I have been experiencing a frustrating issue for a little while now. My rig is 2 years old and I built it myself. So far, no major problems.
However, I do get this problem in WoW where I go to a certain area that really doesn't require much in the way of CPU or GPU to be able to render and the processor randomly jumps about in the region of 1-2ghz then shoots up to 4ghz+.
The frame rate is also impacted at the time of the drop by halving from 140 to 70ish. Bearing in mind that it is a solid 140fps when processor is clocking at a normal 4ghz+. If I go into an 80 man raid then I get anywhere from 40-70FPS and that is to be expected, but this sort of environment where these is nothing going on should not cause such massive fluctuations in FPS, or processor clock speed for that matter.
When this happens, I am standing stationary, monitoring frame rate and task manager and it all happens in perfect sync, as soon as frame rate drops, the clock speed of the processor drops.
I have completely reset my bios, flashed my bios, changed power plans, uninstalled ryzen master, manually changing the TDP and voltage curves etc etc etc, even run the game without any addons running and I have not had any success at curing the issue. Like I say, there are 1 or 2 specific areas where this happens. I can run a 25man raid, with 25 other people at a solid 120+fps without any issues whatsoever, then when it comes to this basic, static scenery this happens. It's puzzling me massively.
One thing to add is that I could fly around this problem area for 10+ minutes stuck on a solid 140fps ... then it happens once (fps drop and processor clocks down) and shoots back to 140fps. Then it will constantly do it and the processor and fps will remain locked at 1-2ghz and 70-100fps, maybe going back to 140fps for a brief few seconds then back down.
Lastly, it cant be temps. I dont recall ever getting above 70c on my processor as I have a top class case and many, many fans. When it does this downclock problem its around 50-60c.
My specs are as follows:
RTX 3090 (downclocked to be stable, no changes running at stock)
3600Mhz DDR4 Ram
Aorus Elite x570
1000w Corsair PSU
WD SN850 2TB (Primary drive)
Sabrent Rocket 2TB
If anybody has any tips for things to try i'd really appreciate it.
Have been monitoring this more closely recently as its the only thing really niggling me about my build and i've realised it does actually do it a small amount during other activities in games - just not as severe as it does in the other scenario I told you about. It can clock down to 2-3ghz in 25 man raids for example. Why is this happening?? Any help please.
I don't recommend using task manager to check clock speed. Use either Ryzen Master or HWINFO64. If you use the later, check as well the "effective clock" and post ss after doing raid or complex scenario.
SS of temps and if you are using optimized defaults in BIOS?
Why downclock GPU, isn't it stable at default clocks?
Kpal11, please post a screenshot of Ryzen Master (RM) running Cinebench R23. What processor cooler are you using? Thanks end enjoy, John.
Hi guys, thanks for responses. I've been really trying to cure this over the last few days and believe I have sorted it. In Ryzen Master Game Mode there is an option: Legacy Compatibility mode that... on the off chance I did a quick google to see what it did - found a lot of ppl complaining about it. Turned this off and Boom, sorted. Have not managed to trigger it again since! Mental that a setting optimised for games seems to cause such a frustrating and elusive problem. I finally have my PC running on its A-Game and am finally happy with it lol. If I had more time off work I probably would of found this sooner but i've been very, very busy.
Btw, GPU is downclocked just to reduce heat and power. It's undervolted as best as it possibly can be whilst being stable. I want to do this with CPU now and will be figuring out optimal single core curves soon, if I can be bothered as its a lot harder to get right than GPU. I have -6 on all cores being stable for months but I know -7 crashes. Probably lost the draw on silicone lottery there!
Thanks again anyways.
If you are only using Curve Optimizer alone, then you are not undervolting at all. CO is only a curve shift in magnitude, where it tells your CPU that it can do higher clock in mhz "- negative value" for the same voltage. So, if you are not limiting the clock frecuency of your CPU your voltage is the same if you are using Curve Optimizer. The easier way to understand this is for example if your CPU on stock gets 1.45Vcore for 4.9ghz and then you set a negative curve optimizer of -10 then when your CPU gets 1.45V it will do higher clock than 4.9ghz for the same 1.45V. If you limit your CPU clock frecuency, then yes you should see a lower Vcore. For heavy multi thread load Curve Optimizer won't help you with temps, and will probably cost you some performance "cores will try to boost higher for the same voltage and fail" stability issues. You can bench with cinebench r23 and check effective clocks on HWINFO64. Then, you can try Curve Optimizer disabled, and set custom PBO limits limiting the actual CPU power draw, which in fact will impact your heavy multi thread load "lower temps" if set below stock values. I don't know how the 5950X behaves on this, but my 5800X does the same performance, for less power and lower temps.
This Video is very handy: https://www.youtube.com/watch?v=c45KXBnbbpk