• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

Custom VRM/VCORE and Motherboard Cooling is it worth it?

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

givmedew

Member
Joined
Sep 3, 2010
Location
Chicago
PICTURES AT BOTTOM

I would like to test how big of a difference actively cooling the VRM or other has on the CPU and VRM temperatures. I would also like to show an easy method of mounting the fans and what size fans work the best. Before I go any further I want to say that I have already done this and the results where fantastic. However I didn't really do it for the benefit of my fellow over clockers. I did it for myself in this I did not record enough data to prove as much as I would have liked. Also that test was directly pointed at the Sabertooth 990FX, however this time I would like to open it up to other motherboards.

Below is a link to the original data collected for both air and water cooled CPU.

A few months ago I posted just a little bit of data from a couple things I played with.

A little info on the old data:

What I played with was:

Fans on the back side of the motherboard pointed between the VRM heat sink bracket and the back of the socket. This requires a case with a hole cut out for access to the cpu cooler bracket and for permanent installation would require you cutting into your side panel. The expected results where little to no cooling. The results where noticeable and valuable cooling to the VRM and in turn the CPU.

Fans pointed directly at the VRM heat sinks on top of the motherboard. I tested both 40MM and 80MM fans. I also tested 1 or 2 fans. I found that although obviously 2 are better that just 1 single 40MM fan produced a significant amount of cooling. Temperatures where taken from the Sabertooth 990FX extra temp sensors. It has sensors all over including 2 VRM, USB, PCIE, and more.

The tests were performed for me not an audience and they proved to me without a doubt that actively cooling the VRM even just slightly reduced VRM and CPU temps by more than just the margin of error.

What made this even more interesting to me was that this change was noticeable with both air and water cooling. Plus any drop in CPU temp is welcome to overclockers.

The data I did save was just a few screen shots that I gathered for myself. I did have a T-Balancer BigNG recording 4 other temps including 1 probe under the heatsink backplate on one bank and 1 probe under the heatsink backplate on the other bank I did not screen shot any of it because although the temps reported were different the drop in temp was about the same percentage so I blame that on the exact location of the probes.

What I want to achieve this time around:

Bobnova pointed out to me via thread responses and PMs that the ASUS thermometers could be incorrect if air was passing over them. Since there is no way for me at least to find out exactly where these probes are I have to agree with him that the data def could be skewed. If someone knows the exact location of these probes and wether or not they are completely enclosed or if they are just laying about freely that info would be great.

So this time around I would like to record for evidence the data from non motherboard probes placed in ways that they would not be affected by air moving over them (in other words directly touching components or thermal tape on one side and a heat-sink on the other side).

I would like to show which size of fan is best for the backside and which is best for the front side and would like to weight it against total performance vs fans rated CFM at 100% on one graph and then performance at a weighted dB level on another graph. I believe at this moment that a 40mm to 60mm is best for above and that I would choose 40 over 80 and that 60-120 is the preferred choice for behind with 80mm being my exact choice. Definitively figuring this out may have little importance to most people and it is going to be the thing that takes the most time to record so I am not certain that this will happen.

Finally I would like to show an easy way to mount 1 medium to large fan or 2 small fans so that people know exactly what to get to get similar results. Depending on the results I might test the differences between a full board block and doing this as I have a full board block on another mobo and I have always questioned wether adding heat to the loop by something I could have cooled with air is worth it. Since I have noticed any cooling to the VRMs results in lower temps on the CPU it can't hurt much to add the VRM heat to the loop but who knows for sure.

The previous thread only got 1800 visits but I got a few thank you's in that thread as well as a couple PM's. So if it helps even just a few people it will be worth it and if it is more definitive and leaves less room for question I can sleep tight knowing I didn't waste anyones time or money.

Bobnova said he would be willing to lend some ideas so Bob if you see this and got anything to add as far as how the testing is done to keep it as definitive as possible please chime in.

I'd like to start taking some measurements Nov 6th which will give me 2 days to get any input from bobnova or other people. It may be possible that tonight I post up the first test results which is simply do the fans being in the on position vs the off position noticeably affect the temperature of the cpu as reported by the cores. This going to be hard to do if it is only a few degrees but would obviously show that it does or does not help on an overclocked water cooled intel. I'd say it is certain it helps the AMD but not sure of the intel yet. One thing that may be an issue is that I doubt the VRM and the CPU are anywhere near as hot as they were on the AMD 6-Core.

Wish me luck!

Here are some shots of the old setup and the new setup.
960t vrm cooling.jpg

00003387.JPG
 
RESERVED

db.JPG

I plan to take my sound measurements from here. I shut that fan completely off so that it would not be blowing air past the meter. The other 2 fans are running at the absolute lowest speed possible which is the speed they will stay for all the tests.

I am going to try to post any differences in sound by percentage to make it easier to understand. As dB can not be used to relay any truly useable information without making every person who reads it plug it through a difficult formula. The only thing that I can really gather from dB ratings just by looking at them is that 10dB is perceived as 2 times as loud and the 3dB is the smallest increment the average person can discern.

Some of the info I will use to figure it out is in the following link. Also there is someone on this forum that knows the equations and hopefully he will help if I need it.

Here is a link to about the easiest explanation I could find on the dB subject on the internet.

As of right now I have run preliminary tests on a Asus P8Z68-V Gen3 w/ a 2500K @1.288v there is not enough change to be outside the margin of error. At 1.4v and 7-10c hotter the fans still did not give any measurable results. Upon touching the heat sinks in the case they where not even hot where as on my sabertooth running 1.4v w/ 6 cores was so hot it would cause a blister if you touched it too long. The support bracket on the back of the motherboard however is too hot to touch.

If I can not get any helpful changes with any of these quick tests I won't do the more in depth tests for obvious reason. So far I feel that unless you are an extreme extreme overclocker I can not see the point of a full board water block on any boards with this socket. I think all it would do is increase the temps of your water therefore increasing the temps of everything else in your loop. Although they do look awesome!

Next test will be back of the motherboard followed by a few top of the VRM again but this time with 2 different fans. I found that these fans do not seem to be actually blowing much any anything out of them. They are not the fans I used for the AMD board they are ones that a purchased so I could get the RPM wire which the other fans did not.
 
Last edited:
Back side of the motherboard could be a nifty little project; cut out an area of the side panel for a low profile fan, mount the fan and cover it with a dust filter.

good luck with your tests!
 
At 1.288v on a 2500k I can say flat out it is not worth it to cool the topside of the VRM heatsinks. When I am all done I will be removing the fans. At 1.288 with the fans at 100% there was only a 2-3c variance in the core temps. With the constant up and down nature of the core temps I would call this pretty close to the margin of error.

The computer will be much warmer once I increase the core voltage. We will see after that. I will not go into the extreme OC voltages.

EDIT UPDATE 11:57pm

At 1.4v the processor gets much warmer but even with the fans on again there is not enough change to measure. One note though is that the heatsink on the VRM is not so hot that it hurts to touch it. On the AMD rig they are too hot to touch w 1.4v and 6 cores.

I may not get around to doing any of the full testing because so far all the preliminary tests are showing no gains.
 
Last edited:
On my 990XA-UD3 PD rig, setting a fan to blow directly on the two head spreaders surrounding the CPU lowered socket temps by 4c (Which would go back up if I jammed the fan within about 3 seconds)-- This is at a pretty high OC for a 8320. The NB heat sink gets quite alarmingly hot on this MB when a chip is OC'd.. Last time I thermal gunned it, I got a reading of 80c. This reads as 68 right after jamming the fan and taking a measure.

I'm also looking at getting air drawn away from the back of the board to lower temps a bit more :D
 
Where did you find that slotted metal bracket stuff, and how did you mount it? That stuff looks fantastic!
 
On my 990XA-UD3 PD rig, setting a fan to blow directly on the two head spreaders surrounding the CPU lowered socket temps by 4c (Which would go back up if I jammed the fan within about 3 seconds)-- This is at a pretty high OC for a 8320. The NB heat sink gets quite alarmingly hot on this MB when a chip is OC'd.. Last time I thermal gunned it, I got a reading of 80c. This reads as 68 right after jamming the fan and taking a measure.

I'm also looking at getting air drawn away from the back of the board to lower temps a bit more :D

I am quite certain that with a hi OC on a 990 board that you benifit greatly from cooling the VRM or even the back of the board (VRM BACKPLATE). Bobnova has made a very good point however. If any of those thermometers are exposed to any air at all then their readings could be suspect to the cooler air being blown around them. Now I had thermometers wedged in between the back plate and thermal tape and they where not exposed to air and on top of that they where on the opposite side of the board. I def saw the temps go down on the VRM but because I was using an unlocked 960T I can not verify that the chips temps went down. The socket temp went down but where is the thermometer for the socket temp? If air can go past it from up top then it could have a suspect reading.

I am fairly certain on the 990s with big OCs or x6 procs that it makes a big difference but I would like to hear from someone else that they verified that the chips temp went down not just the socket temp!

One other thing I would like to point out. If those thermometers are in fact exposed to air they are only going to be exposed on one side of the board top or bottom and I got positive results regardless of wether the fan blew on the top (heatsink) or bottom (heatsink bracket).

What sucks is I do not have the Sabertooth Board and Proc anymore. So I cant really go back and do any more thorough testing.

What I do know is pretty or not cooling the P8Z68-V from front or back doesn't do a thing. At the same time though it is also a good 20c cooler than than the Sabertooth 990FX VRMs where. I can touch the heatsink on the intel where as on the Sabertooth my finger would have burst into flame 2 inches from the heat sink LOL.


Where did you find that slotted metal bracket stuff, and how did you mount it? That stuff looks fantastic!

Microcenter it is a magnetic fan bracket or something like that. The magnets are bogus total junk. So I removed them then used the bolts and nuts that came with it to mount it through the holes in the mesh for the exhaust port and then put the fan back over the exhaust. It's nice and tight.
 
Last edited:
I am quite certain that with a hi OC on a 990 board that you benifit greatly from cooling the VRM or even the back of the board (VRM BACKPLATE). Bobnova has made a very good point however. If any of those thermometers are exposed to any air at all then their readings could be suspect to the cooler air being blown around them. Now I had thermometers wedged in between the back plate and thermal tape and they where not exposed to air and on top of that they where on the opposite side of the board. I def saw the temps go down on the VRM but because I was using an unlocked 960T I can not verify that the chips temps went down. The socket temp went down but where is the thermometer for the socket temp? If air can go past it from up top then it could have a suspect reading.

I am fairly certain on the 990s with big OCs or x6 procs that it makes a big difference but I would like to hear from someone else that they verified that the chips temp went down not just the socket temp!

I haven't seen a substantial change in the chip temps by doing this, but I use a thermal gun for surface temp readings when blowing air on these areas-- I know readings can get messed up when air's blowing on the sensor :p

I feel like getting rid of any heat in that area makes a diff. but man do those little heat sinks get grossly hot without a fan (I was actually afraid of frying something on the board.. 80c??)
 
I haven't seen a substantial change in the chip temps by doing this, but I use a thermal gun for surface temp readings when blowing air on these areas-- I know readings can get messed up when air's blowing on the sensor :p

I feel like getting rid of any heat in that area makes a diff. but man do those little heat sinks get grossly hot without a fan (I was actually afraid of frying something on the board.. 80c??)

In my book 4-8c is a substantial reduction in temperature on the CPU from simply blowing a little air on the VRM and I do mean a little. I ran only a single 40mm fan from a fried xbox 360 cooler and it was enough air to significantly drop the socket temp. The VRMs dropped even more of course.

Now on the Sabertooth I had done everything possible to keep the temps down on the VRM including the fact that I was up to 4GHz stable at 1.392v but I ran at 3.9GHz just to be safe. So my temps without a fan where usually around 70c. However w/ a 555BE rocking 4GHz I did get the VRM temp up to 80c they are supposed to be just fine up to 90c but it is kinda scary. With water cooling on the 960T I only saw VRM temps in the mid low 50s without extra cooling mid high 40s with extra cooling.
 
In my book 4-8c is a substantial reduction in temperature on the CPU from simply blowing a little air on the VRM and I do mean a little. I ran only a single 40mm fan from a fried xbox 360 cooler and it was enough air to significantly drop the socket temp. The VRMs dropped even more of course.

Haha, yeah, you're right! I keep thinking the numbers small (4-8 is pretty darn substantial when I think about it:D), but I'm probably getting something similar overall (The larger drop on that heatsink also put my mind at ease)
 
Yeh the reason it makes since that the CPU temp drops if the VRM is that high is because they are basically directly connected to one another via copper.

I am just a little surprised I can't get any measurable results on the Intel even if I jack up the vcore. Thing is I am not willing to try over 1.4v, I know I know what a wus but seriously I don't have the money to replace it.
 
I'mn not, really.
71w TDP chip (plus 24w of IGPU) vs 125w chip. The AMD bits are working a bit harder :D
Plus there is the question of what kind of MOSFETs are being used, most 1155 boards use Driver MOSFETs that are more efficient than their discreet counterparts.

Are the two rigs in the same model case? The Intel bits may have more airflow due to case fans too.
 
Yeh the reason it makes since that the CPU temp drops if the VRM is that high is because they are basically directly connected to one another via copper.

I am just a little surprised I can't get any measurable results on the Intel even if I jack up the vcore. Thing is I am not willing to try over 1.4v, I know I know what a wus but seriously I don't have the money to replace it.

Aye, I'm the same way-- 1.390 is the ceiling for me, even though people say they can go higher with no problem.

I think intels have a lot less leakage than AMD processors-- might be why lowering VRM temps isn't as dramatic on intel rigs. God the new AMd procs absolutely melt the socket T_T
 
I'mn not, really.
71w TDP chip (plus 24w of IGPU) vs 125w chip. The AMD bits are working a bit harder :D
Plus there is the question of what kind of MOSFETs are being used, most 1155 boards use Driver MOSFETs that are more efficient than their discreet counterparts.

Are the two rigs in the same model case? The Intel bits may have more airflow due to case fans too.

The AMD may have had slightly better airflow than the Intel. The intel only has 3 120mm excaliburs on top pushing through a radiator and into the case. One of the 3 fans turns off and the other 2 are around 600RPM when the system is cool. When I ran those tests there was a 120MM exhaust fan. I took that out though along with the 2 VRM fans because they made no real difference.

In the AMD though it was awesome how much of a difference it made.

I almost wish I had my am3 back.
 
Back