What heatsink are you using on 775 ?

[NappyBishop]

What heatsink are you using on 775 socket ?

[davefr]

Thermaltake Silent Tower. See my comments in the below thread.

[Ralf Hutter]

Thermalright XP-120 with a 120mm Panaflo L1A running at 5 volts.

This is keeping a 3.8GHz Prescott at an idle temp of 41°C and a full load temp of 63°C.

[Leonidas_80]

Heiz,
I bought the Zalman CNPS7000B-Cu cooler yesterday..but something is realy wrong... Im getting 68C when running my P4 3.0Ghz (socket775) on an Abit AS8-V, while the cooler is set at 2400rpm..!! The cooler is put on the ZM-CS1 clip which is needed to support socket 775. If i adjust for 1300rpm my motherboard starts warning me of too high temperatures, over 80C. Can anyone help me on this? Ive checked the installation video at http://www.zalmanusa.com and read the instuctions many times, should be very simple..i havent used artic silver 2, just the one thermal paste provided with the cooler.. Another thing: the cooler is moving a bit if i try to move it, even though its screawed firmly on the clip

[davefr]

I seriously doubt the HSF is making 100% contact with the CPU.

Take it all apart, clean off all the old compound, re-apply just a small amount, and place the HSF on the CPU without any clips or mounting hardware. Just hold it down with with firm hand pressure.

Now boot the system and go into the BIOS and monitor the CPU temp. Is it significantly lower? If so your mounting is at fault.

You didn't say how long it took to get to 68. If it gradually build up temp then you might have deficient case cooling.

I have a Zalman 7000 Al/CU on a P4C800E - Dlx with a Prescott 3.2 that's overclocked to 3.6. CPU temp idles in the mid 30's and in a Prime torture test it still stays below 50. (I think my RPM is set at about 2000-2200).

Originally Posted by Leonidas_80

Heiz,
I bought the Zalman CNPS7000B-Cu cooler yesterday..but something is realy wrong... Im getting 68C when running my P4 3.0Ghz (socket775) on an Abit AS8-V, while the cooler is set at 2400rpm..!! The cooler is put on the ZM-CS1 clip which is needed to support socket 775. If i adjust for 1300rpm my motherboard starts warning me of too high temperatures, over 80C. Can anyone help me on this? Ive checked the installation video at http://www.zalmanusa.com and read the instuctions many times, should be very simple..i havent used artic silver 2, just the one thermal paste provided with the cooler.. Another thing: the cooler is moving a bit if i try to move it, even though its screawed firmly on the clip

[LightSpeed]

I think the problem is lack of fresh air in the case, which is turning the PC into a convection oven with all that copper.
Keep in mind that P4C800 temps are off by at least 10C, and even if you have a thermal sensor on the heatsink, you still have to add 10-15C to get the actual core temp.
I would definitely not recommend holding down that massive heatsink with firm hand pressure that is for sure, especially with a socket 775 CPU! The last thing you need is a little slip or nudge and that will be the end of the computer, with something burnt/dead for sure. I would never put a zalman on a prescott, and in my experience with xp-120 and prescott, it has not been good to say the least. I don't consider 63C under load to be a sufficient cooling solution, when the thermal throttling starts to kick in at 67C. Using the stock intel heatsink and a different fan I get better temps than 62C at 3.9GHz+, 1.4675v (60C load, 38C idle), and with Tt Bigwater that same CPU went down to 52C load/41 idle. That is what I consider to be in the safe zone. Just my opinion.
I also think putting a high speed 120mm fan is nothing short of a disaster with the XP120, and that is why they recommend the L1A panaflo over say the H1A. With all that hot air constantly recycling due to the 100+CFM, ambient temp rose to 34C after 9 hours of 100% CPU load, which was the highest I had ever seen. I removed the XP120 from the loop shortly thereafter. I will re-test with the L1A but I am definitely not impressed with XP-120. Keep in mind that it only seems to work properly with the pipes facing to the right. I know they sure don't work facing upwards.

[wallijonn]

I believe that the pipe bends have to be on the right or the bottom.

As to the XP-120 & the latest Prescott, one reviewer said that the Intel thermal pad or AS5 has to be used and under no circumstances white zinc oxide thermal grease is to be used.

If you are having an oven effect do you have a 120mm fan top blowhole blowing out, with a 120mm intake fan in front?

As to Ralf's PC, remember that he is running it at 5V. That probably relates to about 1000 RPM and 20-25 cfm. You'd have to ask him if he is seeing throttling. What is 63C in F? 145 degrees F. That is a lot less than 181 - 191 degrees F. My 3.0C with a stock Intel heatsink was hitting 146F at load, with white thermal grease and a blowhole. Before the Blowhole I was over 160F (more like 170F).

And yes, when I removed my stock Intel heat sink with white paste the whole cpu chip came out with it. It was a heart stopping moment. Luckily I had no bent pins as I had pulled it straight up. I allowed it to cool as I installed MOSFet heatsinks with Alumina epoxy paste. I figured that it would be as easy to remove as my PIII 1G when it is cooled. I was wrong.

I lapped my XP-120.

ambient temp rose to 34C after 9 hours of 100% CPU load

34C/93F as an inside case temp is a little high. Was it Tomshardware that suggested keeping it under 90F? If that is the case what is one to do when the ambient room temp goes over 90F? Not even bother turning it on? During the summer I keep my room temp at 85F.

.

[LightSpeed]

I know the pipes don't work in any direction, contrary to the original claim of Thermalright. I think the only way that heatsink will work ideally is with the pipes to the right, and if you take the time to look at the design you will see why. In the up/down position, there is a ridiculous overhang either on top of the CPU or below it. With pipes in the downward direction, the tip of the pipes are contacting the CPU, which should theoretically be the coolest point of the heatsink. However the laws of thermodynamics don't really work that way, and the heatsink doesn't either. With the pipes facing to the right, the heat is evenly dissipated across the base, but still off-center. The base of the pipes come in contact with the HS pipes facing right, and that is why it works best facing in that direction
You couldn't pay me enough to use the intel TIM, or AS5. I think it is ridiculous that one could claim not to use white zinc oxide paste, as there is an entire TUBE of white zinc oxide paste included with the XP-120. It looks like arctic alumina to me; too runny to be Ceramique, but just a guess. For that matter I refuse to use anything BUT Ceramique, as it is the safest of all compounds to use, and has excellent thermal capacity, not to mention there is no way it can short out your components.
I don't have a blowhole in the case I used, and am not about to chop a hole in my case for a heatsink. I didn't have any such problems using the stock cooler with 92mm fan in the same case, actually mobo temp never went higher than 29C after a day of CPU load, and CPU temp was still lower at a higher speed. At 3.84GHz using 1.4v, the max temp was 52C using my modified intel solution.
4C is not very far off, and doesn't give much room for breathing. To me that is like running an engine at 6300RPM when the redline is 6700RPM. Sure you can do it, but I don't think it's a good idea. You can even drive with your feet, but you don't see people doing that either, nor does it make it a good idea.
In Ralphs review, he tested the XP-120 in an open environment (essentially a room instead of a PC case), which will never reflect the true values of the XP120 in real world situations. Who do you know with a 12'x16' case?
As a reference, with my 3.0C @3.9 using a swiftech MCX478 I never saw load temps >38C, and mobo temp was 25C.
From my understanding you used thermal epoxy on your CPU which is the worst thing you can do. If you applied epoxy and expected it to be removable, well here's some news...read the instructions
I have pulled my s478 out of the socket attached to the heatsink, and it didn't make my heart skip a beat, maybe because I know there is nothing wrong with that, and I have had it happen several times with no ill effects. My best 3.0C had at least 20 pins bent right out of the box, but hey it overclocked better than all the rest. Keep in mind that to remove a heatsink easily all you have to do is take it off while the grease is warm (give the CPU some load first).
I didn't see any reason to lap my XP-120 as the base reflected quite well.
Intel validated solution states that ambient air temp should be <38C in the case, and at the heatsink/fan inlet. I was not about to sacrifice my entire system to prove a point, and found it a little strange that after a day of testing, coincidentally 4 caps on my mainboard were leaking and burnt.
I don't believe in coincidence, bad timing...*maybe*. I know for a fact that there was nothing wrong before I installed it because I took pictures of the installation, and also afterwards when I pulled it out.
Lets not forget what a pain in the *** the XP-120 is to install on socket 775 mainboard. If you try to do it yourself, well good luck, but ideally you need 2 people to do it. Oh...and if you think that a CPU is hard to remove with ceramique, don't even bother trying to remove the adhesive pad from the back of the mainboard if you use the s775 bracket. It really bakes on there and is a pain to remove, basically if you use the XP-120 for say 3 days, that backplate is there to stay unless you soak your board in alcohol. Ease of use, I really don't think so.

[J Cool]

Using an XP-120 with a 120mm Panaflo L1 in a Lian Li V1000B case:

Idle 43 C
Load 59 C, Prime95
Ambient 21 C

I'm happy with this combo and tried a Swiftech first with mediocre results.

[Ralf Hutter]

Originally Posted by LightSpeed

I know the pipes don't work in any direction, contrary to the original claim of Thermalright.

FWIW, when I got my XP-120 I put that claim to the test. Using an 875PBZ + 3.0GHz P4 NW, I positioned the board (I was running it outside the case on my test bench) in it's "normal" (standing upright, with the CPU at the top of the board) orientation, as well as turning it 90° clockwise & CCW, and I also ran it laying flat on it's back. Full, 2xCPUBurn load temps varied no more than 1-2°C from one orientation to the other, so I'd say Thermalright is correct about that.

[wallijonn]

LightSpeed,

So, putting the XP-120 on a Socket T is a no-no? Broken caps.?; ouch.

Next time around I'll use Ceramique for the cpu seatsink.

I used to use the original Acrtic Silver and I put too much on a Northbridge (or it dribbled off when it became hot (runny)) and killed an ABit BX board. I've never used it since. That was a $125 lesson. By the time I upgraded to a BE6v2-II, Arctic Silver was no longer in the house - I had thrown it away.

I took a MicroCool Southbridge heat sink and sawed it into nine smaller heatsinks which I epoxied onto my PBZ MOSFets. I'm now looking for a shorter Southbridge heatsink (about 1/2" high) as the one I bought was too high (about 1+3/16") and would have hit my video card. Yeah, my MOSFet heatsinks are 1+3/16" high.

I had the cpu open for over a day (I let the epoxy set overnight). While it was drying I was lapping the XP-120.

Had I known that it would've come out when the white zinc oxide was cold (and it did seem as if it was a thick coating) I would never had let it cool. I don't remember having to apply any undue pressure to install the stock cooler. Heck, I would've removed the fan and retainer mechanism and then run a thread or wire thread under the heatsink to break the seal.

I had absolutely no trouble installing the XP-120 on my PBZ, even with the heatsinks which are longer / higher than the large caps. around the CPU. I was very careful installing it.

Here's where Tom said that white thermal paste shouldn't be used (and that the Intel thermal pad or AS5 should be used hwne putting on accessory heatsinks):

http://www.tomshardware.com/cpu/20041114/index.html

Notice that he used white zinc oxide thermal paste on the stock cooler. That is exactly what I did with my stock 3.0C cooler. Beside not trusting thermal pads if you ever remove the heatsink you should replace the thermal interface with a new one.

It would seem that if you have a 560 Prescott you will have to have a thermometer positioned over the case exhaust fan because if it ever goes over 100F / 38C you could cause problems.

http://www6.tomshardware.com/cpu/200...ages/case2.jpg

That's the same type of case I have. I removed the lower disk caddy and lower front purple 80mm fan holder, made an enclosure out of Tupperware and installed a 120mm fan. I blocked off the rear two 80mm fan holes and installed an external 120mm fan blowing out of a top blow hole. No, my wiring looks nothing like that. All my cables are dressed in, with tight bends; everything is tie wrapped down.


.

[LightSpeed]

Ralph, don't get me wrong; your review was one of the deciding factors in my purchase, and I thought your review was very concise. The only thing I didn't take into consideration at the time was the difference in CPU, and lack of it being in a case.
I am very curious how your tests would have turned out had you used your same prescott rather than northwood in your tests, and on a socket 775 mobo. I highly doubt that the results would be the same with the same CPU that runs 63C under load, especially when inside a case, not in an open room.
Northwood don't get hot enough to require such a monstrosity, and that heatsink is overkill to do the job IMO. The only true way to test a heatsink like that is with one of the scorching hot CPUs that are available, not one that can be easily cooled with even the stock cooler at up to 3.7GHz, like a northwood.
I never had as much of a problem installing a heatsink in my life as I did when installing XP-120 on socket 775, and I have tried more than I can keep track of.
The original Arctic silver wasn't as bad as some of the more runny versions such as AS3. What I like most about ceramique is that it won't become runny under any circumstance. It may be a little harder to spread due to its thickness, but at least you have the assurance that it won't be going anywhere it shouldn't be going. Swiftech bundles Ceramique with their socket 775 waterblock as well.
I also have a Microcool southsink which is incompatible with 875/865 chipset; at least any that use PCI cards. Installing it on my P4C800 would mean no use of slots 2-4 which is out of the question. I wonder how hot the southbridge gets on intel mobos, but doubt it is as hot as AMD solutions.
I will be chopping mine up but have no use for the pieces yet, as I got mosfet heatsinks at the same time.
Tom's hardware is so one-sided lately that it makes me wonder how much AMD is paying him these days
The problem he encountered was that the intel pad acts more like an insulator than a heat transfer compound; you can see the same effect on your northbridge, which by default will not get hot if you leave the pink crap underneath, but if you replace it with a decent compound like ceramique, you will see how hot it gets as a result, and also that it now requires active cooling as well. With zinc oxide or AS5, heat was being transferred too fast for the fan to keep up on the stock solution. If he used a constant speed fan which moved a decent amount of air (~50CFM) he would see a world of difference; I sure did (15C to be exact).

[wallijonn]

Originally Posted by LightSpeed

your northbridge, which by default will not get hot if you leave the pink crap underneath, but if you replace it with a decent compound like ceramique, you will see how hot it gets as a result, and also that it now requires active cooling as well.

You mean the PBZ Northbridge can be removed? I thought it was epoxied in.

Then I probably had AS3 because it was runny. Snce then I always wait 5 minutes when applying white thermal paste - just in case the silicon starts to separate. I want air circulation to dry it out before I mate the parts.

[Ralf Hutter]

Originally Posted by LightSpeed

I am very curious how your tests would have turned out had you used your same prescott rather than northwood in your tests, and on a socket 775 mobo.

But I have used the XP-120 with a Socket T 3.8Ghz Prescott, as I posted about at the beginning of this thread:

Originally Posted by Ralfie

Thermalright XP-120 with a 120mm Panaflo L1A running at 5 volts. This is keeping a 3.8GHz Prescott at an idle temp of 41°C and a full load temp of 63°C.

An L1A at 5V is probably putting out a whopping 20-25CFM and that combo still keeps that Prescott cool. I can only guess that running the same fan at 12V, or running something like 60-80CFM fan like so many do, would keep the CPU much cooler than with my very quiet setup.

Originally Posted by LightSpeed

Northwood don't get hot enough to require such a monstrosity, and that heatsink is overkill to do the job IMO. The only true way to test a heatsink like that is with one of the scorching hot CPUs that are available, not one that can be easily cooled with even the stock cooler at up to 3.7GHz, like a northwood.

Note that I write for a website called SILENTPCReview. For us it's about running very quiet systems. In order to do that you need quiet (generally slow running) fans. These fans don't put out tons of airflow so you need to put them on a very efficient heatsink. That's exactly where the XP-120 comes in. It's the most efficient heatsink I've ever used (and I've used hundreds) and this lets you cool the CPU very well with only a very slow running (and very quiet) fan. I couldn't care less how well the XP-120 cools a massively OCed CPU, I only want it to give me at least adequate cooling with the lowest possible noise. And because it's so darn efficient, it does a wonderful job of this.

Originally Posted by LightSpeed

I never had as much of a problem installing a heatsink in my life as I did when installing XP-120 on socket 775, and I have tried more than I can keep track of.

It is a bit of a pain to install, and I would certainly never try it on a CPU without an IHS, but that's something you don't need to worry about with a non-mobile P4 so for me, it's not a super big issue.

[wallijonn]

Originally Posted by Ralf Hutter

I can only guess that running the same fan at 12V, or running something like 60-80CFM fan like so many do, would keep the CPU much cooler than with my very quiet setup.

I read an article which said that higher air flow helps only to a certain output level. Once that level is reached, usually due to the design of the heatsink, no amount of extra air flow will help to cool it down further.

Along with that test / review they did a funnel test, the reasoning being that a fan sitting on a heatsink will usually position the the center of the fan over the center of the cpu which is where the most heat is generated. What is logically generated is a dead spot where the most cooling is required.

They found that it really didn't make a difference. The reason being that you have a flat surface (the bottom of the air stack / heat sink) and no matter how much air you force in there will be air bounce at the bottom / back pressure created. This is why some heat sinks have a fan blowing accross the heatsinks (as opposed to blowing down onto the heatsink) and why other heatsinks have round fins versus square fins.

So, yes, you can put a fan that outputs 130cfm on a cpu heatsink. But you may only need 40cfm to cool it to a point where any more air really doesn't cool it any further. You reach a point of diminishing returns.

If I were to install a 130cfm fan on a cpu I would have a duct which only input fresh air, there's no way I'd have it re-circulating the hot air inside a case. And unless I have a 140cfm exhaust fan I would just leave at least one hole without a fan due to slower fans acting as back pressure. Not only that, but if I do not have a 140cfm fan I'd have a 240mm hole so that the case would again not act as an air baffle. (Look at the Lian V1000 case).