Is the CPU Throttling?

[Draken21]

Whenever I play demanding games like Arma 3 or PUGB and stress the system every 1-2 mins I am getting temp frame drop and the following graph on HWMonitor.

Anyone recognises it? The CPU is a 8320 and the GPU a Sapphire Tri-x R9 290. The bottom line is the CPU dropping from 59c to around 45c and then up again.

Just changed the cooler to a H45 from a 212 as I got an offer I couldn't refuse on the H45.

Temps are hovering below 60c for the CPU and around 80c for the GPU which I find normal.

My question is: In theory this should not be thermal throttling, as all temps are within the norm but what else could it be? The PSU is an OCZ 750w so I don't think it is that either.

 

[sizzling]

What motherboard? That graph is it cpu clock speed or cpu usage?

 

[Draken21]

What motherboard? That graph is it cpu clock speed or cpu usage?

The top one is GPU temp - hovering around the 80c
The bottom one is the CPU temp dropping every few mins when the frame rate drop occurs. I have disabled the cool n quiet feature in case it is the one responsible for that but other than that I cannot think of anything else so any suggestions appreciated.

 

[sizzling]

You need to monitor cpu usage and cpu clock speed, with this information we can see if throttling is happening. If the usage stays high but the clock speed drops then throttling is happening.

You didn't answer which motherboard? There are plenty of low end motherboards that struggle to run 125w CPU's under prolonged use such as gaming. Often the motherboard VRM's overheat and then throttle the cpu to stop the motherboard frying.

 

[Draken21]

Yeap will do today and update the thread. Mobo should not be an issue as I have had it 3+ years without any problems. It is an 970 Extreme3 - not the most modern one but still pretty reliable.

 

[sizzling]

Is it this board http://www.asrock.com/mb/AMD/970%20Extreme3/

I am not familiar with it but it only has a 4+1 power phase setup and from the picture I don't think the VRM's have any heatsinks. I know it says up to 140w so maybe ok but the power design looks very low end. Anyway let's see if we can prove that throttling is or is not happening first.

 

[mdd1963]

blow a house fan inside the opened up case across/onto the mainboard, and see if throttling-like behavior changes...

 

[Draken21]

OK there is definitely some throttling going on but I cannot understand why.

Below are the PCs vitals.

CPU temp is within limits, GPU temp is within limits, PSU is sufficient. I have all the latest drivers. Cool n Quiet is disabled...any ideas?

Plus I have 2xfans blowing air in and 3xfans extracting air out. All of them Arctic F12 PMW.

I could use a house fan but according to temps overheating should not be an issue.

 

[ccampy]

It's the motherboard vrm Temps that are likely the problem rather then the CPU temps

 

[sizzling]

I'm guessing motherboard.

 

[Draken21]

Hmmmm I wouldn't think so as the mobo was working fine for 3 years. Failing I do understand, but temps spiking all of the sudden I find no reason.

 

[sizzling]

Hmmmm I would think so as the mobo was working fine for 3 years. Failing I do understand, but temps spiking all of the sudden I find no reason.

From what I have read VRM's deteriorate with use and overheating ages them faster. That's my theory but I cannot prove it.

 

[Draken21]

I think I found the ah heck... So Open Hardware Monitor and the rest were reporting the temp very wrongly.

Tried HWinfo and CPUID and both reported the CPU temp as

Which is exactly the thermal limit of the processor.

It does not help that the room temp atm is around 30c+ and its 23.00 at night around here.

Anyway now I have to find out why that is happening with a brand new H45. The only thing I can think of is that I have not connected the water pipe 3/4 pin wire which would be pretty silly from my end but still doable as I hear no noise except the fans.

Off to check if Corsair has a monitoring utility. Will keep you posted.

 

[maxwellmelon]

http://www.tomshardware.com/faq/id-2122665/understanding-temperature-amd-cpus-apus.html

use amd overdrive. if your cpu is near 0c or negative number then you are throttling as 0c is throttling start point. its best to be 10c or higher under max load test such as prime 95 small fet test ("thermal margin" works in reverse its how much more you can go to max temp). hwinfo and cupid go by motherboard temps and not the die temp sensor. but for the temp posted above amd recommends not to be over 62c so chances are your cpu is throttling but only looking at "thermal margin" can you know for 100% sure.

if your "thermil margin" is good then chances are its just the VRM causing the issue and because of there location they contribute to the socket temp being over 62c only way to fix it would be to get better airflow around the cpu socket.

 

[Draken21]

So I think it is safe to say it is the VRM causing the issue. And I think the fact that I added the H45 radiator above them did not help much.

AMD Overdrive is giving me min 11c margin. And I read various threads that the mobo has serious issues with 8 core AMDs. Off to install some custom fans blowing on the VRM as a short term solution.

 

[mdd1963]

Typically, once adding an AIO/liquid punp, you then remove some of the cooling of the chipset the CPU fan was providing....; add a fan pointing across the chipset/VRM components....

 

[Draken21]

Yeap that was it.

Couldnt fit any fans near the VRM so did the following.

a. Turned the H45 to face to the front
b. Added another fan below to push hot air out of the box
c. Added 3 intake fans (one at the back, two at the top of the case)

Now the mobo is running nice and cool although I messed up the air flow in my case so now my GPU is overheating. Do you think removing the top intake fans would make any difference?

 

[mdd1963]

normally , most prefer to intake air from the bottom and/or front of the case, and exhaust out the rear and/or top...

You can experiment with configs , monitoring your assorted temps...

But, you just need a little air across the VRM circuitry....

 

[Draken21]

normally , most prefer to intake air from the bottom and/or front of the case, and exhaust out the rear and/or top...

You can experiment with configs , monitoring your assorted temps...

But, you just need a little air across the VRM circuitry....

Yeap I know...that was my setup before as well...Unfortunately there was absolutely no way to cool the VRM down.

And now with the new setup although the CPU runs really cool the GPU is running 5c hotter but most importantly the fan is at 100% which is annoying.