Sign in with
Sign up | Sign in
Your question

Temps a bit high for water cooling. Help??

Last response: in Overclocking
Share
January 23, 2012 3:39:39 AM

I recently bought a refurbed X58 SLI LE board as my main board is going through rma.Here is my build:

i7 950 with apogee gtz
16gb g.skill
120gb vertex 2
xfx6950 with EK copper block
X58 sli le
corsair 850
1tb
250gb
mcp655
swiftech micro res.with silver coil
480 rad 120mm nidec betas at 3200rom each on a push


Now my previous board was a gigabyte ga-x58a-ud3r. i had it going from res>pump>rad>cpu>gpu>res. using primochill 1/2ID tubing. Also if this may help, I RMA my cpu as i could not figure out if it was the cpu or mobo gone bad. anyways intel sent me a brand new chip.
so i have the same loop going in with the sli le board except in trying out the feser tubing 1/2ID.
My temps with the giga board and older cpu were around 32C and my graphic card 33C.

with this sli le board, my temps in real temp show 56C-62C and gpu at about 49C. I reseated the cpu and reapplied mx4 twice.
now on the board display, it reads at about 48C.

tubing is at same length at the other tubing used for the giga board.

Any thoughts.
a b à CPUs
a c 325 K Overclocking
January 23, 2012 1:57:44 PM

Quote:
with this sli le board, my temps in real temp show 56C-62C and gpu at about 49C


Idle or load? I am assuming load? You mention 32-33C from the older board, which would seem to be idle...so just curious on what is what. Depending on a few things, it's normal to get a bit different temps with different hardware. It's possible that the 2 boards regulate voltages to the CPU a bit differently, and one gives a bit more, resulting in a little higher temps.

Did you use RealTemp on the old setup? Tubing length wouldn't have a play in this, so you can count that out. (if the tubing were several yards/meters longer in length, you could see temp difference from thermal dissipation of so much extra tubing, but not even within 2-3 feet of difference.)

Do you have any partial tubing kinks? Any noticeable restrictions?
m
0
l
January 23, 2012 4:33:04 PM

those temps are idle. thought i said that. sorry. The sli board is an 09 model and the giga is a 2010. the giga has more copper in it so im assuming that plays a role too. and yes, maybe the boards to reg. voltages different. I did use real temps on the old setup.

There are no restrictions or kinks. I replicated the exact same loop from the previous build, thats why i mentioned same length tubing. This feser tubing actually bends better then the primochill tubing and feels a bit stiffer than it too. i get my giga board back Wednesday so i will place that board back in and see how the temps play out. right now as im typing this, real temps shows 56C at the lowest and 62C on another core. My gpu is at 51C. Would there really be a difference of 20 to 30C between the two boards?
m
0
l
Related resources
a b à CPUs
a c 325 K Overclocking
January 23, 2012 5:11:13 PM

Quote:
Would there really be a difference of 20 to 30C between the two boards?


No, there is likely something wrong. Are you getting good flow rates? It's possible that you have a decent amount of air in your loop in your rad(s). If everything is working as it should, I would start pointing blame on air in the loop or some kind of partial restriction...

Are you certain your blocks are making good contact?
m
0
l
January 23, 2012 5:28:43 PM

yes the blocks are making good contact. i reseated the cpu twice along with new paste. i did the same for the gpu. i bleed all the air out the loop. There are no restrictions either. I may switch back to the primochill tubing to see if anything changes. Im sure that wont make a difference but i wont know if i dont try. And if that doesnt change anything, i guess i will have to wait until wedns for my giga board to come back. Like i stated, everything is exactly the same, other than a different mobo.
m
0
l
a b à CPUs
a c 325 K Overclocking
January 23, 2012 5:37:21 PM

That's just really odd...tubing won't make a difference though. This is one of the more odd scenarios we've had in a while...hmmm...
m
0
l
January 23, 2012 5:48:04 PM

i know the tubing wont make a difference, but it gives me a reason to rip it all apart and put it all back together again. Im wondering if it has to do with the CPU. Like i said, intel sent me a new CPU that i am using for this evga board. I know new does not mean no problems.
m
0
l
January 23, 2012 5:51:03 PM

It is quite high for an idle temp. And you're using bigger rads. Mines stays at 35 to 37 c on idle on both CPU & video cards. I run a fully separate loop on each blocks. A thin 120 mm rad for the CPU & a thick 140 mm rad for the video card.

I was curious of the possibility of using small radiators to cool things down. I pointed the fact that all these hybrid liquid cooling products made by CoolIt & Asetek & rebranded by Corsair & Antec all used small rads. After reading the review of PNY GTX 580 Liquid Cooled GPU & CPU I decided to go & build a PC using small rad & to add (if needed) a secondary external passive radiator for added performance.

My test without the secondary external passive radiator proved what I believed that it's possible to cool things without using massive set ups. Otherwised, both CoolIt & Asetek don't put this hybrid liquid coolers on the market.

It's proven to be true. I've tested & tested the temps even gaming for several hours & still getting good temps. For my cpu (Core i7 2600K overclocked to 4.8 GHz) with gaming it goes up to around the low to mid 40s c & for the video card at around the 50s c. The temps will go down at 35 to 37 c after I stopped gaming.

I even replaced my CPU water block (OCZ Hydroflow) with Swiftech Apogee HD when I read that the right water block plays a significant role in cooling. I barely saw any significant change in temp with my new CPU water block. I think it's just pure BS if you ask this (so called) experts.

There's something there in your set up that getting you those high temps & need to look at it very closely.
m
0
l
January 23, 2012 5:59:13 PM

yea this is really puzzling to me. those idle temps are on air temps. though a process i dont want to do right now, i might place the heatsinks back on the cpu and gpu and see what temps i get on air. test, trial and error is all i can do at this point.
m
0
l
a b à CPUs
a c 325 K Overclocking
January 23, 2012 6:00:18 PM

Quote:
Mines stays at 35 to 37 c on idle on both CPU & video cards. I run a fully separate loop on each blocks. A thin 120 mm rad for the CPU & a thick 140 mm rad for the video card.


You'd see better temps if you ran a single loop with these rads and blocks.

For CPU blocks, you are only going to see 1-2C difference in load temps, at most, between any of the top 5-7 blocks out there at any given time. Performance gains have worked out so that most competitors are almost on a level playing field...most of it is aesthetics and personal preference.
m
0
l
January 23, 2012 6:13:45 PM

rubix_1011 said:
Quote:
Mines stays at 35 to 37 c on idle on both CPU & video cards. I run a fully separate loop on each blocks. A thin 120 mm rad for the CPU & a thick 140 mm rad for the video card.


You'd see better temps if you ran a single loop with these rads and blocks.

For CPU blocks, you are only going to see 1-2C difference in load temps, at most, between any of the top 5-7 blocks out there at any given time. Performance gains have worked out so that most competitors are almost on a level playing field...most of it is aesthetics and personal preference.


Actually this is my next plan. I was just being curious about running (2) separate loops. I actually have a passive radiator setting in my basement waiting to be used. My plan is to do a single loop then add the ability to add an external passive radiator (on the fly) using a quick release connector so I can either use it with or without an external passive radiator (if I want without a passive rad I'll just have to add a small connecting tube to connect the outlet & inlet).
m
0
l
January 23, 2012 6:16:34 PM

I was just also curious about this (so called experts) grinding the bottom of the cpu block, etc. & using this testers & graphs that it works.

My test turned out differently. It barely able to change the temp at all.
m
0
l
a b à CPUs
a c 325 K Overclocking
January 23, 2012 6:18:30 PM

I don't want to hijack this thread for your project, but having a splitter like this wouldn't be that beneficial, especially on a passive radiator. You are much better off running fans on a rad, even if they are low speed. For a radiator to work at all, there has to be some kind of airflow...otherwise, to be 100% passive, you'd need to have a very large surface area rad to dissipate enough heat to make it a worthwhile addition to your loop. If you want to open a discussion on it, let's start up a new thread. :) 

Quote:
I was just also curious about this (so called experts) grinding the bottom of the cpu block, etc. & using this testers & graphs that it works.


Do you mean lapping? For lapping to be effective, you really need to lap both surfaces- the contact area of the block, and the CPU IHS.
m
0
l
January 23, 2012 6:32:54 PM

hijack away. i always wanted to lap but my cpu is still in warranty, speaking of warranty and going off topic, if i used the warranty, do i still get to use it until its period is over, or does that vary by company? example i sent out my cpu and mobo. so i got a new cpu and my warranty period still has 2 years, so next year, if my cpu dies, can i rma again?
m
0
l
a b à CPUs
a b K Overclocking
January 23, 2012 9:41:36 PM

They generally tell you, but in general you should be able to RMA a second time if you receive a defective product in return.
m
0
l
January 25, 2012 2:08:59 AM

so here is an update on my testing. i took the setup out the case. i took the block off the cpu and placed the stock heatsink. so i just have the bock on the gpu. with the stock heatsink the temps flux. between 42 and 50 idling. thats lower than what i was getting on water. the gpu started at 32C, ran MW3 for maybe 5 minutes on max setting. gpu constant at 37-38C.

on prime 95 ran for 10 minutes and cpu never passed 76C. on water it was nearing real close to 80C.

now as far as the gpu, i can live with 5C difference as on the giga board it was constant at 33C idle, but that was in a loop with the cpu. right now it is by itself in the loop. But like rubix said, maybe my board is giving different voltages that my other board. I get my giga board back tomorrow so i will post an update.
m
0
l
a b à CPUs
a c 325 K Overclocking
January 25, 2012 2:31:03 AM

Quote:
on prime 95 ran for 10 minutes and cpu never passed 76C. on water it was nearing real close to 80C.

Quote:
gpu started at 32C, ran MW3 for maybe 5 minutes on max setting. gpu constant at 37-38C.


If this is with the same board, just using the air cooler vs water block, you have real issues with something in the loop, but let's see what it is via process of elimination....

If your GPU is seeing load of 37-38C, this is very normal for load temps, even for 5 mins...if something were wrong, you'd see this go up as well. So, this tells me the loop is operating just fine...pump, rad, GPU block and minimal air in the loop, if any.

The issue comes down to how the CPU block is mounted. It isn't restriction or anything with reduced flows or no flow...this would impact the GPU temps, directly. I am assuming that the CPU block is still in the loop, but no on the CPU...even so, you didn't mention GPU temps being high, so bypassing that.

CPU block mount issue of some sort.

$5 on it.
m
0
l
January 25, 2012 4:25:58 AM

It could be a thermal paste problem. I've experienced thermal paste problem before. I was using this expensive Antec Nano Diamond something thermal paste. At first it's doing good. I got better temp. Then when I was doing something inside my PC, somehow when I turned on my PC later on the temp when so high (90s) that my PC sounded an alarm & then shut off. I have hard time figuring it out until I removed the CPU liquid cooling block. I found out that the thermal paste somehow dried up & just cracked. I replaced the thermal paste & then the temp went back to normal.
m
0
l
a b à CPUs
a b K Overclocking
January 25, 2012 1:21:41 PM

It could also depend on how you're applying the TIM. I usually just do a "plus sign" or an "asterisk"-type of shape, since those tend to get maximum coverage while eliminating the possibility of air bubbles.

The temperature differences are definitely interesting...
m
0
l
a b à CPUs
a c 325 K Overclocking
January 25, 2012 1:26:15 PM

Even with excess TIM, it should squeeze out what isn't needed if the block is making good contact. It just seems very odd that the CPU temps are high, but GPU temps are very good. This leads me to believe there is an issue with how the CPU block is mounted (which I guess also encompasses TIM application :)  )
m
0
l
January 26, 2012 4:39:32 PM

im using artic mx4 as my thermal paste. it is the same paste i use for my previous build. so i got my giga board back, hooked everything up exactly how i had it before is shipped out the board ad cpu. fired it up and gpu was at 38 which is fine. cpu was at about 42-45ish mark. ran it for a few hours, did some gaming, some video encoding and such. as of right now, 3 hours after putting it all together and using it, my cpu is jumping between low 50's to low 60's. this is affecting my gpu as it is now up to 50C since i have the cpu going to the gpu. im eally starting to think the chip intel sent me had a problem.

could the cpu be the problem? Rubix i apply TIM the same way as you. i even tried with a card and gave it a very very thin smooth layer. still the same. or could a block go bad? i opened the block again to make sure it was ok inside and it is. i hand tighten the thumb screws on the gtz with a very very small twist with a screwdriver.

IDK im thinking this cpu has an issue. one core is at about low 50 while another is up at around 60ish. this is all idle. im sure this is normal for the cores to jump though as one is more active than the other.

The cpu temp is affecting my gpu temp. Im really thinking this chip has an issue. again i remount the block 3 times to make sure it is on properly and the setup is the exact same before i shipped out the board and cpu.
m
0
l
January 26, 2012 4:43:18 PM

The i7 950 i had, i bought in dec of 2010. so i know the new one they sent me has to be from a newer and/or different batch. anyways. if this cpu is having issues with its temp, it does not shoot up to points intel would deem unsafe so im not sure if i could rma this chip. more so, i dont know their stance on liquid cooling their chips so im not sure that me stating my old chip was 33C on water and the new one i got is now at 50ish/60ishC would do me any good.
m
0
l
a b à CPUs
a c 325 K Overclocking
January 26, 2012 4:47:49 PM

I still think there is an issue with how the block gets mounted.

When the block is mounted and then you remove it, is the TIM spread evenly where the CPU IHS made contact?
m
0
l
January 26, 2012 4:58:34 PM

rubix_1011 said:
I still think there is an issue with how the block gets mounted.

When the block is mounted and then you remove it, is the TIM spread evenly where the CPU IHS made contact?




yes it is spread evenly. see the issue is, if it had no temp problem with my older cpu, hence i mounted it properly, how could i all of the sudden mount it improperly with this new chip. im using the same block, same bracket, same TIM, making me think it is the new cpu. This is very puzzling.
m
0
l
a b à CPUs
a c 325 K Overclocking
January 26, 2012 5:04:30 PM

Not all CPUs are the same.

Please confirm- with the new chip, you are seeing good temps with the air cooler, poor temps with water block.

-if so-

It isn't the chip, its how the waterblock is or isn't seating correctly. Your GPUs are still running cool with them in the loop, so it isn't flow or lack thereof. Take a razor blade or flatedge against both the CPU and then against the bottom of the block. This should tell you how flat, convex or concave each are. You could also reinstall the older CPU (if your board is the same socket) and test the old CPU temps vs new ones that way as well.

If you can post a pic of your loop, maybe there are other things at play here, but your GPU temps tell me that the loop is working as it should other than the CPU block making good contact with the CPU.
m
0
l
January 26, 2012 5:13:43 PM

I will try that today and come back with pics and results. I cant test with the older cpu as i sent it out to intel to get tested, and in turn the sent me a new(same chip) cpu. so if this cpu has a curve on it that would be the issue correct? the block is flat as i razor tested that last night. didnt cross my mind last night to razor test the cpu.
m
0
l
January 26, 2012 5:15:40 PM

oh to answer your first question the temps i see on water are the temps i see with the stock cooler. stock cooler has normal stock temps for such a small heatsink.
m
0
l
a b à CPUs
a c 325 K Overclocking
January 26, 2012 5:20:34 PM

And your CPU is at stock speeds, no overclocking correct? (Just trying to rule out variables for temps).
m
0
l
January 26, 2012 5:24:45 PM

yes the cpu is at stock temps.
m
0
l
January 26, 2012 5:30:09 PM

i mean stock speeds. sorry
m
0
l
a b à CPUs
a c 325 K Overclocking
January 26, 2012 5:32:06 PM

You shouldn't see that CPU at stock idle temps of ~60C with any cooler, including the boxed Intel cooler. So, if you are seeing this bad of temps regardless of the cooler being used...something is definitely up.

Thinking more...you shouldn't see this on any stock CPU with any cooler at idle, unless you're outside and its 120F in the desert.

Quote:
3 hours after putting it all together and using it, my cpu is jumping between low 50's to low 60's. this is affecting my gpu as it is now up to 50C since i have the cpu going to the gpu


This concerns me...the CPU temps shouldn't be getting warm enough to impact your GPU temps like this if the loop were functioning properly.

Just for clarity:

Can you state idle (stock speeds) of your CPU and your GPUs in the water loop

and

load temps of the CPU and GPUs in the loop?

You'd want to run some stress tests that don't require launching a graphics benchmark- so FurMark and Intel Burn test would do the trick. (I know there are concerns about how much FurMark can actually load certain cards, but for this, not as big a deal at this point).
m
0
l
January 26, 2012 6:10:27 PM

the cpu is stock at 3.06ghz (i7950) 52 lowest on one core, 58 highest on another core idle.
the radeon 6950 is at 47/48C right now with stock speeds of 800mhz and 1250 on the memory

on load, using intel burn test, cpu dances with 73/72 but during the 10 minutes test stood constant at 71/70. GPU only went to 52 from the water.
m
0
l
a b à CPUs
a c 325 K Overclocking
January 26, 2012 6:28:04 PM

These temps are all high, especially the idle temps...by at least 10C.

I wish I had asked this question sooner. Ok, I doubt its CPU block mount, its air or restriction in the loop. All components are showing idle temps higher than they should be, not just CPU. You have air in your loop and/or an issue with flow or restriction...has to be it.
m
0
l
January 26, 2012 7:43:03 PM

the cpu temp should have an effect on teh gpu temp though correct. the loop is from the cpu to the gpu. so if the cpu is sending hot liquid to the gpu, the gpu is hopeless to get lower temps. thats how i see it.

also, i should have mentioned this but its prolly not a big factor, is i have one of those koolance caps that bleeds air out a loop. its on my res. let me post some pics of before the rma's and after of my rig. images will help.
m
0
l
a b à CPUs
a c 325 K Overclocking
January 26, 2012 8:35:11 PM

Quote:
the cpu temp should have an effect on teh gpu temp though correct


Not really...if so, you have a really bad delta. The water should never be hot...it should at very most, be lukewarm even at full load runs for several hours. There should never be any 'hot' water at all.

Yeah, pics might help identify anything we might potentially see.
m
0
l
January 27, 2012 3:06:57 AM

Just to add my own 2cents, have you made sure you've got good airflow going through your rad? I remember reading about a few people that tried to do a push pull setup through their rad and had the fans set up in opposite directions so almost no air went through.. or I remember someone that put the radiator against something solid where air couldn't flow through it. Just trying to throw an idea out there :( 

Also, I doubt this is really your main problem.. but you mention that you used a card to apply your thermal paste.. and assuming I didn't misunderstand you, you can get better contact by just putting a small amount (grain of rice or pea size) in the center.. and let the pressure from mounting the heatsink/waterblock spread it for you. Pre-spreading the thermal compound will inevitably trap air bubbles in there, which will reduce performance.


video showing how thermal paste spreads
http://www.youtube.com/watch?v=EyXLu1Ms-q4

some tests comparing different application methods
http://www.hardwaresecrets.com/art [...] -1/1303/10

Good luck!
m
0
l
January 27, 2012 6:38:54 AM

gbkinum1 said:
Just to add my own 2cents, have you made sure you've got good airflow going through your rad? I remember reading about a few people that tried to do a push pull setup through their rad and had the fans set up in opposite directions so almost no air went through.. or I remember someone that put the radiator against something solid where air couldn't flow through it. Just trying to throw an idea out there :( 

Also, I doubt this is really your main problem.. but you mention that you used a card to apply your thermal paste.. and assuming I didn't misunderstand you, you can get better contact by just putting a small amount (grain of rice or pea size) in the center.. and let the pressure from mounting the heatsink/waterblock spread it for you. Pre-spreading the thermal compound will inevitably trap air bubbles in there, which will reduce performance.


video showing how thermal paste spreads
http://www.youtube.com/watch?v=EyXLu1Ms-q4

some tests comparing different application methods
http://www.hardwaresecrets.com/art [...] -1/1303/10

Good luck!



i pressure tested my rad and im getting good flow through it. i have 2 fans pushing air in it on the top, and two on the bottom pulling air off of it and the bottom two are hitting the gpu. like damn near kissing the gpu so its getting the air form those fan.

i was thinking about the push pull but seen the downfalls and gains and its not really worth it.

here is a link to the video of my rig on youtube. things look a mess as i was testing it. but that is my setup before the cpu and board rma. thats when it worked fine, temps where were they should be on water.
http://www.youtube.com/watch?v=evYg5tFgKaI

so an update. i figure i do each component in a loop by itself. i did the gpu earlier and it was fine. so i now have the cpu in the loop by itself and the temps are still too high for it on water. so learning towards rubix's idea of the mount, this block is old, screws seem worn, maybe i can try a new water block, or get all new mounting hardware for this block. i do remember now i had to cut the screws short cause they were hitting the back of the case.

I have tried applying paste everything which way that can be done other than throwing the damn tube at it. lol. even then i wont see a high hike from what i previously had. now i wish i had two different 950's to test. i got two boards and both boards show same results. i will work with the rad again but i see its down the the cpu area, be it the mount or cpu, but it has to be in that area.

m
0
l
January 27, 2012 7:01:19 AM

At 13 seconds in your video it sounds like you've got air in your loop (and not a small amount). Try physically shaking/tilting your case with the pump running to bleed your loop. It also looks like you only have half of your radiator getting any air flow through it. Maybe I just can't see it from the video, but it looks like the bottom half is obstructed and is unlikely to be doing much of anything.
m
0
l
a b à CPUs
a c 325 K Overclocking
January 27, 2012 2:33:02 PM

I think the :13 seconds thing might be airflow past the mic pickup on the camera...it sounds like wind, but could be wrong.

Even with some air in the loop, temps shouldn't be that bad...you'd have to really have almost 1/2 a rad full of air to get as bad as he is reporting...which is almost impossible unless the loop is that low on water. (Let's hope not).

If you have the CPU and GPU in the loop, does the GPU stay cool at load temps, regardless of the CPU temps?

aka...GPU nice and cool vs CPU running a bit warm?
m
0
l
January 27, 2012 7:34:58 PM

yes you hear is the wind from those fans. loud and fast they are. loop is full of water. i do tilt the rig to make sure i get as much air out as possible. like i said that koolance port helps with air release also. the gpu was was at 40 when powered on and at load hits only 48. its a very short trip from the cpu to gpu so i was assuming the water from the cpu was somewhat effecting the gpu. the water was very warm, warmer than it should be. so i have to take it all aprt and redo everything. now that the cpu is on the loop on its own, temps are still way up there for water cooling. i think i will buy a whole new mounting kit for my block and see if it is a mounting issue. but holding the board eye level i can see the mount making very good contact. it is also nice and even so it has good coverage across the cpu. i just turned on my pc right now and jumped on here.

temps right now are 38C on one core, 48C on another. if i were to load the cpu, as i have done these last few days, it hits up to the mid 50's. then it stays in the 50's for a long time. never really goes back down to the 40's.
m
0
l
a b à CPUs
a c 325 K Overclocking
January 27, 2012 8:02:31 PM

When I load my CPU on Intel Burn test, all 4 cores easily hit the mid 50's C...they also drop down to low 40's or so, depending on what ambient is at the time. Of course, my CPU is overclocked pretty heavily vs. stock, so this is why my temps are like this.
m
0
l
January 27, 2012 10:03:52 PM

those are nice load temps. but mine are just so high at stock speeds, at idle and on water. when i overclocked to 3.83, my temps were 33C on the cpu. i forgot to say that when they stay in the 50's, ill to the intel burn test again and they jump in the 60s. i just got done gaming from the time of my last post to this post. temps right now are 44 to 52 jumping around on the cores. let me burn test right now and see what happens.

this house is always at 68/70F so that is not a factor. im in the Chi so its cold. might just set the damn pc outside in an igloo with long cables. lol.

m
0
l
January 27, 2012 10:16:32 PM

So the gurgling sound I hear is just camera trickery? You don't hear it at all?
m
0
l
January 29, 2012 10:01:42 PM

gbkinum1 said:
So the gurgling sound I hear is just camera trickery? You don't hear it at all?



yes the gurgling sound you hear is that fan. it is pulling air off the rad into the case going out the back of it. camera got in the area of the fan wind.
m
0
l
!