Lower or equal performance with new 970 than old 760 in some games.

[adc]

Around 6-12 months ago I upgraded from my GTX 760 to a GTX 970, and had this same problem which caused me to return the card. I got another on Black Friday in the hopes the issue would be fixed. Unfortunately, as far as I can tell it isn't.

Running these parts:

Mobo - Asrock 970 Extreme3 AM3+ board.  
CPU - AMD 8350 stock @ 4.0 GHz.  
GPU - EVGA's SC Gaming ACX 2.0 GTX 970.  
RAM - 16 GB DDR3 1600 (G.Skill Ripjaws I believe).  
PSU - Corsair CXM 600W PSU.  

I am around 97% sure this is not a CPU throttling issue; after running DotA 2 (which is a game I have this issue in) for around 5 minutes, HWMonitor reports these numbers:

http://puu.sh/lX2VP/1fbe333f36.png

None of my cores are ever maxed out, and most of the time they're between 20% and 50% usage. Also, I realize my GPU usage is reported as 0%; I don't know why this is (and it happens in both EVGA Precision X as well as MSI Afterburner as well), though I assume it's a faulty "sensor" (can that happen for GPU usage reporting?). For reference, my previous 970 showed around 30-35% usage with the same performance, so I assume it's roughly the same here.

I've turned kboost on in EVGA Precision X, as well as turned all power settings to max performance; however, I still max out at 80 FPS with all settings (except render quality) off or low, dropping to 50 fps or lower in teamfights.

Let me know if any more info is needed, I'm at a loss at this point.

[Amazingasian]

Around 6-12 months ago I upgraded from my GTX 760 to a GTX 970, and had this same problem which caused me to return the card. I got another on Black Friday in the hopes the issue would be fixed. Unfortunately, as far as I can tell it isn't.

Running these parts:

Mobo - Asrock 970 Extreme3 AM3+ board.  

CPU - AMD 8350 stock @ 4.0 GHz.  

GPU - EVGA's SC Gaming ACX 2.0 GTX 970.  

RAM - 16 GB DDR3 1600 (G.Skill Ripjaws I believe).  

PSU - Corsair CXM 600W PSU.  

I am around 97% sure this is not a CPU throttling issue; after running DotA 2 (which is a game I have this issue in) for around 5 minutes, HWMonitor reports these numbers:

http://puu.sh/lX2VP/1fbe333f36.png

None of my cores are ever maxed out, and most of the time they're between 20% and 50% usage. Also, I realize my GPU usage is reported as 0%; I don't know why this is (and it happens in both EVGA Precision X as well as MSI Afterburner as well), though I assume it's a faulty "sensor" (can that happen for GPU usage reporting?). For reference, my previous 970 showed around 30-35% usage with the same performance, so I assume it's roughly the same here.

I've turned kboost on in EVGA Precision X, as well as turned all power settings to max performance; however, I still max out at 80 FPS with all settings (except render quality) off or low, dropping to 50 fps or lower in teamfights.

Let me know if any more info is needed, I'm at a loss at this point.

Did you update drivers?  If you did, try reinstalling them.

[adc]

Yes, I did multiple clean reinstalls last time too. I'll give it a try, though.

[dragoon20005]

i think your mobo is thermal throttling

 

it only has a 4+1 power phase

 

and looking you are running a FX8350

 

those CPU need a lot of power

[adc]

i think your mobo is thermal throttling

 

it only has a 4+1 power phase

 

and looking you are running a FX8350

 

those CPU need a lot of power

It's not. I'm directly monitoring my CPU temps and clocks, and as you can see from the picture during my test (where I saw the issue the entire time) it never went more than 40 MHz below the norm. I can very, very easily get my board to thermal throttle - the power phase does suck - but it isn't in this case.

[adc]

Did you update drivers?  If you did, try reinstalling them.

Apparently my original uninstall wasn't a full clean install and something from the previous driver lingered; a full clean uninstall using DDU gave me a noticeable performance boost (10-20 FPS depending on the situation in DotA, and 100~ in CSGO). I still feel like there's something strange going on because none of my CPU cores ever break 90% usage; should it not be peaking at 100% and thermal throttling instead of stopping at 90%~?

[Amazingasian]

Apparently my original uninstall wasn't a full clean install and something from the previous driver lingered; a full clean uninstall using DDU gave me a noticeable performance boost (10-20 FPS depending on the situation in DotA, and 100~ in CSGO). I still feel like there's something strange going on because none of my CPU cores ever break 90% usage; should it not be peaking at 100% and thermal throttling instead of stopping at 90%~?

Thermal throttling doesn't seem to be an issue.  30C is a good idle temp let alone load temp.

[adc]

Thermal throttling doesn't seem to be an issue.  30C is a good idle temp let alone load temp.

 

Oh for sure, my current temperatures are great (largely because it's only using 1-3 cores most of the time). That said, I don't know why it would only use 90% of those cores, as opposed to pushing them to 100%.

[Amazingasian]

Oh for sure, my current temperatures are great (largely because it's only using 1-3 cores most of the time). That said, I don't know why it would only use 90% of those cores, as opposed to pushing them to 100%.

Well... what stress test/game are you running?

[adc]

Well... what stress test/game are you running?

 

DotA 2 (70-90 FPS depending, 1600x900 all settings low except shadows/texture/render quality) and CSGO (150-600 FPS depending on map/players, 1080p lowest with max shadows), as well as Furmark (rock solid 88 FPS on a 1600x900 no MSAA stress test). I'm definitely expecting some CPU throttling, but I expected it to use all of the cores the games are running on at the very least.

[Amazingasian]

DotA 2 (70-90 FPS depending, 1600x900 all settings low except shadows/texture/render quality) and CSGO (150-600 FPS depending on map/players, 1080p lowest with max shadows), as well as Furmark (rock solid 88 FPS on a 1600x900 no MSAA stress test). I'm definitely expecting some CPU throttling, but I expected it to use all of the cores the games are running on at the very least.

DO you have a CPU and GPU heavy game in your library?  Otherwise, download 3dmark and run the Firestrike test.

[adc]

DO you have a CPU and GPU heavy game in your library?  Otherwise, download 3dmark and run the Firestrike test.

http://puu.sh/lXoDi/9e5816abdf.png

[Amazingasian]

http://puu.sh/lXoDi/9e5816abdf.png

That's what I would expect from a pic like yours. Maybe try overclocking and flashing a custom bios on the gpu. Check out the GTX 970 Owners club on the LTT forums too. It's great.

[SkilledRebuilds]

http://puu.sh/lXoDi/9e5816abdf.png

Graphics scores seem normal (for an non-overclocked 970)

Score seems low due to Physics scores, so you should only be looking at the Graphics score to compare GPU to another 970.

[adc]

That's what I would expect from a pic like yours. Maybe try overclocking and flashing a custom bios on the gpu. Check out the GTX 970 Owners club on the LTT forums too. It's great.

 

 

Graphics scores seem normal (for an non-overclocked 970)

Score seems low due to Physics scores, so you should only be looking at the Graphics score to compare GPU to another 970.

 

Is it normal for my CPU cores to not be totally maxed out, then? Would that have to do with other programs running on those cores reserving a few percent or something?

[SkilledRebuilds]

Is it normal for my CPU cores to not be totally maxed out, then? Would that have to do with other programs running on those cores reserving a few percent or something?

In the Physics test my CPU is using 4x 95-100% during the test, its weird if yours isn't doing the same as thats the intended method.

[adc]

In the Physics test my CPU is using 4x 95-100% during the test, its weird if yours isn't doing the same as thats the intended method.

 

Sorry, I didn't mean in the Physics test - I just meant running games and such, where it seems my cores only go up to 85-90% usage.

[SkilledRebuilds]

Sorry, I didn't mean in the Physics test - I just meant running games and such, where it seems my cores only go up to 85-90% usage.

Yeah thats normal (I know its stupid to say that but its all too common), even if a game only used 1-2 cores, that does not mean they will be 100% usage.

Plus once they hit 100%, other things also trying to use that CPU would slow down so its not ideal to do.

IMO you'd always want a 1-2% spare.