Jump to content
Corsair Community

3D Rendering, Heat and Noisy Fans


Travel Matt

Recommended Posts

I have an H110iGTX Cooler in my system running an i7 5960X Chip.

 

I do a lot of 3D work which means rendering out frames that use a lot of computing power.

 

Up until a few weeks ago my H110iGTX would reach a comfortable 37/38 degrees whilst doing overnight renders but it has recently developed a heating issue.

 

Now, within minutes of setting off a render the H110iGTX temperature races to 40 degree's and continues to rise and making the fans come on at full blast and staying on until the render is complete, even on the simplest of scenes..

 

The noise makes it impossible to concentrate during the day whilst doing test renders or hard to sleep at night and I an concerned that this is a heating issue that will only worsen.

 

In the past I have blown the dust from the radiator grills on the H110iGTX which usually helps but this doesnt help any more.

 

I havent changed the H110iGTX settings which are H110iGTX in quite mode and the pump is in performance mode.

 

I suspect that another issue has developed which has caused this change in fan/cooling behaviour but any suggestions of how I can get back to the quieter running would be appreciated.

 

Cheers

Link to comment
Share on other sites

I suspect the H110iGTX pump is either failing or has failed, but from what you have posted this is nothing more than an inspired guess.

 

You should use http://forum.corsair.com/forums/profile.php?do=extra to add your full PC specs to you profile and if you also attached the CL 4.6.0.86 [Home] tab screen shot I could comment further.

 

See http://forum.corsair.com/forums/showthread.php?p=853900 for how to attach a file/screen image to a post.

Link to comment
Share on other sites

Red Ray

I have updated my system profile, however I am running CL 3.2.5695 here and I am a little reticant to update. I'm not the most technically minded in the event that the update throws up an error of some sort!

 

Is there something I can screen grab from CL 3.2.5695 for you to look at?

Link to comment
Share on other sites

We are looking for H110 GTX Temp (coolant temperature) data for cool idle and your rendering loads. You mentioned 37-38C on prior jobs. What is it now? Does the coolant temp go up and stay elevated?

 

Also keep in mind your base coolant temperature is affected by room and environmental factors. Some unusually warm US weather the last few days and a +10C change in room temp translates into a +10C change in coolant temp.

Link to comment
Share on other sites

Is there something I can screen grab from CL 3.2.5695 for you to look at?

 

I can't really remember what a CL3 screen looks like, but I guess it must have all the information, so post the one with all the sensor information.

 

I prefer to see all the temperatures, fan and pump speeds as quite often I spot more than from a list of some of numbers.

 

Is there are reason for you using CL3 rather than CL 4.6.0.86?

 

Also as c-attack said tell us the room/case temperature and also where the H110iGTX is mounted and the general airflow in the case. I guess you are also in the UK, so it's not that warm.

Link to comment
Share on other sites

Thanks fellas

 

Red Ray - CL3 was installed when I had the machine built and I'm unsure of the update process for v4 (do I have to reconfigure all my settings etc). I have read that v4 can have a few issues and I simply dont have a enough tech know how to navigate any potential update problems!

 

As as far as environment factors go, my home office has no heating. Even with the cold UK weather I can render with the window open and the system overheating issue persists.

 

Essentially the PC's environment has not changed which is why I noticed when the problem started.

 

I will get some grabs and upload.

 

Thanks chaps

Link to comment
Share on other sites

I know only too well what CL4 is like, but it's not as poor as CL3. If you do switch to CL4 you will need to setup the cooling configuration again so make a note of what it's currently set to.

 

Further if it is the H110iGTX pump that has failed I suspect you will get a different cooler when you RMA it that CL3 will not support and you will have to switch to CL4. I recommend the H110i.

 

That is "cold UK weather" in °C ?

Link to comment
Share on other sites

Chaps

 

So here are my screen grabs:

 

These are the values at idle:

 

Idle.thumb.JPG.f83a968967bdb25fd7806bcf1854982c.JPG

 

 

 

These are the values after rendering for 7 minutes:

 

Rendering.thumb.JPG.79fc162bb74eeda24276b7576ebf9604.JPG

 

 

 

You should know that the scene being rendered was not as heavy as some I have and the average overnight rendering session would be approx 10 hours.

 

Also, the H110iGTX cooler temp continued to rise up until I quit the render and I assume it would have continued to do so for some time.

Link to comment
Share on other sites

attachment.php?attachmentid=28382&d=1490635360

 

Looking at all the numbers I expect the H110iGTX pump has failed. Log a call via https://corsair.secure.force.com/home/home.jsp and ask for an RMA. Remember to upload the original invoice and you may wish to ask for an express RMA so you get the new cooler and return the failed one. The UK RMA location is Scan in Bolton which is where I guess the system came from.

 

I suspect they will not have a H110iGTX and will offer a different cooler. I advice you to ask for a H110i.

 

I now remember all the issues CL3 has with creating ghost devices. I assume you have 1 x GTX 960 rather than the 3 x GTX 960 that CL3 reports :(:.

Link to comment
Share on other sites

Thanks so much red ray.

 

Annoying result but so useful to have you run your eye over it.

 

Thinking back I did switch my machine on after the Xmas break and I heard a strange whirring noise that cleared after about 5-10 minutes. Perhaps this was the start of the pump fail?

 

Scan did indeed build my system. Its only just over a year old which is frustrating too.

Link to comment
Share on other sites

Thinking back I did switch my machine on after the Xmas break and I heard a strange whirring noise that cleared after about 5-10 minutes. Perhaps this was the start of the pump fail?

 

Yes, I suspect this was the beginning of the end :[pouts:.

 

Given Scan built the system you may have other RMA options and also wonder if you are happy to change the cooler yourself.

Link to comment
Share on other sites

I'm not confident to go poking around with cooling systems if I'm honest.

I have been having a similar dialogue with them as I have on this forum. They have been suggesting different fan profiles but they clearly havent been working.

 

I'll have to push them on a suspected pump fail I think and hope they come up trumps.

I do have a 3 year warranty on the build.

 

One other question - The grabs I posted did show that the pump had a variable rpm. Could this have been a false reading? If the pump has died shouldnt the rpm be zero?

Link to comment
Share on other sites

If it's not a pump failure I will be rather surprised, I am hoping c-attack will post what he suspects.

 

I guessed as much from your CL4 comments, but there nothing I can do unless you live near GU12.

 

It's quite common for the pump to report a speed when it's not pumping. If you feel the radiator does it feel like it' 50°C (assuming the H110iGTX is 50°C)? If not then it's a wonky pump for sure.

Link to comment
Share on other sites

I don't know if the pump has actually failed, but there does appear to be a flow problem. The 35C idle coolant temp is a bit high, but might be expected on higher fixed voltages with no c-states. The 12C rise in 7 min seems quick as well and my similar 280mm will only go up 5-7C on a 5930K@4.5 in that amount of time.

 

A secondary way to confirm is what happens when you stop the load. In a fully functional set-up, the coolant temp should start dropping as soon the load quits. You should drop the majority of the heat back to near start levels in a few minutes (when the load duration was short). If the coolant temp stays up in the 40's for a prolonged period of time, that would also indicate a problem.

 

Regardless, your system was working and now it is not. Absent a voltage or environmental change, it is no longer effective for your work. I would start the RMA process.

Link to comment
Share on other sites

Thanks for replying c-attack.

 

The heat does start to drop back as soon as I stop the render process and it takes a minute or two before the temp falls to 40c and then the fans spin down.

When the fans are going at full speed I would expect the rad to cool anyway (once the load is off of course) whether the pump is working properly or not.

 

Scan are suggesting I try and renew the thermal paste but its not something I'm confident doing and to be honest I feel that they should be a little more proactive with a machine that has a 3 year warranty.

14 months seems a bit quick to be renewing paste doesnt it?

 

But you are right c-attack, there is a distinct change in the machines behaviour and it needs resolving.

Link to comment
Share on other sites

If you do a lot of heavy work, I think changing the TIM every 6-12 months is a good idea and you should probably look to do it anyway. However, the paste affects the thermal conductivity between the CPU heat spreader and the cooler's cold plate. A loss of effectiveness would result in less heat conducted into the cooler (or at a slower rate) and higher CPU temperatures, but not necessarily a measured rise in coolant temperature. It takes a specific amount of heat (watts) to raise the volume of fluid inside the cooler by 9C (or whatever), less the amount of heat the radiator and fans can remove. Since the fans are running even harder and pegged at max, that rate is constant and the coolant temp is still rising. This suggests something else.

 

This really can only be one of three things: 1) environmental change by the same amount; 2) the flow problems discussed above; 3) a change in the nature of the load. We really haven't discussed the last one and I suppose it would be prudent to do so.

 

While generally similar is nature, I suppose not all renders are exactly the same there may also be the possibility of a program change. Any BIOS or frequency/voltage changes lately? Other changes to memory frequency? Are you at the stock frequency?

 

Run a known stress test for comparison to renders. Intel XTU is a cupcake of stress test, but it loads in a smooth even way. That makes paste jobs and other problems easy to spot when cores jump out of line. Don't use any of the software features to change your settings. We just want the stress test part of the application. At the bottom of the page is a line graph. Before you begin, click the blue wrench (top right corner graph) to change the display settings. Select your 8 cores individually to display, like the picture below. You don't need a lot else, other than load and frequency. 10 or 15 minutes is just fine. Take note of your starting H110 GTX Temp before and after the test. Then again 5-10 minutes after the test stops. I would like to be able to compare the coolant delta between your renders and this program where I have a good idea of the wattage involved.

 

34-png.69066

Link to comment
Share on other sites

Thanks again C-attack

Other than regular Windows updates I cant think of anything that has changed on the system. I try and avoid doing so wherever possible coming from the 'if it isnt broken.....' school of thinking.

 

Environment is the same. In fact I have the heating off and the window open at times which at this time of year makes the room fairly cold.

 

My 3D app is the same and I have notice the same heating issue when converting videoformats in an app called Handbrake. This in the past has caused to sytem to warm up during processing but again is now getting very hot like the rendering.

 

I'll give the stress test a go and see what happens.

Link to comment
Share on other sites

Thanks again C-attack

Other than regular Windows updates I cant think of anything that has changed on the system.

 

Actually, that is something I am wondering about. I've already had a strange conversation with a trio of HW-E owners this week about bizarre core activity after the Tuesday Windows update. All of us experienced a prolonged steady load from the SVC Host (number) where the OS was clearly trying to do something substantial and using a lot of resources. I uninstalled my update and then re-downloaded to address it. This went on for a lengthy period of time with the others.

 

If you have a problem with the cooler, it should also be present at idle. The 35C seems a bit warm for your environment and in the default BIOS conditions, your C-states, speedstep, and adaptive voltage are all on. Everything should step down when no doing something. It might be a good idea to open your task manager (right click menu bar) and sort the processes by CPU usage. Look for anything using a good chunk and it would also be present in the CPU usage graph.

Link to comment
Share on other sites

So here are the test results:

 

 

Idle Pre Stress Test:

1503039360_PreStressTest.thumb.JPG.f7b9a802d788a7bebacd496ed136c0ed.JPG

 

8 Minutes after Test Finished:

25023305_8MinsAfterStressTest.thumb.JPG.5247b4c4e2eebfeb1b37bd607216738d.JPG

 

Intel ETU:

1086708275_IntelETU.thumb.JPG.bd72935c964a2c1c33a93ccfb78f6d9f.JPG

 

I would say that the pattern of heat increase was comparable to when I'm rendering, both the rate of increase and fall off after the process.

 

I'll take a look at those CPU processes now

Link to comment
Share on other sites

What did the coolant (H110i GTX Temp) get up to during the XTU test? The "8 min after" has the coolant right back where you started. That probably means there is not a flow problem or at least a meaningful one. However, your CPU core temps are high for XTU. I would like to know the load coolant temp to get an exact delta, but for comparison my 5930K set to 4.5@1.275 and 4.0 on the cache will pull 185-205W on XTU. That is more than the 140 TDP your 8 core should use. I won't break 62C and my coolant to CPU temp delta is lower than yours. That should not be the case with my much higher voltage and power draw.

 

I do think you should change the TIM based on the coolant to CPU temp delta. That does not fully explain the rise in coolant temp, but perhaps tackle one thing at a time. Most of these pastes are all the same and there is no meaningful difference. Generally, you are looking for ease of application (not super viscous). I generally recommend Noctua NT-H1, but there are plenty of other good options. For the 6-10 core CPUs, I use Grizzly Kryonaut, which is now getting a lot of publicity. It is expensive, but they now sell a 3.5 oz mini that should last you 6+ changes and likely the life of this system. I think it is worth the extra couple 2-3 pounds for a one time purchase.

 

Once you get that completed, take another look. Don't worry about the 3rd and 5th cores popping off +10C on the others. That is an extremely common circumstance on HW-E and related to the physical location of those cores on the CPU.

Link to comment
Share on other sites

Thanks again for such a considered response C-attack.

 

At its highest the coolant temp was 46c during the test which is consistant with the rendering that I was doing. It did steadily climb and I suspect it would continue to do so it I had done an overnight render causing the fans to roar away for the duration.

Generally in the past the fans might spin right up for a couple of minutes then down again suggesting that the temp peaked around 40-42c and that was on a warm summer night where the environment was hot to begin with.

Link to comment
Share on other sites

Archived

This topic is now archived and is closed to further replies.

×
×
  • Create New...