Core temp issue

[Acebmxer]

While my system was down waiting for RMA on PSU I decided to redo my water loop.

Here is a pic of the before



And here is the new setup



The other thing I changes was TIM's I found the little bit of MX-2 that came with the EK water block and decided to try that.

After starting the system up idle temps look fine. I loaded up LinX while it was at 99% load temps looked great but once it hit 100% temps are about 10c higher then they should have been. So it then switched back to good old TX-2 and temps are still off. I am using the easy mounting system with the EK block which does not allow you to over tighten but still just as tight as it was before I moved everything. Also I am positive there is no air in the system as I had it running out of the case for a day running it on and off.

I could try another remount of the block but I did it the same as I always do. I changed it to the new layout thinking I would see better temps if at all as the tube coming from the rad to the cpu is shorter. Also forgot that I switched the directions of the fans so they are blowing out instead of in as there was to much dust in the case.

Any ideas?

[boondocks]

Re: Core temp issue

Well your water system definitely looks much better! MX-2, is that Arctic Silver? Haven't used that one. I have swapped back and forth between TX-2 and the paste Noctua includes with their coolers and found no difference in cooling.
There always seems to be a tradeoff between dust and cooling. My place is extremely dusty, and it's mostly pretty dry here. It's not unusual for me to clean filters every 2-3 weeks and blow the case out. Have you tried setting the fans up with a positive pressure?
i.e. plenty coming in, not as much going out.

[Acebmxer]

Re: Core temp issue

The MX is Artic cooling I believe. In those pics it is hard to tell but the rad is setup in push pull so a total of 6 fans running at max. These fans Yate Loon 120mm x 25mm UV Reactive LED Fan - UV Blue (D12SH-124UB) - FrozenCPU.com

Positive pressure or negative pressure has no effect on the cpu while its water cooled as far as I know.

The cpu used to top out at 82-84c while running 4.6ghz HT off @ 1.45v not its hitting 90-93c. Room temp is 70F or 20c.

[wazza300]

Re: Core temp issue

could be your fans,those are the main changes blowing out now instead of in?

pick some dust filters up for the top mounted fans,cant think of what else it could be

[boondocks]

Re: Core temp issue

Yeah, I'm just saying for dust control you might try positive pressure. As far as your temps, haven't got a clue.

[Acebmxer]

Re: Core temp issue

I should find some dust covers but honestly this cooling system sorta amazes me but at the same time thought it would do better. In the past up in till now has always been constant from time put together till 3-4 months for cleaning with the radiator would have a 1/4 think coating of dust and the fans covered slightly in dust temps never changed maybe 1-2c but thats a stretch.

also another thing its hard to see but old setup I actually only had 5 fans, 2 push on the bottom and 3 pull on top. I could not fit the 3rd fan as I had 2 45 fittings on the rad. The new setup I rotated the rad and changes the fittings around and allowed me to use the 6th fan, and still I have temp issues.

Ill wait to see if anyone else chimes in to see if the fans blowing in opposite direction rely make that much difference.

[Acebmxer]

Re: Core temp issue

Yeah, I'm just saying for dust control you might try positive pressure. As far as your temps, haven't got a clue.

That's why I swapped the direction of the fans to see if it would cut down on the dust as other way was killer.

[wazza300]

Re: Core temp issue

if the outside air is cooler than the air inside your case it makes sense to have them drawing air in from outside? only way to know would be to test it,any womens tights hanging around?(ohh kinky) pull one over the top of your entire pc case,quick and easy dust filter

[Acebmxer]

Re: Core temp issue

For ships and giggles I did another re-mount and maybe helped 1-2c drop. Doing more test since the rebuild I loaded Optimized defaults and have entered overclock manually and only ran LinX for maybe a 1 to 2 passes as it didnt take long for temps to show a problem. This time I tried loading a profile that was known to be stable and it failed each time asking for more vcore, 4.6ghz has been stable with 1.45v. I did notice that gflops have been more constantly on the higher side then the past.

Is it possible that the new PSU is able to push a little more or have some sorta effect on stability and higher core temps?

[Lsdmeasap]

Re: Core temp issue

MX-2 temps will be a little less than TX-2, maybe 3-5C sometimes, so that would have a little to do with it at first. I love both MX-2 and TX-2, TX-2 offers better temps but I can only buy 3-4g at a time, so I often end up using MX-2 more because I can get 30g tubes for around $22 or so shipped.

Same Vcore/Vtt/Vccsa as before, are you 100% sure, and same BIOS? I mean it sounds like you haven't used the board in a while so maybe you aren't certain. Also, same graphics cards, memory, ect? I see you also switched pumps, that may add some into the differences as well. And looks like you also changed the flow direction, or does it just look that way?

And yes, fan direction could make that much of a difference, also the amount of fans has changed too. The only way you will know for certain if this is the issue is to put everything back exactly how you had it before, otherwise it's just wasting time trying to figure it out if this might actually be the issue.

New PSU shouldn't affect Vcore/stability or temps at all, except for adding it's own heat into the case.

[Acebmxer]

Re: Core temp issue

Positive everything is the same except the layout and the PSU same model just bran new. Same bios just loaded optimized defaults on start up to do some memory benchmarks for wazza. Pump is the same with the top just sitting in different postion. Flow is the same res>pump>radiator>cpu>res, the in and out ports of the rad did get switch only for ease of routing the tubing. I dont think that should affect things that much.

I did have to pull the ram out to remove and install the rad but they are installed correctly maybe not in the same physical slot they came out of. I loaded a profile that was know stable or was at least 10-15 pass stable now its crashing 1-3 passes.

Tomorrow I guess I will swap the fans around to blow back into the case like the used to and see how that helps.

[Lsdmeasap]

Re: Core temp issue

Thanks for the further info, guess I was all wrong about most of what I thought!

I'd go ahead and reseat the CPU since your stable profiles are not anymore, and maybe clear CMOS for a bit while you do that. Also, are you looking at temps with all the same versions as before? Maybe software or BIOS detections were adjusted, or are you using DMM?

[Acebmxer]

Re: Core temp issue

Well I flipped the fans around on the rad and notice no change.

[Acebmxer]

Re: Core temp issue

I dont know whats going on... So confused. I swear the only thing I touched was the PSU and the water cooling. Never touched the bios before or doing the swap, current bios installed is f5c.

Since swapping the fans around did not help and 3 remounts and 2 tim changes did nothing. I am redoing my overclocks. The chip has been very reliable and predictable after playing with it for what 2 yrs or so. Since what was "stable settings" are no longer I decided to redo my overclocks. I am starting over at 4.2 ghz and either its been so long since I last ran this speed. I typed in settings that should worked and notices still high temps so I started to drop volts. I was able to drop vcore and qpi/vtt lower then what worked in the past.

Maybe the PSU had other issues other then the break in the wire on the 24pin cable.

Another thing I noticed is the difference temps between core1 - core4 are now different. Before in real temp I had to adjust tjmax on core4 to 107 and idle calibration to -4 just to match the readings from core1 for idle and full load; core three was set 104 and -2 i think. Now cores 3,4 are set to 104 and 0. Core2 was only off by 1c compared to core1.

Im still testing will see what happens when I get to 4.5 and then push 4.6 again.

[boondocks]

Re: Core temp issue

Ace, did you ever get your temps back down?