Frame rates lower than expected [980ti]

Yunimo

Distinguished
Mar 29, 2015
33
0
18,530
I recently got passed down a EVGA 980ti SC; I am concerned with one thing. I was getting better FPS with my previous card HD7850 which shouldn't be the case :??:

Listed below is my rig as of now:

Mobo: GIGABYTE 970A-UD3P
CPU: AMD 8350
RAM: 16GB corsair vengeance pro
GPU: EVGA 980ti SC
PSU: EVGA 650W G1

Any help would be greatly appreciated as I would like to get to the bottom of this.

Kind regards.
 
Solution
A 7850 will run LoL fine. FX CPUs arent all great in that game, but check to make sure you haven't set an GPS limit in the games video menu. There is an FPS cap in the game.
If you haven't already, install MSI Afterburner, and run the OSD to show CPU and GPU usage. Now, run any game in which you feel that performance is degraded. Check the GPU usage - if it's in the 96-100% range, there's some problem. If it's below about 90-95%, then your CPU is bottlenecking the GPU. You need a better CPU to remove the bottleneck. You could also try overclocking the CPU if possible.

If the GPU usage is at or around 100%, then it's just the FPS you'll get. Can you tell us the game, resolution, settings and FPS you are at?
 
GPU usage is around 30% same with CPU.
All temps are fine. CPU is sitting at 4.2

The thing I can't get my head around is why my old card, is still out performing the 980ti regardless of a bottleneck.

The game that's mainly played at the moment is League of Legends; I'm playing on medium-very low on some settings and I'm getting around 80 FPS.

~~~ sorry for the late reply I've been busy
 
Coming from a previous FX 8350 owner, it will almost certainly be the CPU bottlenecking the GPU.

My 8350 even bottlenecked a GTX 660 on a previous build.

The reason you got better results from the previous card is because the new one is more powerful so the CPU has to do more to keep up
 
Have you wiped all your old AMD drivers?.

Get Driver Sweeper/Display Driver Unistaller off GURU and wipe it all then reinstall the NVdia driver again.

If you GPU is at 99% usage (GPU's never really show as 100%) then your maxing out your card. Do remember that AMD and Nvidia cards perform different on different games.

If it was a CPU bottleneck your CPU would be in the 85%+ usage area and you would see your GPU usage drop as the CPU can not throw the data over to the GPU fast enough for the card to smoothly push out its max frames for the set video settings.

You mentioned League of Legends as your go to game. This game is know to have frame rate issues with some cards and drivers, most top down DOTA style kind of games do drop frames as there is a lot going on and usually it is down to the server being on the slow side as its being swamped.

Its like if you went and played World of Warcraft and walked into a busy trade area then you saw your frames drop down but your GPU is getting low usage. If you got that card and went and played an older single player game that was not CPU intensive then your GPU would shoot out its full capable amount of frames. It may also be to do with some games not being optimised for CPU's higher than dual core like the Torchlight 1 & 2 and then Fallout 3 (my GTX 1080 can only pull 90fps some older games due to their engines limits)
 


Try turning up your settings. I suspect you were seeing closer to 100% usage on a 7850 to achieve the same.

While in theory, a stronger card should show improvements all else being equal, it's harder to reflect in the low settings.
There;s only so much to render with low, a clearly a 980TI can do it with ~30% of it's available resources

A 980TI (nor 8350 really) are being pushed very hard in very low-medium settings. Try cracking your settings up to high/very high.

With the 7850, high/very high would likely see a sizeable FPS reduction, I would be surprised if you can't achieve 80+FPS from a 980TI in the high/very high preset.
 


I'm in no situation to upgrade sadly, but it is most definitely on my to do list in terms of this rig
 


Already done this :/
 


I'll give that a go; I'll post back
 
 
Yunimo Your newly acquired 980 Ti is certainly bottlenecked by your existing FX 8350.

Solution: Sell your 8350 / MB / RAM as combo on Ebay. You'll get about $250 on Ebay for it. Then purchase a i5-7500 / B250 chipset MB / DDR4 RAM from NewEgg or Amazon.

Addendum: And sell your HD 7850 while you're at it.

Also, if you have a retail Windows license than you should be okay. But if it's OEM, then you can purchase a new key cheaply off this site: https://www.kinguin.net/category/15019/windows/
 
There will be some bottleneck, but it shouldn't be too extreme.

@rcald2000, I wouldn't recommending that site. While it's only a marketplace, the potential to receive a code from a seller there that either won't activate, or gets blacklisted in future is high.
 


That's a definite bottleneck - surprised no one was able to confirm this before me. If your GPU usage is not at 100% fixed, it simply means your CPU is bottlenecking the GPU. You'll either have to overclock the CPU, or increase the settings. Overclocking will increase FPS, but not enough to make the GPU usage 100%, there will still be bottleneck. Increasing the detail settings will increase GPU usage, but it still won't get you higher FPS - you need more CPU power for more FPS. You'll have to either be satisfied with current FPS, or OC your CPU, or upgrade. Since you can't upgrade, and Ocing will only help to some extent, the only option you have left is turn up the settings till the GPU is being used 100%(or as high as it can go), and just bear with whatever FPS you are getting. Sorry :) .
 
@Shektron, except the OP confirmed "GPU usage is around 30% same with CPU.".... too me that means CPU usage is also around ~30%.

So, while I don''t disagree that there's some aspect of a bottleneck, it's not like the CPU is hitting 100% usage, holding the GPU....... not yet, anyway.
 
CPU usages do NOT have to be at 100% to confirm CPU bottlenecks - this is because games do not always use all the cores of a CPU, so the percentage will not be a 100%. Let's say you have an 8 core chip, and a game uses 4 of them. So the usage will be 50% or around 50%. But, if the 4 cores are slow(which is the case here), then though the usage is only 50%, the CPU will still bottleneck. Seems to me you don't know much about the relation between usage and bottlenecking...
 


in games that dont use all cores, like online ones

he would never hit 100% usage

still a bottleneck, because of the very weak per core performance of the FX
 
Again, I didn't say it *wasn't* a bottleneck.

Poorly worded on my part re: 100% CPU usage. I meant 100% "per core/thread". 30% of an 8core/thread CPU would be maxing 2.4 cores at "max", so probably 2 maxing for the game itself. IIRC, LoL can utilize 4 cores/threads, so the FX8350 should be able to see a little over 50% utilization.

Again, not saying it's not a bottleneck on an IPC basis. Stronger single/"main" core performance would show gains..... but I don't believe the 8350 is maxing out what it can do, even at stock speeds currently.

 
Oh no, I made a boo-boo 🙁

https://youtu.be/tAf8jM9ehZI

Looks like we do have a problem here. Something's not right with your FPS, and my guess is it's the CPU(just stating the obvious here). Please report temperatures, and run some stress tests, maybe Prime95 and then AIDA64. Also run memtest, though RAM issues are usually more...extreme. Also report GPU temperatures. All of these are to be noted when running the stress test, not the game.

Also, this may sound silly, but I hope you don't have Vsync on.
 
As per the notes section of those results: "֍ Benchmark Settings
→ Max Settings"

That's what I suggested earlier. Sounds like there's some kind of FPS 'cap' in place on low setitngs, which would explain virtually no difference between a 7850 that was (probably) maxed, and a 980TI barely being worked at all.

The OP did return at say there was no change, but didn't definitively state whether that was at high settings.
 




At max settings I'm still seeing around 90fps~

As for the stress test (AIDA64); what tests are you looking for me to run exactly?
 


Ran both tests for about 2 hours, best I could do; temps are pretty much identical in both.
WT98kZW.png