The ACC GPU Benchmark Thread

Thank you.

Doesn't look too promising, but I guess it will have to do.
I think higher refresh rate and more performance (paired with freesync) on the other games will be worth it, especially since ACC currently sits behind 2 other games I play (way) more.

 

For ACC AMD is a totally unreasonable choice right now. But well, you were informed ;-)

Make sure to check the actual performance in your favourite games. Also, get the 590...not much more expensive than the 580, slightly better performance.

 

Well, it would be 170€ for the 580 8GB vs. 240€(+) for the 590, so IMO the 580 is the better deal considering the 590 is 10-15% faster (at best).

 

or the new 1660 TI ???

 

Thanks for the help and suggestions, but as I said before, I'm not going to buy anything from Nvidia any time soon.

 

Before we can get back to GPU testing, Kunos needs to do some Homework it seems. The results with maxed out mirror are crazy:

2019-03-04 21:50:53 - AC2-Win64-Shipping
Frames: 1941 - Time: 60000ms - Avg: 32.350 - Min: 23 - Max: 52

Even with mirror Q and Res to low, it doesn't look good:

2019-03-04 23:09:03 - AC2-Win64-Shipping
Frames: 4508 - Time: 60000ms - Avg: 75.133 - Min: 54 - Max: 121

I previously had 78 FPS with everything set to epic!

When deactivating the mirror:

2019-03-04 23:10:48 - AC2-Win64-Shipping
Frames: 5749 - Time: 60000ms - Avg: 95.817 - Min: 68 - Max: 166

@pankykapus are there any "low hanging fruits" to improve this issue? Please note that in this benchmark, the mirrors aren't even visible on the screen - it seems unreasonable that they take this much of a toll on FPS.

 

2019-03-06 14:33:49 - AC2-Win64-Shipping
Frames: 3859 - Time: 60000ms - Avg: 64.317 - Min: 51 - Max: 89

 

My friend, you're running custom settings. Please post the results here:

https://www.assettocorsa.net/forum/index.php?threads/the-acc-benchmark-thread.53257/

The GPU benchmark (this thread here) will have an instruction update on post 1, very soon.

 

Little spoiler: I already ran the test with settings that are equal to the ones we had before (clarified this with pankykapus). The AVG FPS are exactly the same ;-)

 

Post 1 updated with new test instructions. They're equivalent to the ones we had before so results will be directly comparable.

The old table remains attached to post 1.

Please, everyone: re-run the tests for a good database!

Here's my result:

2019-03-12 06:16:56 - AC2-Win64-Shipping
Frames: 4555 - Time: 60000ms - Avg: 75.917 - Min: 57 - Max: 116

The 4 FPS AVG improvement over my last result entry comes from AMD driver fiddling. The ACC performance has remained exactly the same.

 

Maybe show hardware also for people who don't know what everyone is using!
Im on mobile and sig's don't show.

 

From post 1:

HW is noted in the table, too. Even comments section for notes on driver settings and OC.

 

Ryzen 5 1600 @3,6GHz (all core OC)
16GB DDR4-2933 CL16
Vega 56 Pulse (Undervolting + OC, Core (Boost) ~1500-1550-ish MHz, Memory 930MHz)

2019-03-12 11:10:09 - AC2-Win64-Shipping
Frames: 3662 - Time: 60000ms - Avg: 61.033 - Min: 47 - Max: 85

 

So you went for a stronger AMD GPU, after all ;-)

For ACC, you should set filtering quality to "performance" and override anisotropic filtering in the driver with the value 2. Should give you a ~10 % FPS boost with no notable quality decrease. The other settings don't make a big difference.

 

Thanks, I'll try that driver setting and run the benchmark again.


I already ordered a 580, but the next day there was a Vega 56 Pulse for just 130€ more, which IMO is well worth it considering on how much longer I can probably run the Vega until the next upgrade and the Vega has a way better cooling system than the 580 I ordered. After selling the old GPU and the game bundle V56 came with it was somewhere around ~160€ I've spend to upgrade from my 970 to V56.


Edit: BTW: If I have Fraps running I can't start ACC, it gives me a fatal error crash.
I have to start ACC first and then launch Fraps. Anyone else having this issue?

 

New results:

2019-03-12 14:15:11 - AC2-Win64-Shipping
Frames: 4160 - Time: 60000ms - Avg: 69.333 - Min: 52 - Max: 108


Thanks @LeDude83, that's quite a performance gain.

 

You're welcome. That's ~14 %, good stuff!

It was the right move to go for a Vega.

As for your FPS issue: I get a warning dialogue as I start the game that FRAPS can crash the application. But it never did.

 

I also get the warning when I launch it. But so far never had a problem.

 

The cursor gets the blue loading circle for around 2-3 seconds and then I get this error message:



Happens every time (so far) if Fraps is running and I launch ACC.

Well, it's not really important since I don't use Fraps and only installed it for this benchmark.
I use the Steam overlay for fps.

Well, I remembered between the 2 run that I also have ReShade running, but only Curves and Tonemap. I don't think those 2 are a (noticeable) performance influence, but could also have increased the result on my last run of the benchmark. Although I think the driver settings were the bigger part of that gain.


But on a side note (or maybe more relevant note?!), isn't filtering quality and AF set to 2x a bit of cheating since we don't run the same benchmark then?

 

i7-8700k gtx1080ti 16gb ram no overclocks.

Current result looks exactly the same to me.
2019-03-12 20:10:27 - AC2-Win64-Shipping
Frames: 6793 - Time: 60000ms - Avg: 113.217 - Min: 84 - Max: 169

Previous result (I think post #6 here, but not 100% sure....)
2018-11-24 12:21:09 - AC2-Win64-Shipping
Frames: 6863 - Time: 60000ms - Avg: 114.383 - Min: 85 - Max: 168