The ACC Benchmark Thread

theres no point because everyone is using their own setttings anyway. You cant correlate anything unless you have the same exact setup as that person does.

 

2018-11-22 09:13:54 - AC2-Win64-Shipping
Frames: 9462 - Time: 183938ms - Avg: 51.441 - Min: 29 - Max: 80

i7-3930K/48GB/RX580-8GB

 

That´s true.

@LeDude83 Maybe we should do a second benchmarklist with fixed video settings, if it isn´t too much work?

Edit: I mean a second list beside the new one

 

Let me know.... it's the only leaderboard I seem to be near the top!!

 

Post 1 is up-to-date with instructions on the new, faster and even more reliable benchmark method.

I'll start:

2018-11-22 20:45:05 - AC2-Win64-Shipping
Frames: 3281 - Time: 60000ms - Avg: 54.683 - Min: 43 - Max: 74



View/Camera: dash
FOV: 34

 

i7.8700k
gtx1080ti
16gb
no overclock
win10

view cockpit (not dash, not helmet)
height 7
distance 4
fov 50
pitch 0

2018-11-22 21:47:38 - AC2-Win64-Shipping
Frames: 3487 - Time: 60000ms - Avg: 58.117 - Min: 45 - Max: 76

 

My advice is to not use fraps and also remove cpu from the equation.
Choose absolutely maximum graphic settings, single player and report fps while standing in pits at specific camera and fov.
Otherwise too many parameters.

Στάλθηκε από το m2 note μου χρησιμοποιώντας Tapatalk

 

It's a system benchmark so removing CPU from the equation is absolutely not what I want. The game will be run on a lot of different systems/settings so that is also the goal for the benchmark.

A standstill screen like you suggest is a poor choice IMO. You leave out moving objects, shadows, reflections, dust and water spray - far from a realistic use case.
Like I said earlier - if you want that very much narrowed-down comparison, pick one system from the benchmark list and try to fix all variables you can with the respective person.

This thread is (also) to find out optimum settings for different systems, so fixing all settings per se would be a fail.

And why not use FRAPS?

Edit: But I can see the wish for a GPU benchmark. It should be a 60 second benchmark in hotlap mode. Can be the same settings as with the formation lap benchmark. If you want, you can fix the settings and start benchmarking - I'll add it to post 1 later and make yet another list. Resolution should be 1920*1080 I guess as that's what everyone can run.

 

I can't read the numbers mate. Please write them down for me.

 

2560x1440
Full Screen Enabled
vsync disabled
Resolution scale 110%
view distance epic
shadows high
anti aliasing epic
temporal
effects epic
post processing epic
sharpness 150%
foliage mid
texture epic
frame limit off
mirror quality epic
opponents all
motion blur disabled
saturation 90

 

Here is my result using the new benchmark details from post 1.



2018-11-23 10:12:32 - AC2-Win64-Shipping
Frames: 2703 - Time: 60000ms - Avg: 45.050 - Min: 31 - Max: 62

Dash View FOV 34

Intel i5 4670K @ 4.2
Nvidia GTX 970
8 GB RAM
Win 10 Pro

 

I've now spent some time playing around with the settings, and I've managed to find a combination which will allow me to hit a consistant 60 FPS (day and clear conditions) with a full grid. I hope it helps others with a similar setup as my system is now a few years old, but is capable of running ACC with a few concessions.



I've also enabled 4x antialiasing and anisotropic filtering in the NV panel and left all the other settings on default. The mirrors I'm not worried about as I use the proximity arrows and left/right views to determine what is happening around me.

Thanks to LeDude83 for starting this thread, it encouraged me to really try and figure out where the problems were with my settings.

Intel i5 4670K @ 4.2
Nvidia GTX 970
8 GB RAM
Win 10 Pro
OS/Steam Library SSD

 

2018-11-23 19:15:34 - AC2-Win64-Shipping
Frames: 2928 - Time: 60000ms - Avg: 48.800 - Min: 36 - Max: 70

Intel i5 3750K @ 4.4
Nvidia GTX 770
16 GB RAM
Win 10

Cokpit view (FOV 45)

 

I'll just chime in about 2 settings:
1) Benching with mirrors OFF: even if it gives a huge fps advantage benching, it shouldn't be used in a race in a sim were you are expected to race clean, so it distorts results. Low looks terrible, has cars disappearing, but at least it can be useful when defending most times.
2) Number of opponents: Special events are using 20 (even the last Special Events released on AC where set to 20 too). This is a lot, and very taxing, and it's a good reason why a gen 4 i5 is now on the lowest specs for the game. IMHO, going under 20 isn't a good measure on how the game was intended to be played, or how the same combo will behave on default special events - even if it's a necessity on older hardware.
Edit: The problem with this (benching with 20 AI), is the way it distorts GPU comparisons under a huge CPU bottleneck, so a second set with light CPU usage (practice with fixed high to epic details at 1080p so all can use it) could be used to directly compare GPUs, and how much only the CPU render thread limits avg/max fps (true IPC on single thread, not number of cores for physics threads that are stressed under AI). Other than that, the current benchmark is OK to tell how user are managing their custom races, so don't pay much attention to this post.

 

View: Cockpit ? - the first one out of the three in car views - perhaps should clarify what to call each in the OP. [cockpit - dash - helmet ?]
FOV: 45

Up to date system specs in signature

Ran test 3 times:

2018-11-24 13:57:47 - AC2-Win64-Shipping
Frames: 2338 - Time: 60000ms - Avg: 38.967 - Min: 31 - Max: 51

2018-11-24 13:59:33 - AC2-Win64-Shipping
Frames: 2361 - Time: 60000ms - Avg: 39.350 - Min: 35 - Max: 43

2018-11-24 14:00:59 - AC2-Win64-Shipping
Frames: 2383 - Time: 60000ms - Avg: 39.717 - Min: 35 - Max: 49

Good work Mr Dude

Edit: after some mentioning vsync=on gave some improvement, I turned on vsync and ran again:

2018-11-24 14:12:36 - AC2-Win64-Shipping
Frames: 2344 - Time: 60000ms - Avg: 39.067 - Min: 35 - Max: 43

2018-11-24 14:13:59 - AC2-Win64-Shipping
Frames: 2418 - Time: 60000ms - Avg: 40.300 - Min: 35 - Max: 49

2018-11-24 14:15:22 - AC2-Win64-Shipping
Frames: 2429 - Time: 60000ms - Avg: 40.483 - Min: 33 - Max: 49

strange how the results improve slightly each time I run the test again, might have to run 5* next time and see if it stops improving at some point and levels out

Edit again: looked at the spreadsheet and filtered by people with my GPU and CPU, then copied the settings from best result and ran again:



2018-11-24 14:27:59 - AC2-Win64-Shipping
Frames: 3631 - Time: 60000ms - Avg: 60.517 - Min: 52 - Max: 79

2018-11-24 14:29:36 - AC2-Win64-Shipping
Frames: 3583 - Time: 60000ms - Avg: 59.717 - Min: 52 - Max: 77

2018-11-24 14:30:51 - AC2-Win64-Shipping
Frames: 3637 - Time: 60000ms - Avg: 60.617 - Min: 53 - Max: 77

2018-11-24 14:32:11 - AC2-Win64-Shipping
Frames: 3715 - Time: 60000ms - Avg: 61.917 - Min: 55 - Max: 79

2018-11-24 14:33:24 - AC2-Win64-Shipping
Frames: 3688 - Time: 60000ms - Avg: 61.467 - Min: 53 - Max: 79

So, quite interestingly the result number 4 was the best and then it started to drop back down.

 

With the old settings from here...

https://www.assettocorsa.net/forum/...cc-benchmark-thread.53257/page-5#post-1010511

...I set the driver profile in Radeon settings to Turbo. This gained me 3 FPS AVG:

Standard driver profile:

2018-11-22 20:45:05 - AC2-Win64-Shipping
Frames: 3281 - Time: 60000ms - Avg: 54.683 - Min: 43 - Max: 74


Turbo driver profile:

2018-11-24 15:09:31 - AC2-Win64-Shipping
Frames: 3509 - Time: 60000ms - Avg: 58.483 - Min: 47 - Max: 80

So I wanted to "spend" that performance on some higher settings and changed post processing and foliage from high to epic. This had very little impact on the FPS - just 1 AVG.
So, my latest settings and result:



2018-11-24 15:20:44 - AC2-Win64-Shipping
Frames: 3433 - Time: 60000ms - Avg: 57.217 - Min: 45 - Max: 81

That's a pretty good sweetspot for me. The image looks very good and the lowest FPS drop is just on the lower limit of my FreeSync monitor.

 

5760x1080 FOV 30 (prefered resolution)
2018-11-24 18:26:47 - AC2-Win64-Shipping
Frames: 2772 - Time: 60000ms - Avg: 46.200 - Min: 37 - Max: 64

1920x1080 FOV 30
2018-11-24 18:30:17 - AC2-Win64-Shipping
Frames: 3674 - Time: 60000ms - Avg: 61.233 - Min: 51 - Max: 74

AMD FX8350
GTX 980ti
16GB Ram DDR3

My old benchmark:
Post #43

 

i7 3770 3.4-3.9(turbo)
Asus R9 290x DirectCU II 4gb ddr5
16 gb ram 1600MHz
SSD

2018-11-25 13:41:56 - AC2-Win64-Shipping
Frames: 2454 - Time: 60000ms - Avg: 40.900 - Min: 34 - Max: 51

Camera: Dashboard

 

VR run, daylight not dark. 20 competitors and me at back.

i9 9900K (OC 5.1) RTX 8020 ti

Frames Min Max Avg
5388 83 91 89.8

I just built this computer to be able to run ACC in VR at full 90 fps. Needless to say I am gratified to see this result.

 

Good to see those result, but this PC is one the 0,00001% who can run it on the old CV1.
If you go further with Vive Pro, Samsung Ody, Pimax 5k+ then even this fully overpriced PC wont make it ^^

Could you start on the 10. or 11. place?
This seems harder for the CPU than on the last place IMO.