Benchmark results, food for thought

Ahh that makes sense, we are in full agreement with regards to the motion blur, usually not something I enable in games either.

Yeah the x64 app thing is a bit of a nuisance. For the record, I would leave AC on your mechanical drive, I didn't see any difference in performance by putting it on an SSD aside from loading times. Now that we have this handy benchmark, I should try it again to be sure though.
Steam library tool makes quick work of changing a directory without making steam go crazy.

Much appreciated! I think your high logical core count is the only reason your CPU percentage is low then, it's just resting very comfortably Also, thanks for saying what I should have said before Jason, I went off on a tangent and didn't explain properly when I asked him to test without the HQMR!

"also of note, the MIN fps certainly does not do any justice here, my FPS after the initial start, merely 2 seconds in was 103 and it never dropped below that"
Ah yes, this makes sense, kinda confirms what I was thinking with the benchmark needing a tiny time delay before it begins logging, otherwise the minimum isn't going to provide an accurate value in terms of how the game is playing.

Performance is looking very nice Jason, excellent work. I think your findings will prove very useful for many, including Kunos. Nailed down the culprit settings very comprehensively.

 

I play Assetto Corsa thru my laptop's HDMI out to my 46" LCD TV. I keep the laptop next to me with Chrome open and other stuff running. I always play with everything maxed out:

AC VERSION: 1.3.4 (x64)
POINTS: 10036
FPS: AVG=68 MIN=21 MAX=132 VARIANCE=2 CPU=42%

LOADING TIME: 14s
GPU: NVIDIA GeForce GTX 965M (1920x1080)
OS-Version: 6.2.9200 () 0x300-0x1
CPU CORES: 8
FULLSCREEN: ON
AA:4X AF:16X SHDW:4096 BLUR:0
WORLD DETAIL: 5 SMOKE:5
PP: QLT:5 HDR:1 FXAA:1 GLR:5 DOF:5 RAYS:1 HEAT:1


I've noticed that some people turn the SMOKE setting down to 3, so I decided to run another Benchmark with my SMOKE set to 3. In my opinion, the improvement was so small I might as well leave it maxed out!

AC VERSION: 1.3.4 (x64)
POINTS: 10223
FPS: AVG=69 MIN=22 MAX=153 VARIANCE=1 CPU=43%

LOADING TIME: 12s
GPU: NVIDIA GeForce GTX 965M (1920x1080)
OS-Version: 6.2.9200 () 0x300-0x1
CPU CORES: 8
FULLSCREEN: ON
AA:4X AF:16X SHDW:4096 BLUR:0
WORLD DETAIL: 5 SMOKE:3
PP: QLT:5 HDR:1 FXAA:1 GLR:5 DOF:5 RAYS:1 HEAT:1

As you can see, the MAX Frame Rate jumped 20fps, but the AVG Frame Rate increased by 1fps. Nothing to write home about.

 

But smoke on or off. might be important at tire burnouts or spins of course (such situations). In the benchmark scene there is not much smoke, so it has almost no impact of course there.

 

Btw.: Why is the used reflection quality, reflection rendering frequency & mirror res. missing in the bench logs ? That are also important values to know.. .
So the whole settings from the "Reflections" Tab is missing in the results atm. . Would appreciate, having this corrected / added in a future AC update.

I used 'mirror res.: normal' / 'reflection quality: low' / 'reflection rendering frequency: static' in my posted benachmarks.


Found out that turning off the mirrors increases the avg FPS to 86 for me, but driving without any mirrors.. .
Glad to see that 'reflection quality: low' (on static) results in the same avg FPS as 'reflection quality: none'.
So you can use reflections (on low) without having any performance impact, that's a good thing ! On the next higher setting 'medium' the FPS then starts to drop of course.

 

Just changed Reflection quality from Medium to High setting and look what happens to CPU %

AC VERSION: 1.3.4 (x64)
POINTS: 16259
FPS: AVG=110 MIN=27 MAX=186 VARIANCE=3 CPU=70%

LOADING TIME: 12s
GPU: AMD Radeon HD 7900 Series (1920x1080)
OS-Version: 6.2.9200 () 0x100-0x1
CPU CORES: 8
FULLSCREEN: ON
AA:4X AF:16X SHDW:2048 BLUR:0
WORLD DETAIL: 5 SMOKE:3
PP: QLT:3 HDR:0 FXAA:0 GLR:3 DOF:3 RAYS:1 HEAT:1

Reflection quality Medium

AC VERSION: 1.3.4 (x64)
POINTS: 14220
FPS: AVG=97 MIN=26 MAX=164 VARIANCE=0 CPU=59%

LOADING TIME: 12s
GPU: AMD Radeon HD 7900 Series (1920x1080)
OS-Version: 6.2.9200 () 0x100-0x1
CPU CORES: 8
FULLSCREEN: ON
AA:4X AF:16X SHDW:2048 BLUR:0
WORLD DETAIL: 5 SMOKE:3
PP: QLT:3 HDR:0 FXAA:0 GLR:3 DOF:3 RAYS:1 HEAT:1

Reflection Quality High

 

Same for me, once I set reflection quality at high (and mirrors also), I (surprisingly) got a lower CPU usage of 57% (instead 70%-72% with lower reflection quality setting).

But maybe it's simple to explain: More FPS on lower settings = CPU has more to do in the same time, somehow because of the higher framerate.

 

Main PC Results

AC VERSION: 1.3.4 (x64)
POINTS: 14987
FPS: AVG=102 MIN=60 MAX=121 VARIANCE=0 CPU=59%

LOADING TIME: 13s
GPU: NVIDIA GeForce GTX 760 (1920x1080)
OS-Version: 6.2.9200 () 0x100-0x1
CPU CORES: 4
FULLSCREEN: ON
AA:4X AF:8X SHDW:1024 BLUR:0
WORLD DETAIL: 5 SMOKE:3
PP: QLT:3 HDR:1 FXAA:0 GLR:3 DOF:3 RAYS:1 HEAT:1

 

That would be the main reason for that, also one thing I think should be changed in the benchmark is for me it seems as though the min fps is as the bench is loading. I have seen min fps much lower even when the bench never drops. I did a run capped at 61 with avg 61 yet min was about 30. I never so the drop in the actual run. I will do a couple runs and compare it the the msi trace to see if thats the case as it kind of makes the min figure a bit usueless if thats the case.

 

AC VERSION: 1.3.4 (x64)
POINTS: 7108
FPS: AVG=48 MIN=32 MAX=96 VARIANCE=0 CPU=29%

LOADING TIME: 12s
GPU: NVIDIA GeForce GTX 780 Ti (5120x1080)
OS-Version: 6.2.9200 () 0x100-0x1
CPU CORES: 8
FULLSCREEN: OFF
AA:8X AF:16X SHDW:4096 BLUR:0
WORLD DETAIL: 5 SMOKE:5
PP: QLT:5 HDR:1 FXAA:1 GLR:5 DOF:5 RAYS:1 HEAT:1


Im satisfied.

 

AC VERSION: 1.3.4 (x86)
POINTS: 8792
FPS: AVG=60 MIN=21 MAX=110 VARIANCE=4 CPU=62%

LOADING TIME: 70s
GPU: AMD Radeon R9 200 Series (1920x1080)
OS-Version: 6.1.7600 () 0x100-0x1
CPU CORES: 4
FULLSCREEN: ON
AA:4X AF:16X SHDW:2048 BLUR:0
WORLD DETAIL: 4 SMOKE:0
PP: QLT:5 HDR:1 FXAA:0 GLR:1 DOF:5 RAYS:0 HEAT:0

its enough for me

 

Pretty interesting results, and might actually be beneficial, despite my first reaction...



AC VERSION: 1.3.4 (x64)
POINTS: 16187
FPS: AVG=110 MIN=23 MAX=264 VARIANCE=7 CPU=45%

LOADING TIME: 39s
GPU: NVIDIA GeForce GTX 970 (1920x1080)
OS-Version: 6.1.7601 (Service Pack 1) 0x100-0x1
CPU CORES: 4
FULLSCREEN: ON
AA:4X AF:16X SHDW:2048 BLUR:0
WORLD DETAIL: 5 SMOKE:3
PP: QLT:5 HDR:1 FXAA:1 GLR:5 DOF:5 RAYS:1 HEAT:1

You guys with 12-13s load times, are you running on SSD?

 

Yup

 

no

 

My AC is on a regular HDD, and a slow one at that; it takes about 30 seconds to load the first time but then if I change a setting and load it again it only takes about 15 seconds.

 

I think loading times seem to have quite a bit to do with X86 vs X64 too

 

Could it be that the benchmark is stored in RAM after the first run?

 

I did do some quite extensive benchmarks on my setup and came to pretty much the same conclusions as I did simply benchmarking using afterburner in the past.

Here are the regular ingame settings I run but also I normally cap at 61fps and it pretty much never drops a frame at these settings.

AC VERSION: 1.3.4 (x64)
POINTS: 10156
FPS: AVG=69 MIN=22 MAX=156 VARIANCE=2 CPU=55%

LOADING TIME: 12s
GPU: NVIDIA GeForce GTX TITAN (8060x1440)
OS-Version: 6.2.9200 () 0x100-0x1
CPU CORES: 8
FULLSCREEN: ON
AA:4X AF:16X SHDW:4096 BLUR:0
WORLD DETAIL: 5 SMOKE:3
PP: QLT:5 HDR:1 FXAA:0 GLR:4 DOF:4 RAYS:1 HEAT:1

Now the min frames thing is clearly a bit bugged in the benchmark as in the test my fps never dips that low other then when it starts, I think the benchmark should load the grid then wait a few seconds maybe 5 or so for the race to start. As it is it loads pretty much on the green and the fps is just starting to build up at probably the heaviest time fps wise back of grid with wheelspin. Below are multiple benchmarks at various settings and it clearly shows no drops to the same lows once the bench has started that the results are indicating.



Here are all the Benchmarks I ran with afterburner showing some key info, note fps the min fps is never dipping from the time the fps builds which is just after the benchmark starts. CPU usage is also marked higher in the benchmark then afterburner has it but when looking at each core they are all getting a bit of use.



I have also included a rar file of the 11 benchmark runs shown in the afterburner screen shots.

 

Dont know what settings we're supposed to use, so just did the settings that I play with.

AC VERSION: 1.3.4 (x64)
POINTS: 36425
FPS: AVG=248 MIN=54 MAX=355 VARIANCE=48 CPU=82%

LOADING TIME: 9s
GPU: NVIDIA GeForce GTX 980 Ti (1920x1200)
OS-Version: 6.2.9200 () 0x100-0x1
CPU CORES: 8
FULLSCREEN: ON
AA:4X AF:8X SHDW:2048 BLUR:0
WORLD DETAIL: 5 SMOKE:3
PP: QLT:5 HDR:1 FXAA:1 GLR:3 DOF:3 RAYS:1 HEAT:1

 

I get pretty much the same load times each run on an SSD, if it was storing in ram I think it would load much quicker then it does.

 

AC VERSION: 1.3.4 (x64)
POINTS: 12036
FPS: AVG=82 MIN=62 MAX=92 VARIANCE=0 CPU=37%

LOADING TIME: 11s
GPU: NVIDIA GeForce GTX 780 (2715x1527)
OS-Version: 6.2.9200 () 0x100-0x1
CPU CORES: 8
FULLSCREEN: ON
AA:4X AF:16X SHDW:2048 BLUR:0
WORLD DETAIL: 5 SMOKE:2
PP: QLT:5 HDR:1 FXAA:0 GLR:5 DOF:5 RAYS:1 HEAT:0