[SOLVED] AMD FX-8350 clock speed dropping every 20 seconds

[tempDLUser]

Hello everyone!
I'm new here and hope this sub-forum is the right one for my question.

I've lived happily with an AMD Phenom II x4 955 BE for a long time and recently upgraded to an AMD FX-8350.
Since then I have been experiencing problems with GTA V, the game that challenges my rig most: Every 18-20 seconds there are strong performance losses (frame drops, jerks), after another 18-20 seconds everything is fine again and runs very smoothly. The whole thing repeats itself constantly and indefinitely.

The attached screenshots show my task manager, which shows how the clock speed of my CPU falls and rises in exactly this rhythm between 4.1 GHz and 1.39 GHz. My GPU (NVIDIA GeForce GTX 750 Ti, worked with the old processor) does exactly the same.

I've read elsewhere that this might be caused by cooling problems. However, I have an Artic Freezer in it that keeps my CPU at 13°C at idle and 32°C at full load. That shouldn't be a problem.

Via BIOS I tried to force my CPU to 4.1 GHz, apparently without success. The Windows 10 energy saving plan speaks of "high performance".

What else could be a reason? Why does performance rise and fall in exactly this regular rhythm?

Most important information about my crutch:

• OS: Windows 10 Home Premium (64-bit)
• CPU: AMD FX(tm)-8350 Eight-Core Processor, 4100 MHz, 4 cores, 8 logical processors
• GPU: NVIDIA GeForce GTX 750 Ti
• RAM: 8 GB (5.9 in use with GTA V running)
• BIOS version/date: American Megatrends Inc. 2101, 02.04.2015
• MB: ASUSTeK Computer INC. M5A78L LE

And especially the GTA V settings:
Total average graphics settings, no special performance requirements


I hope you can help me!

 

[rgd1101]

make and model of the psu
could be the motherboard can't handle

 

[tempDLUser]

make and model of the psu
could be the motherboard can't handle

PSU is a LC-Power LC420H-12, should be delivering 420W. Online calculators told me that I only need about 285W for my setup.
My motherboard lists FX-8350 in its supported CPU list, should be able to handle it, doesn't it?

 

[sizzling]

It’s the motherboard, this is a common problem with 125w cpu’s on cheap motherboards, really the 125w CPU’s were only suited to higher end motherboards. From a quick Google your motherboard only has a 4+1 VRM with no heat sink.

What cpu cooler are you using? The stock cooler with top down airflow was one of the better for cooling the motherboard vrm’s.

 

[tempDLUser]

It’s the motherboard, this is a common problem with 125w cpu’s on cheap motherboards, really the 125w CPU’s were only suited to higher end motherboards. From a quick Google your motherboard only has a 4+1 VRM with no heat sink.

What cpu cooler are you using? The stock cooler with top down airflow was one of the better for cooling the motherboard vrm’s.

Thank you for your answer!

cnet says: The ASUS M5A78L-LE supports FX processors with max. TDP of 125 watts. Can this really become a bottleneck because there is not enough leeway?

My cooler is an ARCTIC Freezer 13 (92 mm PWM fan with up to 200 Watt cooling capacity, according to the manufacturer) on the CPU, plus a small fan on the back of the case to remove the warm exhaust air.

Edit: I just made a test run with CPU clock limited to 3.4 GHz (17 x 200). Temperature never reached 42°C under full load, drops still occured but now every 70 seconds. I'd be glad if I could be sure whether my mainboard or my PSU is probably a bottleneck.

 

[sizzling]

Supporting and running properly was unfortunately 2 different things with the old FX series. This issues was quite common.

A psu cannot cause a cpu to throttle, it can cause stability issues but not throttling.

 

[hftvhftv]

Reset your BIOS settings and make sure they are on normal, not Asus energy saver, optimized, or auto overclock or anything like that.

 

[tempDLUser]

Supporting and running properly was unfortunately 2 different things with the old FX series. This issues was quite common.

A psu cannot cause a cpu to throttle, it can cause stability issues but not throttling.

Thank you for that answer!

Reset your BIOS settings and make sure they are on normal, not Asus energy saver, optimized, or auto overclock or anything like that.

I restored default settings in BIOS, then disabled all the ASUS energy related features, forced the CPU to max. 3.2 GHz and made the CPU-Fan spin at 100%. The result: HWMonitor shows brilliant values (cf. screenshot), MB temperature is 52°C (125°F) max, CPU temperature is 42°C (106°F) max. That shouldn't be a problem at all, should it?
The speed drops (lag spikes) still occur, but now only every 120 seconds. I don't see the point of throttling the CPU any further and apparently the motherboard has no reason to interfere, because the temperatures are completely within bounds (assuming the sensor is at the right place). Then why is it doing this anyway?

 

[sizzling]

Thank you for that answer!




I restored default settings in BIOS, then disabled all the ASUS energy related features, forced the CPU to max. 3.2 GHz and made the CPU-Fan spin at 100%. The result: HWMonitor shows brilliant values (cf. screenshot), MB temperature is 52°C (125°F) max, CPU temperature is 42°C (106°F) max. That shouldn't be a problem at all, should it?
The speed drops (lag spikes) still occur, but now only every 120 seconds. I don't see the point of throttling the CPU any further and apparently the motherboard has no reason to interfere, because the temperatures are completely within bounds (assuming the sensor is at the right place). Then why is it doing this anyway?

Those motherboard temperatures are not related to the VRM's. Setting the fan to 100% has likely improved airflow around the motherboard and VRM cooling. You could try rigging a fan pointing directly at them which is what some people have been known to do with varying results. As a test you can try taking the side panel off and pointing a desk fan at the CPU area and see if things improve.