Made the switch from Nvidia to Amd R9 390 but have low frame rates

Heres another. It seems to be even worse.

Edit: My clocks seem to be dropping as low as 980mhz on core

 

Do you use temperature target? I noticed that Fury X downclocks a bit in case I set temperature target to something too low (like 30°C) as cooling can't keep it there.
(That could explain that small drop in GPU clock.)
Rest looks like CPU/HDD bottleneck. That can be identified by mentioned sitting in low fps area for long time. If you have low fps for long time while watching same place, it is CPU bottleneck. If fps improves by itself in few seconds, then it is caching (HDD).

 

When i test my cpu with occt there is only 100% cpu utilisation for a couple of spikes and thats it. the majority of the time its around 60-70% so surely that cant be my cpu bottlenecking as it would be hitting 100% for the majority whereas the frame rate drops are constant. A hard drive bottleneck wouldnt really be causing low constant fps as well surely? Hdd's stutter if they cant keep up with the load normally. You might be onto something though as keeping it at the same spot without moving the frame rates are around the 50's in low framerate areas but as soon as i move forward and for the whole time im moving the fps drop to around 30 looking at the exact same are. If it was hard drive caching once it was done the framerate would go up surely while i was moving but it doesn't. Its only as i move.

Edit: No i don't use temperature target, my gpu only ever reaches 75°c is there a temperature throttle built into the drivers maybe?

Edit: Here is whats happening with my CPU

 

I don't know about Crysis 3, but I can assure you that GTA V's Vsync is bad. A lot of older games seem to behave weirdly under Windows 10, especially ones with full screen modes. The performance killing "Advanced Graphics" in GTA V are most likely the reason for your frame rate drops, along with the double buffered vsync of the game. Try enabling it through RadeonPro, disabling those features, and see how it plays.

 

Heres a gpu-z of me playing the Witcher 3. Same weird gpu utilisation. Even though it feels smooth and like it 50-60fps it stutters horribly every 4 or 5 seconds and i think those stutters are linked to the gpu utilization spikes but im not sure. Ive heard people talk about turning upls off but will that affect anything?

 

It helps only if you dual-gpu configuration but hey, isnt hurt to try.

Did you try adding voltage to the gpu?

 

Already got it set at 50%

 

You are having the weirdest issues. I keep thinking if I've had such spiky usage but nope. That is properly weird. Erm try disabling ulps, forcing constant voltage and keeping that 1050 clock to see if that does anything.

My 290x spikes around in WoT because my gpu usage is around 50%.

 

I know, it's driving me insane. I was so excited about getting this GPU as well as i hadn't upgraded in years and was hoping this would be a god level upgrade. Making me wish i had stuck with Nvidia now but im just not happy about the memory config of the GTX 970.

How do i disable upls and force a constant voltage?

 

MSI Afterburner can do both, check the settings

Really think you should try a clean windows install too though, if nothing else, to rule software out

 

I think what will help a lot is by updating your pc specs so we know exactly what you have. So have you changed the psu out yet?

 

Op, windows 10 still has issues with r9 390 cards and drivers, mayby that is your problem?

 

Awesome thanks for that. Gonna try doing a dual boot a bit later and see if Windows 7 changes things.

 

Ahh ok that might be why. Even more reason to dual boot Windows 7 then. What kind of problems are we talking?

 

He does not suffer from GPU throttling, he suffers from micro freezes/bottleneck of some kind.

I had sh*t episode of my own due to Windows 10 I did not share yet from pure embarrassment that I managed to suffer from it for few weeks. And that it took me 2 days of work to solve once I got mad at it enough:

With video playback I had occasional freeze of system from which OS recovered as soon as I moved mouse, clicked or taped key on keyboard. This freeze had one effect, it stopped video playback and sound did loop in like 0.1s loop. Sometimes it did happen every 5 minutes, sometimes once per hour, pure annoyance.
In games it did happen sometimes too, same recovery method worked. But in Guildwars 2 freeze required <ALT>+<TAB> and return to game which made mmorpg game unplayable.
(I can say that my OS was pretty clean, as always and that it started after one of windows updates, so I thought it would go away, but it did not.)

I did override every windows update driver for each device I could. Even to very old one, nothing fixed that issue.
I went as far as disabling any graphical HW acceleration of browsers, media players. Tested bunch of soundcard drivers. Still nothing.

Reinstalled OS, did my usual minimum software installation without I could not imagine life. Issue still there. Now I started to consider that CPU/memory/board is about to die.

1st step, disabling all power saving on USB and removal of any unnecessary device. Nothing changes. Then updated Firmware in mouse. No change...

I did loosen memory timing, downclocked CPU to 4GHz and objectively observed freeze rate. It went down. I was not happy at all from direction it took.
And I started to think what thing can be affected by 'tight' timing. DLed LatencyMon (that one which actually works with Windows 10) and 1st look was look on HORROR high numbers going up to 2ms, yes milliseconds. Red was everywhere.
I went to detailed look and expected USB or soundcard driver. To my awe I did found HWiNFO64. I immediately killed it and numbers went to lovely green microseconds where they belong.
HWiNFO64 is off now, everything is back to perfection not one hitch since then.

For me microsoft changed something what collided with intrusive method HWiNFO used to read data from chips.

Weirdest thing was: as freeze occurred I let it continue for 15~30 seconds and it would not stop till I sent some input. I would have believed that chocking CPU for 20 seconds would result in BSOD/driver unresponsive error, but no such thing.

On windows 10, even something what works usually to your satisfaction can become your worst enemy. Maybe dennymoofoo has some similar software which touches wrong nerves and Windows panics from it.
Maybe even afterburner statistics can cause issues to some degree.

Anyway, Windows 10 is 1st OS where I have observed time unlimited freeze from which one can recover at will by mere mouse move/press or keyboard input.

 

It's just something I picked up reading several posts in win10 forums.
Mostly complaining about ms delivering an unfinished os and regularly updating it. Drivers for 390 weren't adequate yet, causing crashes and low performance.
Could be solved before the end of the year ofcourse, but in general there is some driver lacking in windows 10 for several hardware components.

As always, some have serious problems while others have non.
Dual booting with win7 would be a smart move yes. At least that's a tested environment.

 

Here Bro -> http://forums.guru3d.com/showthread.php?t=398484

Check it out and see what best for you ;-)
Also it's recomended to Uninstall All nV soft from your RIG and clean the registry !
then go and do a Fresh Start !