HD 7850 Low Framerates -- Need Help

[MichelOutils]

Note: I've posted this on other online forums.

I'm about at wit's end. Yesterday I was running Dota 2 and I was getting 55 to 93 fps. Today it's stuck between 28 and 30 fps.

I have HWinfo64 and RivaTuner running with OSD to monitor everything that I think I should (CPU cores' temps and total usage, GPU temp and usage). GPU usage is stuck at 100% with temps around 70*C; total CPU usage hovers around 45%, 55*C temp. I'm not sure why, but it seems this HD 7850 is always at full load even when it isn't doing anything (i.e. even when all settings are dialed down to the lowest, GPU usage is still at 100% and temp at 70*C).

It is worth noting that on any normal day I'd get around 65% to 85% total CPU usage and around 80% GPU usage when running Dota 2. But today everything just looks odd, as CPU usage doesn't exceed 50% but GPU usage is always at 100%

Things I've tried so far:
1. Uninstalled driver, downloaded latest driver (dec 2016) from AMD website and installed it -- no improvement;
2. Used DDU (display driver uninstaller), reinstalled newest driver -- no improvement;
3. Ran Malwarebytes, 0 malware found -- no improvement;
4. Overclocked via AMD overdrive -- no improvement; Reverted to stock speeds, no improvement;
5. Tested other games, there are noticeable framerate drops (e.g. Sleeping Dogs normally stable at 60fps on custom mid-high settings, but today it's stuck at 28fps with occasional spikes to 55fps; GTA V which is normally stable at 60fps on custom mid-high settings ran 30fps today with occasional spikes to mid 40s; Battlefield 4 which normally runs at 60 fps on custom mid-high settings now runs at around 28 with occasional spikes to mid 40s);
6. Ran 3D Mark Firestrike, got a score of 3262 (significantly lower as compared to the score I got when I ran the same test on Feb 19, which is 4241)
7. Went to BIOS, loaded optimal default settings -- no improvement.
8. Verified Dota 2 game cache, no improvement.
9. Uninstalled Dota 2, reinstalled, dialed settings down to lowest -- got around 35fps, a measly 5fps improvement at most, but looks real ugly now -- GPU usage still at 100% while CPU usage still around 45%, per HWinfo64/RivaTuner OSD.
10. Unplugged my second monitor, made sure Windows only detects a single monitor. No go.
11. Did a system restore (restore point March 9 2016) -- no go.

My PC specs:
- AMD Phenom II X4 965 BE cooled with DeepCool Ice Wind Pro (not overclocked);
- HIS HD 7850 1GB GDDR5 256-bit (via PCI-Ex16 slot -- I bought this used in December of '15, it was running perfectly fine until today)
- MSI 770-C45
- Kingston HyperX Fury 4GB DDR3-1600+Geil Value RAM 2GB DDR3-1333 (dual channel mode, default speed 1333mhz, total RAM 6GB )
- Zalman ZM600 HP 600w semi-modular
- Seagate Barracuda 500GB
- Windows 10 Pro 64-bit activated (upgraded activated Windows 7 64 bit in July 2015)

I have a dual monitor setup, both are Samsung SyncMaster S20B300's.

Anyone ever experienced this? If so, did you get to figure out what was wrong? Or do you all think this could be a sign of a dying GPU? Or does AMD just suck so bad? (I used to be an AMD fan, but meh).

 

[Killrose]

Make sure your CPU is running full speed, check it with Hardware monitor after running some app that will stress it. Maybe your mainboard bios has it locked at a lower multiplier than it should be, ie you've lossed several hundred Mhz on the high end.

 

[TheELF]

On my nvidia it happens sometimes that it stays in 2d frequencies even when running a game,so it will run at like 300mhz and 100% usage...
So make sure that your GPU reaches max freq in game,run a game in windowed mode and run gpuz in the background to make sure.

 

[Bacon1]

Since you have two monitors, have the game on one, and task manager (with cpu usage memory page open), afterburner stats or other program to show GPU usage, GPU Temp, GPU VRAM, GPU Core Clock and GPU Memory Clock.

Its probably not raising the core / memory clocks properly, if thats the case, you can manually "overclock" them back to normal speeds using afterburner.

 

[DAPUNISHER]

Fan speed when running 70c? That PSU you have with the 4 +12v rails sectioned off like that is odd to me. Maybe the one the 6 pin PCIe is connected to is dying. Try it in another +12, and with another 6 pin cable if you can. Looks like V2 and V4 are designated for it.

 

[MajinCry]

I'm amazed you're able to even use the computer, with the 7850 @ 70C. My Sapphire 2GB 7850 died when the temps reached 58C, and began to underclock itself at around 52-53C.

RMA'd it, got a new one, downclocking still occurs a bit past 50C. Took out a janky fan, badly zip-tied a 120mm case fan to the shroud, and now it only breaks 50C when I run it at full throttle (e.g, Furmark, needlessly high ENB settings).

So yeah. I think your card is downclocking itself due to the temps.

 

[SPBHM]

70ºC is VERY healthy, no need to worry,

check if the clocks are running at the correct speeds while gaming

one thing I would make sure is if there is no other software (like web browser) forcing video encoding clocks (lower) while you play a game.

 

[3DVagabond]

I'm amazed you're able to even use the computer, with the 7850 @ 70C. My Sapphire 2GB 7850 died when the temps reached 58C, and began to underclock itself at around 52-53C.

RMA'd it, got a new one, downclocking still occurs a bit past 50C. Took out a janky fan, badly zip-tied a 120mm case fan to the shroud, and now it only breaks 50C when I run it at full throttle (e.g, Furmark, needlessly high ENB settings).

So yeah. I think your card is downclocking itself due to the temps.

I'm thinking something is giving you false readings.

 

[tential]

I'm amazed you're able to even use the computer, with the 7850 @ 70C. My Sapphire 2GB 7850 died when the temps reached 58C, and began to underclock itself at around 52-53C.

RMA'd it, got a new one, downclocking still occurs a bit past 50C. Took out a janky fan, badly zip-tied a 120mm case fan to the shroud, and now it only breaks 50C when I run it at full throttle (e.g, Furmark, needlessly high ENB settings).

So yeah. I think your card is downclocking itself due to the temps.

I'm amazed you've been here for 1 year and have such misinformation to spread on GPU technology.....

 

[MichelOutils]

I'm amazed you're able to even use the computer, with the 7850 @ 70C. My Sapphire 2GB 7850 died when the temps reached 58C, and began to underclock itself at around 52-53C.

RMA'd it, got a new one, downclocking still occurs a bit past 50C. Took out a janky fan, badly zip-tied a 120mm case fan to the shroud, and now it only breaks 50C when I run it at full throttle (e.g, Furmark, needlessly high ENB settings).

So yeah. I think your card is downclocking itself due to the temps.

When I'm running any game, GPU clock speed is 860mhz, fan speed is at 48%. I'm thinking maybe I should flash the video card's BIOS.

Also, I live in a country where there's only two seasons: summer and rainy. It starts to get really hot during March (30 to 35*C) and can get even hotter in April and May. IMO, 70*C isn't too high here, one of my other video cards (Galaxy GTX 560Se) can reach up to 85*C in full load and I've had it running like that since 2012, it still works great.

70ºC is VERY healthy, no need to worry,

check if the clocks are running at the correct speeds while gaming

one thing I would make sure is if there is no other software (like web browser) forcing video encoding clocks (lower) while you play a game.

I'm running Dota 2 right no with no browser or any other app that might be eating up GPU resources. I only have HWinfo64, RivaTuner and Fraps on. It's running at 860mhz, GPU utilization at 100%.

I'm thinking something is giving you false readings.

Would be glad to know if there are other hardware monitoring software more reliable than HWMonitor and HWinfo64.

 

[MichelOutils]

On my nvidia it happens sometimes that it stays in 2d frequencies even when running a game,so it will run at like 300mhz and 100% usage...
So make sure that your GPU reaches max freq in game,run a game in windowed mode and run gpuz in the background to make sure.

It is running at max frequencies when I'm running games, as both HWinfo64 and GPUZ tells me.

Make sure your CPU is running full speed, check it with Hardware monitor after running some app that will stress it. Maybe your mainboard bios has it locked at a lower multiplier than it should be, ie you've lossed several hundred Mhz on the high end.

HWinfo64 says each core runs at 3,408.7 mhz, Core VID for each is 1.400V. I also chose High performance plan, CPU minimum and maximum load always at 100%.

Fan speed when running 70c? That PSU you have with the 4 +12v rails sectioned off like that is odd to me. Maybe the one the 6 pin PCIe is connected to is dying. Try it in another +12, and with another 6 pin cable if you can. Looks like V2 and V4 are designated for it.

Tried it a few hours ago, it didn't work. Unfortunately I'm unable to get the actual fan speed -- could be a limitation on the type of fan connector the manufacturer used.
I do have another PSU that's the exact brand/make/model and I have it in another system running i5-2500 and R9 270. I'm not having any issues with it. Plus if the rail is dying don't you think the GPU shouldn't even come on?

BTW thanks for replying, you guys. I appreciate it. Please keep the suggestions coming. I need all the help I can get.

 

[MichelOutils]

I've fixed it. I opened the case and for some reason the GPU wasn't fully seated, i.e. it's screwed tightly in place on its rear end, but the front end kind of came loose for some reason, must be the latch on the motherboard.

I re-seated the GPU and lo and behold! It's working perfect again. Thanks everyone!

 

[MajinCry]

I'm thinking something is giving you false readings.

Eh. I used GPU Temp, GPU-Z and MSI Afterburner to measure the temperature, back when it was buggering up, and after the RMA.

All of 'em had the same readings.

I'm amazed you've been here for 1 year and have such misinformation to spread on GPU technology.....

Ya what?

'Tis what me experience with the ol' 7850 yielded.

 

[Insomniator]

I've fixed it. I opened the case and for some reason the GPU wasn't fully seated, i.e. it's screwed tightly in place on its rear end, but the front end kind of came loose for some reason, must be the latch on the motherboard.

I re-seated the GPU and lo and behold! It's working perfect again. Thanks everyone!

Interesting, anyone know the technical details about something like this? I would think it either works or doesn't work, not just works with lowered performance. Not getting enough power, no contact with all PCI lanes?

I know the practice of re-seating parts goes back forever.. but a video card half un-seated would seem to show more symptoms than just performing worse.

 

[PontiacGTX]

check the temperature of the VRM, also use 15.11.1 non crimson drivers