Is my H100i GTX faulty?

[notreadyyet]

Does the H70 backplate work with the H100i GTX?

 

I ask because I recently transferred everything to a new rig. Before, with my H70 core in an Antec Lanboy Air, I was getting 18-27C on idle (push/pull). Now, I'm getting 23-35C on idle in my Corsair 540 Air (pull only) , and there are times where it shoots up to 42C (very briefly). Otherwise, temps always float around in the 20's and 30's. I feel like I should be getting way better temps on my H100i GTX, compared to my H70. What gives?

 

Specs:

 

I7 2600k @ 4.7ghz (1.352volts)

ASRock z68 pro gen 3 fatal1ty

Mushkin ddr3 2133mhz (4gb x 4)

Msi 980 Ti

 

Edit: Also, I'm using two Sycthe Gentletyphoon fans (1850rpms) in pull. These fans and the pump itself are plugged into 12v power cables which are wired into my PSU (molex).

[c-attack]

You've got a lot of differences between the two set-ups -- case, cooler, fans (and probably fan speed), and perhaps radiator mounting location as well. Your numbers with the GTX sound about right to me for a chip with 1.35v running through it. I'm assuming you have the same level of C-states on both, but to have anything idle at 18-22C you need a room no warmer than 15-17C. If it is 15C and your idle temps are 30-ish, than there are some things to look at.

 

However, a short duration stress test will be a much better indicator of how the system is performing. You can use anything you want, but the free AIDA and OCCT programs as well as Intel XTU will let you see line graphs of individual core temps that make behavior easier to see, including unusual spikes at idle. A 5 minute run is enough for a basic shake down. If for some reason you get an immediate high temperature spike, abort and then take a look at the block contact.

 

Is the GTX in the roof? Or the front intake? Where was the H70 in the old case? As an intake?

 

On a side note, how recent is the 980 Ti. It has the potential to alter the balance all on it's on.

[notreadyyet]

You've got a lot of differences between the two set-ups -- case, cooler, fans (and probably fan speed), and perhaps radiator mounting location as well. Your numbers with the GTX sound about right to me for a chip with 1.35v running through it. I'm assuming you have the same level of C-states on both, but to have anything idle at 18-22C you need a room no warmer than 15-17C. If it is 15C and your idle temps are 30-ish, than there are some things to look at.

 

However, a short duration stress test will be a much better indicator of how the system is performing. You can use anything you want, but the free AIDA and OCCT programs as well as Intel XTU will let you see line graphs of individual core temps that make behavior easier to see, including unusual spikes at idle. A 5 minute run is enough for a basic shake down. If for some reason you get an immediate high temperature spike, abort and then take a look at the block contact.

 

Is the GTX in the roof? Or the front intake? Where was the H70 in the old case? As an intake?

 

On a side note, how recent is the 980 Ti. It has the potential to alter the balance all on it's on.

 

All the fans that were in my Lanboy Air are now in my Air 540. Ambient room temp is 22.7C

 

My Lanboy Air had two 120mm intake fans in the front, two 120mm intake fans on the side, and the top rear had the H70 Core with two Scythe Gentle Typhoons (1850 rpm -push/pull) as exhaust.

 

My 540 Air has three 120mm intake fans in the front, one 120mm exhaust fan in the top rear, and the H100i GTX's rad is mounted on the top of the case with two Scythe Gentle Typhoons (pull) as exhaust.

 

I've attached my results for the Intel XTU along with RealTemp.

 

http://i.imgur.com/bRcFq5t.png

[c-attack]

So you were able to get those low idle temps with the H70 in the rear exhaust above the 980 Ti? If everything is the same, that's extraordinarily good, but don't underestimate the impact of room and case ambient temperature on idle temps.

 

At the present, I don't see anything wrong with a 60C max for those settings. You do have a 10C difference between cores, but that could be internal to the chip (nothing you can do) or it could be TIM. Did you use the pre-applied paste? If so, it's not likely the paste is the issue, but I suppose at some point you will need to take it off and re-apply.

 

The screen shot looks to be just a second or two after the test is complete, and your instant CPU temps have already dropped 20C off the high mark. This is another sign the cooler is working. Presumably they dropped another 10C or so in the next few seconds. If it stayed at the 40C mark for a little bit, then that would suggest the fans were running very slow during the test allowing the water to heat up more than normal. Probably not since it sounds like you are running them at 12v all the time. If true, more likely the pump is not pushing as fast as it should if the temperature remained high right after that short test. It sounds like you are not using LINK and you don't have to. However, it may be difficult to get a pump reading in the current set-up. With out any other instruction, it should be running at the default setting which I thought was performance mode. Pump speed should be a non-issue.

 

I don't see anything wrong with your results, but the final word is better left to someone who has more experience with a 2600k. As for the idle temp differences, if your room temp in 22-23C, I can't see how you would be able to maintain anything lower than 26-27C, and that would be with the fans churning. 4-9C over ambient is a general range, with a lot of other settings have a small say.

 

Trying to track down the pump speed might be a pain and would likely require you to temporarily change the configuration. I am not sure that's worth the hassle until there is a need. I am a little more curious about the 10C difference between cores. Did this exist before on the Lanboy set up? How do the load temps compare to before (just generally)?

[notreadyyet]

So you were able to get those low idle temps with the H70 in the rear exhaust above the 980 Ti? If everything is the same, that's extraordinarily good, but don't underestimate the impact of room and case ambient temperature on idle temps.

 

At the present, I don't see anything wrong with a 60C max for those settings. You do have a 10C difference between cores, but that could be internal to the chip (nothing you can do) or it could be TIM. Did you use the pre-applied paste? If so, it's not likely the paste is the issue, but I suppose at some point you will need to take it off and re-apply.

 

The screen shot looks to be just a second or two after the test is complete, and your instant CPU temps have already dropped 20C off the high mark. This is another sign the cooler is working. Presumably they dropped another 10C or so in the next few seconds. If it stayed at the 40C mark for a little bit, then that would suggest the fans were running very slow during the test allowing the water to heat up more than normal. Probably not since it sounds like you are running them at 12v all the time. If true, more likely the pump is not pushing as fast as it should if the temperature remained high right after that short test. It sounds like you are not using LINK and you don't have to. However, it may be difficult to get a pump reading in the current set-up. With out any other instruction, it should be running at the default setting which I thought was performance mode. Pump speed should be a non-issue.

 

I don't see anything wrong with your results, but the final word is better left to someone who has more experience with a 2600k. As for the idle temp differences, if your room temp in 22-23C, I can't see how you would be able to maintain anything lower than 26-27C, and that would be with the fans churning. 4-9C over ambient is a general range, with a lot of other settings have a small say.

 

Trying to track down the pump speed might be a pain and would likely require you to temporarily change the configuration. I am not sure that's worth the hassle until there is a need. I am a little more curious about the 10C difference between cores. Did this exist before on the Lanboy set up? How do the load temps compare to before (just generally)?

 

I used the pre applied TIM that came with the H100i GTX. I've used Corsair Link to change the LED color and to confirm that the pump is running max in performance mode, then I uninstalled it.

 

The 10c temp difference was not there before I put on the H100. When I had the H70 on there the temp difference was only about 5 degrees. I'm not sure why there is that much of a difference. I cleaned the chip with 90% rubbing alcohol and a good microfiber cloth. I let it dry, then as steady as I could, mounted the H100i to the chip using the cross threading method. I do not recall moving it once I stuck it on the 2600k.

 

The idea of removing it and applying paste myself is terrifying. I never did it before. Every heatsink I've had always had the TIM pre applied.

 

 

EDIT

 

Here are the temps while my computer was just sitting idle on the desktop for 15 minutes. I wonder why the temps spike to 46 and then settle in the 20's? Maybe the pump has to get going?

 

http://i.imgur.com/Uz5AkDZ.png

 

 

I ran RealTemp's Sensor Test, not sure how to interpret it. Maybe it'll help shed light on the situation.

 

http://i.imgur.com/t0UZPFC.png

 

 

Thanks!

[lucascaju]

I am new here but I have problems with my H100i GTX and I need some help. The fans speed control don't work properly. With corsair link in performance mode or quite mode the fans work fine in idle, but after some stress activity, like games or stress softwares the fans go to maximum after some time and stay there for quite some time, even if I come back to idle again. The custom curve option in CL simple doesn't work for me because go to 100% don't matter the curve points that I put. The fixed mode works fine. I checked cables and everything but with no results. If I use other softwares that can read the cooler, like SIV or HwInfo with corsair link off the fans go to fixed in the lowest rpm possible. Everything is just too weird to my taste.

My cpu temps in idle are 38-40c (ambient 29c). In gaming 55c-60c. In OCCT after 5 minutes 68c 71c. So the temps I think it's fine considering my ambient temperature.

[c-attack]

I guess what I am getting at is with a 22-23C room temperature, those idle temps of 26-27 are about the best possible result. Manipulating your idle temperatures is difficult. Room temperature + current vcore load is going to have all the say. Fans almost none. Pump flow, yes -- but we've checked that one off. The load test suggests the cooler is working fine, but you still have concerns. Something has to be different now, whether it is room/case temperature or what Windows is doing. You can change some settings in XTU to allow you to see a line graph of individual core temps. You can also do the same thing in OCCT or the trial version of AIDA64. This will allow to easily see whether it is one core popping off and doing a simple task or all 4 cores jumping together. At the desktop, there shouldn't be a lot of simultaneous core jumping.

 

Make sure you haven't left High Performance as your Windows Power state after benchmarking. Take a look at your vcore at idle to make sure it is dropping off into .75-.90 range (or whatever it should be for your settings). It sounds like you have done your homework and I don't necessarily suspect a TIM problem, but the individual core activity makes that easier to see.

[c-attack]

I am new here but I have problems with my H100i GTX and I need some help. The fans speed control don't work properly. With corsair link in performance mode or quite mode the fans work fine in idle, but after some stress activity, like games or stress softwares the fans go to maximum after some time and stay there for quite some time, even if I come back to idle again. The custom curve option in CL simple doesn't work for me because go to 100% don't matter the curve points that I put. The fixed mode works fine. I checked cables and everything but with no results. If I use other softwares that can read the cooler, like SIV or HwInfo with corsair link off the fans go to fixed in the lowest rpm possible. Everything is just too weird to my taste.

My cpu temps in idle are 38-40c (ambient 29c). In gaming 55c-60c. In OCCT after 5 minutes 68c 71c. So the temps I think it's fine considering my ambient temperature.

 

Lucas, go to the groups tab in LINK and make sure you have moved the the radiator fans dial(s) into the H100i GTX Cooler group (this is your water temperature). The fans must run from water temperature. You can put the pump in that group as well. If you have tried to manually set a curve for CPU temperature, that would explain the results. Water temp takes some getting used to. It will be different for each person. Minimum is room temp + 3-4C. Maximum can be anything from the low 30's up to 50C, depending upon CPU heat and any other system waste heat, particularly GPUs. See if this helps. If not, copy your original into a new post and we'll look at other things. You have a very different issue than the original poster.

[notreadyyet]

I guess what I am getting at is with a 22-23C room temperature, those idle temps of 26-27 are about the best possible result. Manipulating your idle temperatures is difficult. Room temperature + current vcore load is going to have all the say. Fans almost none. Pump flow, yes -- but we've checked that one off. The load test suggests the cooler is working fine, but you still have concerns. Something has to be different now, whether it is room/case temperature or what Windows is doing. You change some settings in XTU to allow you to see a line graph of individual core temps. You can also do the same thing in OCCT or the trial version of AIDA64. This will allow to easily see whether it is one core popping off and doing a simple task or all 4 cores jumping together. At the desktop, there shouldn't be a lot of simultaneous core jumping.

 

Make sure you haven't left High Performance as your Windows Power state after benchmarking. Take a look at your vcore at idle to make sure it is dropping off into .75-.90 range (or whatever it should be for your settings). It sounds like you have done your homework and I don't necessarily suspect a TIM problem, but the individual core activity makes that easier to see.

 

 

I appreciate your help. Thank you very much!

 

The only thing I can think of is the case. My LanBoy Air was entirely mesh, with two fans directly cooling the video card, two front intake fans, 1 top intake fan, and the H70 core in exhaust (2 fans - push/pull). I must have taken the temps it gave me for granted. At the end of the day I am still within my limits of CPU and GPU (30 degrees of headroom). So I guess this is something I'll just have to get used to.

[c-attack]

The case shouldn't have much of an effect (if any) at idle. Load is another story, but that seems be OK. Environmental differences are still the most likely cause, but I would go ahead and look at what your cores are doing if you are getting spikes.

 

http://www.mediafire.com/view/xqlibirclilcn7h/Performance_Mode_and_C-states_enabled.png

[notreadyyet]

The case shouldn't have much of an effect (if any) at idle. Load is another story, but that seems be OK. Environmental differences are still the most likely cause, but I would go ahead and look at what your cores are doing if you are getting spikes.

 

http://www.mediafire.com/view/xqlibirclilcn7h/Performance_Mode_and_C-states_enabled.png

 

What program is that?

[c-attack]

AIDA64. The trial version has some limitations, but will let you see the basic details. I have found the full version to be a very effective monitoring tool. It's one of the few that lets you see individual core activity.

[lucascaju]

Lucas, go to the groups tab in LINK and make sure you have moved the the radiator fans dial(s) into the H100i GTX Cooler group (this is your water temperature). The fans must run from water temperature. You can put the pump in that group as well. If you have tried to manually set a curve for CPU temperature, that would explain the results. Water temp takes some getting used to. It will be different for each person. Minimum is room temp + 3-4C. Maximum can be anything from the low 30's up to 50C, depending upon CPU heat and any other system waste heat, particularly GPUs. See if this helps. If not, copy your original into a new post and we'll look at other things. You have a very different issue than the original poster.

 

Thanks but i move the pump and the fans to the cooler group with no results. I will create i new Thread.