably both bigger die and bigger process
the mount of the cpu block looks fine, I tried to make it tighter but nothing changed.

and right now the room temp is 31 C but the cpu cores are at 55 - 56 C

[millertime359]

Is your room seriously 95 degrees Fahrenheit? :eek:

That explains part of the for the high temps. I recommend getting an AC unit, or if one is not available, put computer in Fridge. Just make certain you leave room for beer. :yepp:

[monohouse]

miller, no matter what the conditions are, it can't be that the difference between a thermaltake kit and a swiftech non-kit can be so small

[genec57]

Going from air to water, even not so great water is a huge jump. Going from not so great water to better water is not that big a jump. In water cooling the benefit from money expended decreases the higher you go. All that notwithstanding I think your temps are a bit high. The most likely culprit as I said before is the mounting of your cpu and gpu blocks. A good quality thermal compound can make a difference.
Air in the system can mess things up as well.

[millertime359]

The conditions will make a difference. Your room temps directly affect your CPU temps. With a room that hot, your are causeing your cooling system to work a bit harder. What fans are you using? Your next thing to do would be to try getting some higher speed fans.

edit: A GTZ is a bit hard to mess up the mount though. IMO, you would see higher temps if there is a bad mount. What TIM was used?

[monohouse]

yes, it seems there was some grease missing, I have 2 greases, formula 5 and arctic silver 5, im using formula 5, it looks like I didn't put enough grease in the cpu, so now I added more grease, but the temperature did not change, it looks the same...

[millertime359]

I know AS5 is a bit outdated now. I'm not certain about formula 5 though. How much did you use? MX2 is suppose to be a bit better, and sounds like MX3 can lower temps a degree or 2 beyond MX2, but some of the comments I have read say it is a bit of a pain to get a good mount with the MX3.

What fans are you useing?

[monohouse]

various fans, all are different, formula 5 is just 99% silver it's not taking prisoners and doesn't try to be non-conductive, but I heard the grease has no effect on performance, it's like pipe size, also have no effect on performance. but still my grease is good anyway.

I thought about the fans, but after checking that angle, and to see the difference was not high when running at 12V, it can't be.

maybe the pump is running the water too fast, can this pump be lowered to work at 7V ?

[millertime359]

Take a look a Vapor's review on TIMs. They can make a difference. http://www.xtremesystems.org/forums/showthread.php?t=232141

The pump speed will only matter if the fluid is moving too slow. I doubt the pump is making any difference on that small of a loop.

Fan speeds also make a difference. You can see by this graph:

http://www.skinneelabs.com/Radiators/Triple-Rad-10C-Delta.jpg

[millertime359]

i just found this in another thread that might help you understand your room temp makes a difference. Naekuh broke out the science:


okey...

common misunderstand a lot of people make is when they get watercooling your computer does indeed get colder, but your room will get hotter.

Its thermodynamics:
Energy can not be created nor destoryed.

So if your computer is running cooler, it means your pulling more heat out, and therefore the heat needs to go somewhere.
Since water pulls the heat out faster, and more efficiently, unless you have fans in your room to expel the heat, it will build up and eventually toast your room.

Also, think of it this way. Your are outside and it is 100 degrees. You want to cool down. What would cool you down faster, 95 degree air, or 75 degree air?

[monohouse]

I understand that, and also the thing with the fans, but the reason I am sure the fans have no effect, is because the amount of air going from rad extremely changes when set to 12, while the cooling performance does not, so the rad and all the stuff outside the case, can't be the problem, so you are sayin, the friction from the water should not have effect on cooling performance, only if it's too slow, this pump is many things but slow isn't one of them, I tried to plug it into 7V but the pump won't start at that voltage.

strange, I put the pipe more deeper into the water container, and now when the pump shuts down there is no water going to the back of the pipe and air no longer going back into the pipes, and I remounted the block once again without adding grease this time.

the room temp didn't change, but the cpu temps are 51 - 52 now, strange where the 4 degrees came from.
also strange is that video card temperature is the same as cpu, also 51 - 52

[Boogerlad]

millertime, I hope you were joking about the fridge idea. That won't work as that will burn out the refrigerator's compressor. Those compressors were not designed for constant heat loads.

[monohouse]

haha fridge, sounds good, I was thinking of something like it:

put the radiator into water, and every time the heat builds up in the water, take ice from the fridge and put into the water for the radiator.

so, does my temperature make sense now ?

[millertime359]

millertime, I hope you were joking about the fridge idea. That won't work as that will burn out the refrigerator's compressor. Those compressors were not designed for constant heat loads.

:D Yea, I was joking.

I imagine the hard drives might not be happy about it either. There is also the issue of getting cables out of the fridge.

[NaeKuh]

I understand that, and also the thing with the fans, but the reason I am sure the fans have no effect, is because the amount of air going from rad extremely changes when set to 12, while the cooling performance does not, so the rad and all the stuff outside the case, can't be the problem, so you are sayin, the friction from the water should not have effect on cooling performance, only if it's too slow, this pump is many things but slow isn't one of them, I tried to plug it into 7V but the pump won't start at that voltage.

strange, I put the pipe more deeper into the water container, and now when the pump shuts down there is no water going to the back of the pipe and air no longer going back into the pipes, and I remounted the block once again without adding grease this time.

the room temp didn't change, but the cpu temps are 51 - 52 now, strange where the 4 degrees came from.
also strange is that video card temperature is the same as cpu, also 51 - 52

... okey im kinda lost in what your trying to say.

Fan speed will have impact on rad cooling. The more air you get though the rad, the better it performs, (smaller your delta becomes.)

However, u must only look at those charts when you use fans of = or greater specs. (Martin used Yates.)

So if ur using crap fans, which have crap static, then yes, you wont see a jack difference from low to high, because the static on the fan is crap.

When you look at fans for a radiator, you dont only look at CFM. You need to look at STATIC pressure.

Also friction from water almost never plays a role. You have a much bigger heat source then friction, ie PUMP.
And even then the pump is considered a very small displacement in heat.

[monohouse]

looks like those readings were bogus, it wasn't 51 - 52, it's because the system boot at strange speed. the real temperatures were:
core0 56, core1 54, gpu 53.

I have an another fan, a 220 V fan that will fit in the radiator, I could add it, and I could also add the old radiators, and add some more fans, since the pump is good it shouldn't be a problem with all the added resistance. but is that the only reason left for the low performance ?

[rge]

The other thing you need to consider, is you are not testing the limits of either cooling system at that low TDP, ie running a core 2 duo at stock.

My i940 on custom water (PA 120.3, 6 x 1850 GT fans push/pull, ddc3.2 pump) at stock loaded at 50C. That same cpu/mobo is now my wifes computer, but she is on scythe/air cooler with single 1600 rpm fan, and loads at stock about 57-58C. So only 8C difference, and reason is at stock, the TDP is very low about 80W, and does not really stress the air cooler or water cooler. Your dual core is going to be even less TDP.

But at 4.5Ghz and 1.4vcore, water cooling keeps cpu at 75C prime load. My wifes air, those settings hits 100C instantaneously and bsods. Temp difference is greater than 25C.

Point is, a dual core running at stock is not hot enough loaded to show that much difference between two water systems. My guess is it would not show much of a difference even on high end air. If I could send you my fully loaded 4.5ghz, 1.4 vcore cpu at 250+W, you would likely see a large difference in core temps between the two systems.

Same reason why idle temps are not that different between air and water, but rather load is where much larger difference is seen. To little heat to really stress either cooling system. You can always get better fans and drop a few more C, but you will hit diminishing returns quickly cooling such a low TDP.

[ridney]

i would suspect that it is a bad mount on the cpu block or the pump is not flowing water, maybe some trapped air? if you're thinking of the rad fans not flowing enough air through the rad then take out those fans and use a stand/desk fan instead to see if you get lower temps and isolate the problem.

[voigts]

I'm trying to figure out why you are stressing over idle temps between the Thermaltake and Swiftech setup with no comparison of load temps. That is like taking a vw bug and a porsche idling side by side and saying that they run the same.

[Conumdrum]

Yea, need load temps. And what are you using to measure temps for the CPU? Hopefully something like coretemp, realtemp, HW Monitor.

[Caoder]

also don't use thermaltake rads in the loop, they're aluminum........ fyi

Also i'm getting ~42 under load on my oced cpu and gpu. Idle's 35~ with 25 ambient.

[Jamesrt2004]

what order is your loop in?

in my opinion get some decent fans not random ones you have lying around... and DONT judge idle temps they mean jack.

[d50man]

Mono house get your room temps down and add aftermarket fans to your system

[eth0s]

IMHO, lowering idle temps by 5C is a HUGE improvement (especially in a 35C room). Whoever told you otherwise is an ass-jacky idiot. But, regardless of your huge improvement in idle temps, you really should be comparing 100% load temps, as the wise men above me have already pointed out. So if you don't already have a program to stress your dual core cpu, you'll need to find an older dual-thread version of Prime95, which I'm sure you can find it at MajorGeeks.com, or one of those type of free download sites.

Now, for the other issues: (1) You do not want to slow the water down. This is a bad idea, and this notion seems to be the persistent hobgoblin of little minds. Now, even though it's been said 10,000 times on this very forum, here it comes for the 10,001th time, and maybe you'll listen this time: increasing flow rate increases cooling performance, while decreasing flow rate will decrease cooling performance. You never, ever, as in like never, and when I say never, I don't mean sometimes, I mean always no, 100% certainly not, nope, not gonna ever want to slow that water down. Not ever. Not even on Pluto or Comet Haley, not even with green eggs and ham, never, no, no way, no how.

(2) Fan speed matters! Faster fans is better. Muy rapido es muy bueno (as they say down Mexico way). Si, mi gusta el guapo (which the Senoritas always say to me, but I can't speak Spanish, so I don't really know what the heck this means). Anyway faster fans are better. (although to be technically correct, you should follow NaeKuh's advice and go for higher static pressure always.) But I know you won't believe me, because ye are of the slow-water people, so what can I do? Shall I stand on my head? Shall I do a little dance? Or I guess, I could cite physics equations, I could cite other online articles, I could cite actual real-world testing, I could do these things, but would it change your mind? Would you listen? I doubt it.

(3) News Flash: Tubing size matters as well! The bigger the tubing diameter, the more water flows through the tube, and the more heat is absorbed and carried away to the radiator! That's how it works, it's like an eight lane freeway compared to a 2 lane country road. It is just that simple! I can't make it any plainer.

(4) Ambient temp is going to govern you ultimate performance level. I guess what you are saying is that you want a 2-5C air in to water out delta. Well Martin has done plenty of testing, and has come to the conclusion that in order to get that small of a delta, you are going to need a quad rad for the cpu alone, and a second loop with a quad rad for the gpu. Then you will probably get even lower idle temps. How low do you expect to get in a room that is 31-35C? If you really want to improve the cooling of your computer, figure a way to cool the room. Water cooling is really a misnomer I guess, it really should be called "water-assisted air cooling", because that is what it really is. You are still using the air in the room to cool the computer, you are just using the water to help you do it more efficiently.

(5) And finally, welcome to the club! You spent $500 and the results were not as great as planned? Oh man, I've never heard that one before. Oh young grasshopper, someday you may learn to snatch the pebble from the old master's hand, but not today! This is an expensive hobby, and you can spend a fortune chasing another -1C delta, and that is what we do, so again welcome to the club, you are in good company.

[millertime359]

IMHO, lowering idle temps by 5C is a HUGE improvement (especially in a 35C room). Whoever told you otherwise is an ass-jacky idiot. But, regardless of your huge improvement in idle temps, you really should be comparing 100% load temps, as the wise men above me have already pointed out. So if you don't already have a program to stress your dual core cpu, you'll need to find an older dual-thread version of Prime95, which I'm sure you can find it at MajorGeeks.com, or one of those type of free download sites.

Now, for the other issues: (1) You do not want to slow the water down. This is a bad idea, and this notion seems to be the persistent hobgoblin of little minds. Now, even though it's been said 10,000 times on this very forum, here it comes for the 10,001th time, and maybe you'll listen this time: increasing flow rate increases cooling performance, while decreasing flow rate will decrease cooling performance. You never, ever, as in like never, and when I say never, I don't mean sometimes, I mean always no, 100% certainly not, nope, not gonna ever want to slow that water down. Not ever. Not even on Pluto or Comet Haley, not even with green eggs and ham, never, no, no way, no how.

(2) Fan speed matters! Faster fans is better. Muy rapido es muy bueno (as they say down Mexico way). Si, mi gusta el guapo (which the Senoritas always say to me, but I can't speak Spanish, so I don't really know what the heck this means). Anyway faster fans are better. (although to be technically correct, you should follow NaeKuh's advice and go for higher static pressure always.) But I know you won't believe me, because ye are of the slow-water people, so what can I do? Shall I stand on my head? Shall I do a little dance? Or I guess, I could cite physics equations, I could cite other online articles, I could cite actual real-world testing, I could do these things, but would it change your mind? Would you listen? I doubt it.

(3) News Flash: Tubing size matters as well! The bigger the tubing diameter, the more water flows through the tube, and the more heat is absorbed and carried away to the radiator! That's how it works, it's like an eight lane freeway compared to a 2 lane country road. It is just that simple! I can't make it any plainer.

(4) Ambient temp is going to govern you ultimate performance level. I guess what you are saying is that you want a 2-5C air in to water out delta. Well Martin has done plenty of testing, and has come to the conclusion that in order to get that small of a delta, you are going to need a quad rad for the cpu alone, and a second loop with a quad rad for the gpu. Then you will probably get even lower idle temps. How low do you expect to get in a room that is 31-35C? If you really want to improve the cooling of your computer, figure a way to cool the room. Water cooling is really a misnomer I guess, it really should be called "water-assisted air cooling", because that is what it really is. You are still using the air in the room to cool the computer, you are just using the water to help you do it more efficiently.

(5) And finally, welcome to the club! You spent $500 and the results were not as great as planned? Oh man, I've never heard that one before. Oh young grasshopper, someday you may learn to snatch the pebble from the old master's hand, but not today! This is an expensive hobby, and you can spend a fortune chasing another -1C delta, and that is what we do, so again welcome to the club, you are in good company.

nicely said, think you dropped a few zeros there though. ;)

[monohouse]

ok from the beginning:
NaeKuh: static pressure is more important than CFM, I will buy fans of this type if and when I get the chance. but I am sure that if I replace one of the fans on that rad with a 220V AC fan there will be no reason to suspect that fans are the problem ?

rge: agreed, TDP, but here is the funny thing: when I bumped up cpu voltage from default to 1.38V temperature bumped up from ~55 to ~60, which is obviously nothing to compare to what you reached - which I guess you mean I should increase even more and then there will be less difference ?

ridney: bad mount ? I re-mounted 2 times already - the first time I noticed there was grease missing - so I fixed it, the second time to see if I mismounted - and I didn't, im pretty sure there is very little can be improved from trying any more times, air in the water : yes there is some, I can both see and hear it in the pipes assuming that is what I am hearing, there is some strange random noises coming from the block and the pump, but maybe it's just the water turbulence. it's really hard to say that this is because of bad mount, all the remounts showed the same temperatures, with the added grease and without it (the block has good surface I guess that is why there was no change).

voigts: all temperatures are under full prime95 load, not just all cores but also furmark for video card load to affect the cpu temp.

Conumdrum: temps are hw monitor yes.

Caoder: not bad for that, i guess during winter maybe my temperature will be like your, but the thermaltake rads are not aluminium, it's copper pipes with aluminium fins.

Jamesrt2004: the order of loop: tank > pump > rad > gtz > mcw60. the loop is serial.

eth0s: I never mentioned anywhere any idle temps, to clarify again: room temp = ~35C, load temps ~55C for all components with prime95 and furmark with affinity. the attempt to slow down water failed, the pump does not start at 7V and now that you mention it, I guess there is no need to try any further is there. and for the news flash - I have read the water cooling tutorials on this site, all of them - one of which has done extensive testing on pipe size effects on temperature and found no difference, be it as it may regardless: I have good pipe size, not too big and not too small 3/8" so no problem there. it's not 5C idle, it's 5C load.

for the moment I am leaning towards rge's point, it's like mentioned in one of the water cooling guides here, the higher is the temp on the rad and the higher is the difference between the air in the rad and room temp the faster it gets cooled, just didn't have idea that this applies to blocks just the same, so im gonna try to further increase voltage and see how the stress is handled, maybe it won't increase that much beyond 60.

one other point I am wondering might have an effect: parallel loop
I guess that's another thing I could try and see if it affects, the thermaltake system was connected in parallel, so maybe that is why the difference is so small, but maybe not.

as I've seen many wondered, the idle cpu temps are between 1-2 C above the room temp, at around 36-37, but those temps were identical to the thermaltake system, so no improvement there.

I discovered there was a leak in the mcw60, it flooded the sound card below it, strangely the system continued to work as normal, even if the entire sound card was flooded, I guess this could explain why there was air and bubbles in the system all the time, and never went away.

with this opportunity I added the 240 rad from the old system, and made a parallel loop with it, to ensure that from now on rads and their fans are out of the equation.

[Humminn55]

(3) News Flash: Tubing size matters as well! The bigger the tubing diameter, the more water flows through the tube, and the more heat is absorbed and carried away to the radiator! That's how it works, it's like an eight lane freeway compared to a 2 lane country road. It is just that simple! I can't make it any plainer.



I can....tubing size is essentially irrelevant to cooling performance.

In case you missed Cathar's testing of various tubing sizes, the result was, when comparing the same heat load on different tubing diameters, there was a 1C difference between 1/4" ID tubing and 1/2" tubing. 3/8" ID tubing vs. 1/2" ID tubing showed around a 0.25C difference on the same setup, same heat load.

So, tubing size is pretty irrelevant to cooling performance.

Here's the thread for this testing results for your reading pleasure....it's found in the stickies. (http://www.xtremesystems.org/forums/showthread.php?t=147767)

As he said at the end of the testing,


It's (the performance difference between tubing ID sizes) certainly not the 5C figure that people bandy about. I never expected that it ever would be myself. In my own testing with arbitrarily choking the flow-rate in a test-system, I've always been amazed at the low flow resilience of many setups. Below 2LPM is where things start getting pear shaped quickly for most systems.

[millertime359]

Parallel won't make much difference on your loop. I tried a parallel run directly off the pump and temps didn't change.

Also if you want to do a parallel loop, you will need to get a T3 or XSPC res.

For peak temps, they really are not that bad. The 2 things you need to change, which should lower your temps a few more degrees is getting some good fans, and lowering your room temp.

For fans, Yate Loons (http://www.petrastechshop.com/12yalod1cafa1.html) are cheap and work great

Gentle Typhoons (http://www.performance-pcs.com/catalog/index.php?main_page=product_info&products_id=24749) are a bit more expensive, but are high quality fans with a great noise quality compared to Loons.

[monohouse]

ok, powered up with the extra rad using parallel, no change, all the temperature is exactly the same, looks like it's not the rad or the fans after all.

so far only 3600 mhz, 1.4V (1.36V), ~60 C but only with fans at 12V, at 7V it goes up to 64 C

[Caoder]

well one other thing about thermaltake rads are that they are flow killers. along with the fittings that they use.. 1/8? or something like that....

[monohouse]

no, not in this case, it's 3/8" like the rest of the system, maybe the 120 additional is 1/4", but I didn't use that one, only the 240 part, but still it's flow killer for other reasons like 10 passes. I wonder if it gets any heat into it, maybe because the MCR gets more heat into itself not much left for the other to do...

[rge]

@ monohouse, one of points I was trying to make, at higher tdp you will see a larger absolute drop in temps with better cooling than lower tdp, even though percentage is same, you take a percentage of much higher number. So even if you ignore all other variables, you will see bigger difference.

radiator/fans
old rad/fans assume .05 C/W and new system much better at .02 C/W, and will ignore pump and few things since irrelevant for the point.
.05 C/W x (70Wcpu load + 50Wgpu idle) = .05x120 = 6C air to water delta old system
.02 C/W x 120W = 2.4C air to water delta new system
So at low TDP load of 70W, you saw 3.6C improvement in temps from increase rad/fan performance.

And lets say your waterblock with better design improved heat transfer a certain percent dropping your load another 1.5C. So your load temps decrease 3.6C + 1.5C or ~5C in all.

Now same, but with 250W cpu load and 50W gpu idle.
.05C/W x 300W = 15C delta air to water in old system.
.02C/W x 300W = 6C delta air to water in new system.
So you dropped your load temps by 9C just from better fans/radiator, despite percentage improvement 70W cpu and 250W cpu were same.

Waterblock improved the same percentage heat transfer... so 1.5C improvement on 70W load, is now ~5C improvement on 250W.

So 70W cpu sees 5C total improvement with better cooling, the 250W cpu sees 14C with better cooling, even though percentage wise the same....again ignoring any other variables just to illustrate the effect of higher absolute numbers despite percentage same.

Also, now even higher fans on 250w cpu with 6C air to water delta, if improves cooling another 33%, will drop temps 2C.

Higher fans on 70W cpu with already low 2.4C delta air to water, if improves cooling another 33%, will drop temps .8C, so see much smaller gain, though percent same.

as an aside, I just put my system down to stock, and my TDP via everest was 60W load, my delta air to cpu prime load was ~20C. Yours is 22C (55C load with 33C ambient). You did get a significant improvement in performance with your upgrade, just your are using small numbers TDP, so looks smaller. Your current delta air to cpu is pretty good, but if you want lower temps, you need to lower ambients.

[millertime359]

ok, powered up with the extra rad using parallel, no change, all the temperature is exactly the same, looks like it's not the rad or the fans after all.

so far only 3600 mhz, 1.4V (1.36V), ~60 C but only with fans at 12V, at 7V it goes up to 64 C

No, I'm not saying another rad won't help. It may if you can't lower your room temps. I just wouldn't use the Thermaltake stuff, get another Swifty rad.

I'm saying parallel won't necessary help that loop. It is a pretty free flowing loop and a traditional parallel loop actually hurts flow and doesn't improve temps.

You need the T3 or one of the XSPC dual DDC pump res' to do a parallel properly.

As you see, fans do make a difference. You saw 4 degrees difference from undervolting your fans.

[monohouse]

sure, but one of the fans doesn't even start to work at anything less than 12V, so it's like a whole fan shuts down at 7V, left with only 2 fans at that voltage.

so I guess the air to load delta is not so far off, maybe because of the leak and now again reconnected everything back as before without the TT rad and back to serial, maybe if now without leak, there will be no air there will be improvement, but still it's funny how the sound card was flooded and everything including it still works just fine. so I guess nothing is wrong after all, maybe the leak was it, don't know, will see after air leaves the pipes, after that it's just maybe the fans, the ambient is a different story, air conditioning is 2000+ watt business, might as well go phase in that case, but I am not extreme, so I don't run air conditioning just for the computer, though the fridge deal did sound good, too bad it wouldn't work.

im gonna see what happens when I increase more voltage, maybe if I can get 4000 mhz with 65C and 1.56V it will be promising.

still, there is a huge difference in video card temperature, even if cpu didn't improve that much, the video card load temperature dropped from 68 C to 53 C, 15 degress, not like the double from stock to water, but still is very huge, can't compare to the TT W2. maybe reached limit of cpu heat dissipation.

anyway, I bumped everything up, 1.525V, 3880 mhz, in prime (with furmark) it gets very hot ~72 C, but in normal applications it does not go over 60 C... strange

[monohouse]

just to let you know something I just discovered:

I disassembled some of the loop, and by accident discoverd that some of the fittings I used were from the thermaltake system, I used them only for the rad because I didn't get any fittings with the rad, those fittings were surely for 3/8" ID, but after I checked it manually the actual ID of the fitting is 7.5 mm which is not 9.6 (3/8" ID), apparently "thermaltake 3/8" ID" is not the same as 3/8" ID, I checked the swiftech equipment, and it is 1 cm diameter (bigger than 3/8" ID even if it says that it is only 3/8" ID), but that was just the start, since I didn't order a container I used the thermaltake container, and the fitting there was 6 mm ID ! it just gets worse and worse, it seems that you were right, the thermaltake system are 1/4" ID, even if they fit 3/8" ID pipes !

someone once said before "and get rid of ALL your TT equipment" and he was right, I just didn't think that he also meant fittings LOL.

anyway, what I did to fix it is first of all use the "beer mug" method for the container (just put the pipes into the water), and used the 1/2" ID fittings for the radiator which I have left from the gtz block, somehow with some effort I managed to barely fit the 3/8" pipes, I checked the thermaltake pipes - they were actually 3/8" ID so no problem with that. I wonder if that could have a significant effect, because I have yet to check it, probably will later today.

one question though, the GTZ's barb fittings appear to be using some kind of polished metal ? do i have to add anti corrosive because of it ?

[Titan7171]

I understand that, and also the thing with the fans, but the reason I am sure the fans have no effect, is because the amount of air going from rad extremely changes when set to 12, while the cooling performance does not, so the rad and all the stuff outside the case, can't be the problem, so you are sayin, the friction from the water should not have effect on cooling performance, only if it's too slow, this pump is many things but slow isn't one of them, I tried to plug it into 7V but the pump won't start at that voltage.

strange, I put the pipe more deeper into the water container, and now when the pump shuts down there is no water going to the back of the pipe and air no longer going back into the pipes, and I remounted the block once again without adding grease this time.

the room temp didn't change, but the cpu temps are 51 - 52 now, strange where the 4 degrees came from.
also strange is that video card temperature is the same as cpu, also 51 - 52

Im confused here.........you have air in your tubes?:confused: By what Im reading It sounds like you have a system that drains out into a container when you turn it off then fills the tubes when you start it up again? If this is what you are doing you might be allowing air bubbles to acumilate in your water blocks when you start it back up.

[Caoder]

no barbed fittings are coated brass = no effects when mixed with copper. Also get rid of all of ur TT stuff everything hTT uses has the low ID fittings and are flow killers in general. either way.. even though this won't affect you're temps too much it'll inmprove it a bit. i think next step is to get decent fans. From my knowledge... i believe the 3870s ran hotter than 4870s, but the dual core should run cooler than a phenom x4... so i still don't understand the high cpu temps. What kind of TIM are you using? and how much are you putting on there....

[monohouse]

I put just enough grease to make sure it spread everywhere on the cpu and evenly and enough. but even when there wasn't enough grease there was no change in temperature, it's good to know I don't have to add anti corrosive, so I will just use distilled with anti algae additive, im using Antec formula 5 99% silver grease everywhere (like SB and NB), I don't like the grease they put into the SB/NB from the stock, don't like it at all.

I understand now that the sounds I was hearing from the gtz were because of air inside it, everytime there was sound in it I could immediatly see bubbles coming from the out port, I fixed this problem by turning the radiator upside down for start, and then after a while turn it back to normal, now I can't hear anything in the gtz, maybe a little, but very very rarely, the air does not go back into the pipes anymore as long the pipes in the container are under the water.

it would seem my suspicion was correct after all - I did have a problem, but it seems to be all me, from the air in the tubes, the leak, the restriction from the fittings, so yhe, now it looks like I fixed all the problems, now it's just down to fans and proper fittings for the rad...

I should mention I am very impressed with the pump, there is no comparison to the thermalcrap where it took hours for water to get to the cpu, not to mention start flowing through it, with the mcp within less than a second the water was already going out into the container, it's like the water just flies not flows.

[SoulsCollective]

Y'know, I'm surprised noone so far has mentioned IHS issues. If your CPU is badly concaved or has terrible contact between the CPU die and the IHS, your temps won't be fantastic no matter what you do.

[monohouse]

I have no regrets, this equipment is stellar, it's just not the same as what I expected, but we are yet to see the difference with all the fixes, problem is that I afraid to power the system on, because I think I have psychological effect from the burned mobo and sound card because of the leak, so now that I replaced the mobo I think I smell smoke coming out from the new system, but there is no smoke anywhere, and when there was smoke when the old mobo burned, the system auto shut down, now it doesn't im so stressed ? I dont know what to do ! the smell is from the burning of the old sound card and mobo and it is still there a few days later so I afraid that it comes from the new system even if it doesn't and there are no overheating components as far as I could see by touching them, and even increased voltage in the video card now, and the mcw60 is holding.

Im so stressed, I assembled everything correctly it seems because the system boots just fine and works. but because I afraid from what burned last time and the smell of the burned plastic that stayed from the last time I shut it down.

[meanmoe]

I understand now that the sounds I was hearing from the gtz were because of air inside it, everytime there was sound in it I could immediatly see bubbles coming from the out port, I fixed this problem by turning the radiator upside down for start, and then after a while turn it back to normal, now I can't hear anything in the gtz, maybe a little, but very very rarely, the air does not go back into the pipes anymore as long the pipes in the container are under the water.

it would seem my suspicion was correct after all - I did have a problem, but it seems to be all me, from the air in the tubes, the leak, the restriction from the fittings, so yhe, now it looks like I fixed all the problems, now it's just down to fans and proper fittings for the rad...


always fully bleed a system. The air not only impedes the flow and contact between the water and heat exchanger surfaces but it can also act as an insulator. (it's noisy too)

I'm happy you solved the problem, now go put the TT stuff in the garbage.

[monohouse]

anyway here are the new results, at room temp of 32 C, voltage of 1.47 V (hwmonitor) with the GA-p35-DS3R, set to 485 FSB 1:1 * 8 (3880 mhz), I get up to 61 C with prime on both core, but when assigning furmark to core #0 and prime to core #1 core#0 temp goes up to 68 C (improved compared to 72) strange that, anyway looks like this mobo has much worse memory performance, 50 cycles longer memory latency.

but by comparing the same testing method to both cooling systems, then all my expectations (and some more, such as video card at 1.57V DMM at full load no higher than 55 C, the temperature seems like it's voltage independent :)) have been met, I don't have results for the thermaltake cooling system performance at that setting during this room temperature because I could never reach stable at these frequencies so I never set them, but I can be sure of this which I did set: at room temp of 20 C, the thermaltake system with the two prime95 testing method reached 50 C, and now at room temp of 35 C it works at same frequencies and voltages but at 60 C temperatures of cores, which definetly suggests what was mentioned earlier about it - the higher is the heat load the better is the cooling performance.