Build log - project heat pump part 1

My system

CPU - AMD Phenom II X4 965 BE C3 125 watt @3.4ghz
Ram - 4GB Gskill Ripjaw @ 1333mhz in dual channel 7-7-7-21
GPU - ATI 5850 non reference GPU@725mhz | Memory@1000mhz

HDD only a 250GB Seagate ST3250310AS and a 80GB WDC WD800AAJS since all my storage is on a NAS that has 5TB.

Optical Drive - Who cares
Mobo - Asrock m3a770de
OS - Windows 7 ultimate 64 bit



I can get my Phenom to 3.8ghz stable on air cooling so I hope to achieve 4.4ghz with the new cooling system.
(Finally run PCSX2 games at max and I won't need to get bulldozer until a much later date)

Ok, this looks intriguing  . Keep posting  

What did you order and what do you have?

So what I have bought so far is this.

1. 2x120mm Swiftech radiator
2. 2M of Gigabyte 1/2'' tube
3. 2 x 1/4'' BSPP 1/2'' barbs for the radiator
4. Swiftech MICRO Radiator REV 2
5. Swiftech mcp655 pump
6. Arctic thermal compound
7. Arctic Alumina thermal adhesive
8. 2 x 91Watt peltiers
9. 2 x AMD Fan + Heatsinks (ones on the bottom without fan)
10. 2 x Quite Fan + heatsink (ones on the top)

I am just taking a while to decide on the CPU block. So that and coolant is the only thing I haven't baught.

I must be missing something, your making a custom cooler for the water?

2 parts I am making.

1. My actual liquid cooling loop just like anyone else.

2. The ice box that cools the radiator. (There would be tubing coming from the radiator in the renders but I was a bit lazy)


The renders are the ice box.

I have tomorrow off and will upload the rad and pump also the pelts and fans should be in tomorrow.

Bit of delay in hong kong where the fans and pelts are coming from but I expect them here tomorrow Monday Australia.


Also I have chosen the water block and its the EK Supreme HF Acetal.

I have decided to have the entire project as one instead of two and here is some stuff that came in.

Here is the reservoir that came in.




And here is the pump



And the radiator plus barbs I had to buy



And the thermal grease



And the Gigabyte tubing




Just waiting on the water block. Plus the pelts and fans still haven't come in but they are ebay things so it takes a while.

Just tested a simple loop with the pump rad and reservoir and it worked fine. I am liking the speed control on the mcp655.

Also the system help just under 600ml of water so when I have the cpu block there will also be more tube in use so 600ml seems like the volume of the loop.

The MCP655 and EK Supreme HF should be a good combo. My full copper runs very well (~55C while stress testing a 3.8 GHz OC that didn't work) in my loop, and I would have gotten the acetal for aesthetics if I could have found a cheap used one.

I'm considering running my tubing outside my apartment in the winter here; there's snow on the ground from December to March. Natural loop cooler  

Damn I can OC my Phenom 965 to 3.8 on air and have it stable.

But I only see from renders and conversions a 5% increase since its only a 400mhz OC which isn't much.


With LC I think it could be stable at 4.2 and hopefully with my peltier cool box I can hit 4.5ghz which would be a huge 1.1ghz OC which would be significantly faster.


BTW the EK extreme I bought is the acetal one.

I was just messing around and didn't bother to adjust the voltages. It stress tested fine but games were unstable. It's also one of my first forays into OCing as well.

The EK Acetal looks nice  

Well to be honest when my Phenom is at 3.8 it is stable but it heats up my room quite a lot so I put it back to its stock clock.

Also I am on a 64 bit OS and I hear for some reason the 965 can't get past 4ghz on a 64 bit OS. I guess I will just have to wait and see. Luckily I also have my drive partitioned with a 32 bit os.

I was thinking if the results are like this.

64 bit os - 4ghz

32 bit os - 4.5ghz


Would you switch back to 32 bit.

I don't think I'll go back to 32-bit OS again. 64 bit is much more stable, more possibilities...and clock speed isn't everything in the world of computing. I know for some, its a big deal and I admit...it was for me a few years ago. But now, versatility, stability and expansion are far more important than a few extra hundred mhz.

Can you answer this because a bunch of guys in the cpu section I was arguing with were saying a quad @ 2.5ghz would be better than a single core @ 10ghz.

I don't see how that would be possible because the quad has cpu core sync problems where that doesn't factor for the single core.

Also I find having more cores than any game or software is threaded for stupid but they seemed to think the more the better.

As is the case so many times, it depends on what your doing.

Look at any CPU review. Using iTunes or Lame to convert MP3s is a single threaded task that loves clock speed. In this case the single core 10GHz chip would be faster. (assuming the same or similar arch.) Converting a movie however using handbrake is a task that loves cores. Here I'm not sure if the quads would really win out or finish about the same.

Gaming however should be faster on the quads. A single core playing a modern game would finish one thread, then dump the pipeline and begin the next. While the quad would have to flush the pipeline as well, it can still work on the game while some cores flush. You also need to take windows into account. The single core will need to flush to perform the background windows tasks. So the single core might be faster, or it might not.

Here is my last question. If a program is multi threaded and you are using a single core would it be less effective than even if the program was only coded for one thread.

So lets say we have.

Photoshop programmed for one thread using a one core cpu.
vs
Photoshop programmed for 4 threads but using a one core cpu.

Is there any speed difference.

Assuming the same CPU, there should be no difference in speed. 1 thread will get executed at a time no matter what.

It all really depends on the majority of the software being used. Anymore, most software is developed to be multi-threaded, including virus scanners. Speed is not the only variable here...you have to consider the instruction set capabilities of the processor generation/model you are using and the ability for it to complete instructions per clock cycle vs. another chip. When you start talking about an old P4 or Athlon at 3ghz and a current-gen quad core running a single core at the same speed, the newer gen chip will always trump older gen on a 1:1 comparison.

Trust me on this...it's not just about speed or cores...it's also about the instruction set and calculation ability per core, per clock cycle, multiplied by the number of cores.

I knew all that!. I was simply pointing out a quad clocked at a quarter of a single core going up against that single core. Assuming they have the same instruction sets I think the single core would win because it has no sync problems. For example I have a quad and in Xilisoft video Converter Ultimate you see the first two cores at about 90% and the other 2 are hanging around 50%. This is core synchronization problems that come with multi cores. Thus my quad is 3.4ghz and I think if I had a single core @ 13.6ghz which is 4x it would win in converting. Most OC champs OC the most powerful core.

The pelts came in.




Water block tomorrow and who knows when the heatsinks are coming in.


The other images aren't working anymore because hardcore upload exceeded their bandwidth but I am not changing them because they will be viewable again next week when hardcore upload is in a new week.

In theory, I will partially agree with you. However, this is not the case in real-world application, since it is an extreme rarity that a CPU core clocks over 5-5.5ghz...I think I've seen a small handful of 6ghz OC's somewhere, but those were LN2 or Helium cooled benches. I get the point you are trying to make, but for the vast majority of folks, they don't benchmark for a living...they use their PCs for a lot more.

The AMD heatsinks and fans came in, As you can see I disassembled them because the heatsink will be used on the cold peltier side.




Now just waiting for the fans that actually cool the hot side to come in.

Here is the old aquarium pump I am going to use to move the water in the cool box.

It has some stubborn dried algae as you can see. Very small 6 watt pump without the filter obviously.

Just tested the pelter cooler and got it to -21 degrees Celsius.


The setup was a container with about 5cm of liquid in it. The heatsink was placed in the water and the peltier hot side was placed on the heatsink, then the small pump was used to pump the water through the heatsink fins to cool them.

A large bolt was placed on the top of the peltier and filled with water and a thermometer placed in it. After 50 seconds I had -21 C.

Overall I am happy with its effectiveness.

Water block came in.




Also the block didn't come with any barbs but I am ordering a second couple of the swiftech barbs because they fit nice.





BTW what coolant do you all use.


I am thinking of using 9 parts distilled water to 1 part car coolant since it will stop bacteria that thrive in distilled water but the whole mix won't be conductive and with the 10% coolant in the mix freezing won't be a problem even though I don't think I will get to zero anyway.

I read that a lot of coolants made for pc cooling perform worse than distilled water.

That is true, that distilled water will perform better than most coolants... Instead of using car coolant, maybe you could use a couple drops of a biocide like PT_Nuke... Unless it will freeze the water, but if the water is moving it probably wont freeze very quickly.

Can someone please tell me why not to use car coolant as a 10% mix. I hear car coolant is an anti corrosive and also slows the leaching of metals in the loop plus increases the life of plastics.

Though I hear some people say don't use it but never explain why.

Because you don't need to, there isn't any benefit...if you really feel its necessary with mixed metals, you should be changing coolant more often and cleaning your blocks just in case, as well. You can use it if you want, but it isn't going to do a lot other than some minor corrosion resistance and maybe lower the freezing point of your water (which may be a benefit to your build), but there are other alternatives as well.

Sorry I don't know whats up with them fans maybe I got scammed?.


Anyway I have thought of a better idea to my liquid cooling loop instead of this current method which requires the pelts to cool too much liquid.


So I am thinking of neatly cutting square holes in both sides of my reservoir and placing thick copper sheet in their which will be glued in with silicone. Then the pelts will be thermally glued on them sheets. Thus the pelts are cooling a small volume of liquid improving the setup. Opinions.