CPU downclocking itself for a split second then going back to normal speed

[turbodan]

Hey all,

I've recently noticed this issue with 3x AMD processors. Specs (current) are AMD 8350 @ 4.2ghz, 16gb ddr3 Kingston HyperX 1866mhz, Radeon 470, and an older Gigabyte 970a-ud3 mobo (rev 1) and a 630w green series 80plus Rosewill PSU.

I always use HWmonitor. It shows stable temps for my cpu, and 99.9% of the time, stable cpu speeds. For example, upon boot, my cpu core speeds will all show 4200mhz. After a few hours of multitasking, gaming, etc, the HWmonitor will record any low the cpu hits - often times its about 3200mhz. When this happens, which isn't often, it only occurs for a split second. Like an immediate dip and then it's back up to 4.2ghz. Tbh, I would probably never notice unless HWmonitor showed me.

Not sure why this happening, and although my performance isn't necessarily affected and it doesn't stay downclocked, I'd like to know what's causing it. I also have an issue with Battlefield 1 crashing after 30-45min of gameplay now, and wondering if it's a PSU issue.

Should I worry about this? Like I said it's a split second dip - running at 4.2ghz 99.9% of the time then bam, 3200, and right back to 4200. This never really happened before I added more stuff to my comp - at least I don't think so. 630w Rosewill with specs above, but I also added a corsair strafe keyboard (illuminated w/ 2 usb ports), an extra fan, headset, speakers, etc. Do you guys think this is a PSU surge issue? Any suggestions for maximizing CPU power / saving some voltage to allow more power for the PSU? Like I said right now I have the PC plugged directly into the wall, then everything else is on the multiple outlet stick (w/e the f they're called). I also heard I can adjust something in BIOs to make sure the CPU is constantly running at max capacity.

Also, the voltage for the CPU fluctuates between 1.4 - 1.435 or around there. It doesn't stay at a set voltage. Obviously, when the voltage dips to it's lowest, the MHz of the processor may fluctuate too. I feel as if this may be leading to a crash in Battlefield 1. Any suggestions? Thanks

 

[chemmajorp53]

The psu would be the first place i'd look in this case, your voltage should not be fluctuating hardly at all. If you go into your bios you can set the load line voltage to be constant. I'd also consider your motherboard, the 970 chipset motherboards are not made to supply that much power and 1.4V may be too much. Maybe try downclocking a bit, because 8350s are well known for burning the power phases of lower quality motherboards.

 

[GhislainG]

The CPU voltage dips when the frequency drops; that's normal and it shouldn't be caused by the PSU. You have a good motherboard, but the CPU frequency may drop a bit because the VRM gets too hot; how are you cooling the CPU? Top-down blower type cooler to help cooling the VRM? It could also be the CPU that gets too hot; you should monitor the thermal margins with AMD Overdrive. If they get too close to 0°C, then the frequency will drop.

 

[turbodan]

That's the term I was looking for - LLC. Keep reading that that may be the solution for the voltage fluctuations, which in turn is causing the dips in my cpu speed. Will try changing that in BIOs, although I'm not sure where it is. Don't recall seeing it under manual voltage settings.

 

[turbodan]

The CPU voltage dips when the frequency drops; that's normal and it shouldn't be caused by the PSU. You have a good motherboard, but the CPU frequency may drop a bit because the VRM gets too hot; how are you cooling the CPU? Top-down blower type cooler to help cooling the VRM? It could also be the CPU that gets too hot; you should monitor the thermal margins with AMD Overdrive. If they get too close to 0°C, then the frequency will drop.

Have a Cooler master double fan / big heatsink type cpu cooler. Looks like the Evo 212 or whatever it is. Under max load, CPU temps never exceed 44c. I think that's pretty good for air cooled. If that's the case, do you guys think my VRM would be overheating? As far as I know, there's no way to check the "vrm" temp. Have yet to see the CPU go over 44 though, even during hard gaming.

 

[chemmajorp53]

I'm pretty sure CPU-Z let's you check VRM temps if your mobo makes that available. Your cpu temp is not the problem here. Your PSU should be replaced even if it's not acting up, so now would be a good time to do it.

 

[turbodan]

Great I just read the 970a Ud3 rev 1. (mine) doesn't have LLC control.

 

[turbodan]

What I will try is manually adjusting the voltage to 1.4 - something. When I overclocked my 1100t to 4ghz I had it set to 1.450, and it stayed at 4029mhz forever. I reset my bios when putting in the 8350, and now voltage is back to "normal." Either way, do you guys think this voltage fluctuation could result in my battlefield 1 crashing? The computer will just go black then reset.

 

[chemmajorp53]

What I will try is manually adjusting the voltage to 1.4 - something. When I overclocked my 1100t to 4ghz I had it set to 1.450, and it stayed at 4029mhz forever. I reset my bios when putting in the 8350, and now voltage is back to "normal." Either way, do you guys think this voltage fluctuation could result in my battlefield 1 crashing? The computer will just go black then reset.

Yeah, it's one of either the cpu voltage, the gpu crashing, or a RAM error. Pretty much the only options unless your game files are corrupted.

 

[turbodan]

What I will try is manually adjusting the voltage to 1.4 - something. When I overclocked my 1100t to 4ghz I had it set to 1.450, and it stayed at 4029mhz forever. I reset my bios when putting in the 8350, and now voltage is back to "normal." Either way, do you guys think this voltage fluctuation could result in my battlefield 1 crashing? The computer will just go black then reset.

Yeah, it's one of either the cpu voltage, the gpu crashing, or a RAM error. Pretty much the only options unless your game files are corrupted.

Don't want to jump to conclusions, but it most likely was the ram. When I upgraded to the 8350, I also bought 2x 8gb hyperx fury ddr3 at 1866mhz. My motherboard does support up to 2000mhz, but during gaming I would crash after a little while. Adjusted other things first, such as voltage on cpu, which actually did fix my fluctuations and dips - now it goes from 4027mhz to 4200, being normal to turbo - that's good. Either way I thought the issue was battlefield specific, but when playing WoW for about 10min, same exact thing happened - black screen to restart. Proceeded to go into BIOS and check voltage on ram, which was at the normal 1.5v. Figured I'd go ahead and try downclocking it to 1600mhz, and that apparently did the trick. Not sure what it is about 1866mhz ddr3, but it's more uncommon and seems to have the most compatibility issues. Going to leave it at 1600mhz for now.

 

[turbodan]

Turns out downclocking the ram didn't solve the issue and replacing the motherboard/ psu did. My biggest issue was Battlefield 1 crashing after 15-45min of gameplay. At first, in windows 7, I'd get an error saying my performance was low and to change windows aero settings. Even after doing this, it would still crash. I figured Win 10 would help, esp. with dx12, so I upgraded, but same thing - after an hour of play, crash. Voltage / speed on the cpu was still fluctuating too much.

Went to microcenter the other night and got a MSI 970 carbon gaming mobo, and a 750w psu. As of now, everything runs perfect. My thought is that the old Gigabyte 970a-ud3 was half shot, it still ran, but under heavy load (like battlefield 1 completely maxed for an hour +), maybe something, like the VRM would overheat. This could be why the voltage would drop and the cpu speed would go from 4.4ghz to 3.6 sometimes (even with energy saver off).