Finding out whether your FPS are CPU or graphics card limited

[dud]

Do just take the "unreal is cpu intensive" unmeasured blah-blah statements.

Do this:

• get framerate in a reproducible way that you like
• go into the BIOS and downclock your CPU by 1/3rd, repeat benchmark
• download an overclocking utility for your graphics card and use it to downclock your GPU by 1/3rd, repeat benchmark
• for added fidelity, do same for:
• RAM
• graphic card RAM

I don't have time right now but I'll post mine later.

We need a reference thread about hardware-to-fps anyway.

 

[farjam]

If it's single CPU thread that becomes a bottleneck you won't see that.
It's simple, if it's not GPU, it's CPU. I think you are doing fine with 100fps, add VSync or frame limiter and CPU usage will drop.

I'm ok with cpu load as the temps are controlled , but i like to see my GPU using its full potential (like up to 95% in any other games/software/benchmark i'm using) to be able to set the frame for example on 120 but it ends up on 100fps in crowded scenes and 120 when i'm alone on the track with only 60-70% gpu usage

 

[RasmusP]

I'm ok with cpu load as the temps are controlled , but i like to see my GPU using its full potential (like up to 95% in any other games/software/benchmark i'm using) to be able to set the frame for example on 120 but it ends up on 100fps in crowded scenes and 120 when i'm alone on the track with only 60-70% gpu usage

Which cpu do you have? It's definitely the cpu limit. It sucks, I know...

 

[farjam]

Which cpu do you have? It's definitely the cpu limit. It sucks, I know...

It's 8700k , it's so weird that i have this issue only in a game that i love

 

[RasmusP]

It's 8700k , it's so weird that i have this issue only in a game that i love

These 100 fps: are they at 100 all the time or going higher/lower while driving?
Is it with you alone on track or is that Screenshot from a race with AI?

With an 8700k the fps are too low I think!
I'd suspect Riva tuner limiter or ingame limiter active. Or your monitor is set to 100hz and it's vsync being on (windows aero can activate it sometimes when not using full screen.

 

[farjam]

These 100 fps: are they at 100 all the time or going higher/lower while driving?
Is it with you alone on track or is that Screenshot from a race with AI?

With an 8700k the fps are too low I think!
I'd suspect Riva tuner limiter or ingame limiter active. Or your monitor is set to 100hz and it's vsync being on (windows aero can activate it sometimes when not using full screen.

It's around 90-105 fps when the ai cars are near me in sunny weather and 85-95 fps in storm/heavy rain , but when i'm alone it will be easily 120+ fps with the same gpu usage (edit: now i can get 190-220FPS when i'm alone in practice just with shadow on low and foliage on low but the race FPS is still almost the same)
Rivatuner fps limiter is deactivated , vsync is double off (in nvidia panel and in game)

also i tried to force gpu to have more power target and it's just make it to utilize gpu 6-7% more

 

[RasmusP]

It's around 90-105 fps when the ai cars are near me in sunny weather and 85-95 fps in storm/heavy rain , but when i'm alone it will be easily 120+ fps with the same gpu usage
Rivatuner fps limiter is deactivated , vsync is double off (in nvidia panel and in game)

also i tried to force gpu to have more power target and it's just make it to utilize gpu 6-7% more

The FPS with a full grid are totally normal. If I'd try this with my 2600k it will become quite a slideshow...
I guess while it is "okay" regarding optimization without AIs, the AI calculations will get some more work in the future. I doubt that Kunos intends to drop below 90 fps on the best Gaming-CPU available.

 

[farjam]

The FPS with a full grid are totally normal. If I'd try this with my 2600k it will become quite a slideshow...
I guess while it is "okay" regarding optimization without AIs, the AI calculations will get some more work in the future. I doubt that Kunos intends to drop below 90 fps on the best Gaming-CPU available.

i have added another info to the last post
" (edit: now i can get 190-220 when i'm alone in practice just with shadow on low and foliage on low but the race FPS is still almost the same)"
it makes it more confusing, almost 100FPS is the difference, 190 to 220 in practice and 90-105 in race when the ai are around me , and my face is like a question mark right now
and about optimization , i had a similar issue with a ubisoft game , Assassins creed origins , obviously different engine, but the problem was exactly the same and after a month or two the gpu usage was normal after some patchs , i hope it's just a optimization problem...

 

[RasmusP]

The physics are really complex and the AI runs on the same, not on simplified physics!
So 5 AI cars mean 6x your own car with full physics + 5x driving the car instead of you giving the inputs.
Thats why the cpu load increases so much.

It might be that when you're alone on track the graphics card becomes the limit so turning down settings will boost fps. With AI it changes to the cpu limiting and the graphic settings don't really matter.

 

[Neilski]

@Jempy I doubt it will help your current performance but I do suspect from what you're saying that your stock cooler isn't working as well as it should. You could try gently wiggling it to see if it's fully engaged on all four pins. The best OC you're likely to get though, even with a better cooler, is ~ 20% and you might find it's under 15%.
(Btw, the 4670K is more like 5.5 years old )

The problem is that nvidia messed everything up by coming up with an expensive proprietary module.

Agreed. It's really sad that they did this. And I guess I probably have a higher tolerance for fps variations, having had to put up with borderline hardware for many years. (I recently decided to splash out a wee bit more, given that I normally hang onto whatever I buy for ages.)

 

[Pawel Ziolkowski]

(Btw, the 4670K is more like 5.5 years old )

Oi don't you talk down the 4670k! i'm getting 80fps (cpu limited) where farjam had 100 on his 8700K

5 years and at probably twice the price at launch... yeah CPUs are really moving forward aren't they .

 

[Jempy]

@Neilski

Thanks for your answer.
Better than check, I will probably change the cooler .... that will give a better result and more safety finally.

I checked for the socket of my mb ... and it's well what I thought: this is a 1150 one ... so change CPU should need change motherboard as well .... so better stay with the 4670k for now and wait for better times ( in € ! ).

The CPU might be used at 100 % for 2 hours .... and as long as temperatures are fine ( after replacing the cooler ) and fps are +- 60 ... I think there will be no problem and I can already be happy with that.

EDIT and of course thanks to Martin and Rasmus for their answers too.

 

[Neilski]

so better stay with the 4670k

Well, if you were able to get your hands on a 4770K (or even 4790K) for say €100 then it might be a simple and attractive upgrade - no need to change anything else. However, a quick bit of googling suggests the prices for those i7s are still stupidly high. Weird! I'd been thinking the prices would have dropped way down by now.

 

[Neilski]

Oi don't you talk down the 4670k!

Old doesn't mean bad - not talking it down at all!
Chips have barely speeded up in the last 10 years...

 

[metaDp]

I think we should put in place a test protocol.

For exemple a 5 minute race at 21h in the sunset, mid rain with AI. The same level of AI for anyone (maybe 50 agression, 95 difficulty) trying to get a 100% accurate comparable scene.

Then use a benchmark tool to get the MIN, AVG and MAX FPS.

I will post later something like this with my setup (I7 4770k and GTX 980ti) with stock CPU vs some OC, I'm actually testing the stability for a 4,3Ghz. This night with a 4,2 it felt better in the minimum FPS, but it will be sure when I will run my idea of protocol, I will have concret view of the gain.

 

[Paul Bennett]

Downloading soon , running a Xeon 1220v3 ( i5 ) with a 1060 6gb - 16gb ram.
Far as I have read, set FXAA an motion blur off should be good to go ?.. Will set to medium just for starters

 

[Opus131]

Far as I have read, set FXAA an motion blur off should be good to go ?

You should set motion blur off just because it is an effect that makes no sense, never mind gaining performance. Depth of field is also another effect that needs to be instantly disabled in all games but there's no option for that here.

As for FXAA, maybe there's something wrong with my set up but it doesn't do anything. Neither does raising the resolution scale to 120. Jaggies everywhere in both cases. Only thing that removes jagged lines is temporal AA, but that makes everything blurry.

Not happy about this. I have to see if i can force the AA of the card through nvidia inspector but i have a feeling it won't work well.

Jagged lines in the distance was one of my biggest pet peeves in the original rFactor i can't believe in 2018 i'm still fighting this nonsense.

 

[T_rac3]

Hi, this is my spec: FX8300 , GTX1070TI , 24 GB RAM
AI at 90 and aggression at 70, 9 cars with me 10.
And some screenshots with video settings and FPS, both wet and dry

 

[metaDp]

Here are my result.

The setup:
I7 4770k/2x4Go 2133mhz
GTX 980ti

Testing protocol:

I'm trying to have the same scene on each test to be sure to compare apple with apple and orange with orange. I choose to start in the midfield with the rain with the idea to stress the system to get lowest FPS/worst condition "possible". The green lights comes and I just follow the other car until Turn 1, then gently I pass the other car one by one until lap 2. A little gap is formed and I catch the leading pack during the second lap. And follow them for the third and last lap.

Using fraps from the start to the end to get the Minimum/Average and Maximum FPS.
Result will be expressed like this: CPU -Stock or OC-, FPS min/avg/max.

The performance:

I have run 4 test, 2 in Epic and Medium graphic settings/ with and without CPU overclock.

Epic graphics settings

CPU Stock: 21/59/77
CPU @4,3: 24/56/79

Medium graphics settings

CPU Stock: 48/64/77
CPU @4,3: 51/71/85

My conclusion:

During the test, I kept an eye on my CPU/GPU Usage on my second screen. During the 4 tests, the CPU usage was really high on the all 8 threads. (I would say between 85/95%).

On the other hand, the interesting things comes from my GPU Usage:
Without OC (in Epic and Medium settings) the usage were from 45 to 65%.
With OC (in Epic and Medium settings) the usage were from 80 to 100%.

So the game is using a lot the CPU for the physics calculations I would say, I'm not a pro in this work ^^.



Now I will probably do the same thing while being alone on the track, but probably only one lap, I guess it will be enough. And if guys like what I have done, I will probably do the same during each phase of the Early Acces, to see if improvement are made and open my own thread for that ^^.

 

[Opus131]

Hi, this is my spec: FX8300 , GTX1070TI , 24 GB RAM
AI at 90 and aggression at 70, 9 cars with me 10.
And some screenshots with video settings and FPS, both wet and dry
View attachment 269867
View attachment 269868
View attachment 269869
View attachment 269870
View attachment 269871

Hah, another FX owner. Yeah, we are screwed:

Time to upgrade to a Ryzen, except good luck replacing all that RAM. Prices are through the roof

 

[Case_]

Well, if I was to upgrade to some more current i7, it would cost me some 600 euro. Even getting more up to date i5 setup would be fairly close to that, so...I feel for you.