Ashes of the Singularity Bench

Gregster · 4 Sep 2015 at 14:30

1440P - Titan X @ 1430/4002 - 3930K @ 4.625 - 355.82

1440P same clocks but DX11 run

I think I will stick with DX11

TheRealDeal · 4 Sep 2015 at 14:41

I think you guys should have waited for the gtx980ti looking at Andy's score lol. There's obviously something up somewhere.

ALXAndy · 4 Sep 2015 at 14:50

TheRealDeal said:
I think you guys should have waited for the gtx980ti looking at Andy's score lol. There's obviously something up somewhere.

Eh? I'm 1 FPS behind Kaap's Titan X. You do realise I ran the bench at 4k, yes? Every one else is using 1080p and 1440p !

Kaap also has a 5960x and in this bench it seems to matter quite a lot, so I don't think 1 FPS behind a Titan X is too bad really.

TheRealDeal · 4 Sep 2015 at 14:53

ALXAndy said:
Eh? I'm 1 FPS behind Kaap's Titan X. You do realise I ran the bench at 4k, yes? Every one else is using 1080p and 1440p !

Kaap also has a 5960x and in this bench it seems to matter quite a lot, so I don't think 1 FPS behind a Titan X is too bad really.

I am talking about the Titan X guys who ran at 1080p. They are almost 10fps behind AndyBird. Should have typed the whole name. I think something is going on with settings in this bench. The Titans and Fury look well matched but the gtx980ti is so far in front at 1080p.

ALXAndy · 4 Sep 2015 at 14:54

TheRealDeal said:
I am talking about the Titan X guys who ran at 1080p. They are almost 10fps behind AndyBird. Should have typed the whole name. I think something is going on with settings in this bench. The Titans and Fury look well matched but the gtx980ti is so far in front at 1080p.

Something is going on with this bench period tbh dude. Poorly coded. Would be a better idea to wait until something a bit more professional comes out for DX12 IMO.

thebennyboy · 4 Sep 2015 at 15:00

Would be easier to just use 1 preset rather than changing the MSAA, should just tell everyone to use the "High" preset.

andybirds score is way too high, even for his overclock.

andybird123 · 4 Sep 2015 at 15:30

I'll run it again later - my log file and settings ini all say the settings in the OP but I'll run it three times back to back later and see what the variability is like

Gregster · 4 Sep 2015 at 15:37

andybird123 said:
I'll run it again later - my log file and settings ini all say the settings in the OP but I'll run it three times back to back later and see what the variability is like

Looks fine to me. Your settings are the same as mine, so I wouldn't worry about it.

humbug · 4 Sep 2015 at 15:43

Gregster said:
Looks fine to me. Your settings are the same as mine, so I wouldn't worry about it.

lol...

With his 980TI he's pushing 30% higher performance than your Titan-X

Gregster · 4 Sep 2015 at 15:44

humbug said:
lol...

With his 980TI he's pushing 30% higher performance than your Titan-X

I won't self harm, so don't worry

humbug · 4 Sep 2015 at 15:51

Gregster said:
I won't self harm, so don't worry

Clock it to 2000Mhz, you might just beat him....

queamin · 4 Sep 2015 at 15:52

Really I would take results with a pinch of salt until the game is released really as things can change but at moment the 290/390 looks like a dark horse with dx12 but we cannot tell until more games come out.

Gregster · 4 Sep 2015 at 15:56

humbug said:
Clock it to 2000Mhz, you might just beat him....

As I get more frames in DX11 over DX12, I stick by what I said from the very start and that was "Whilst AOTS is a CPU bench and shows how DX12 helps with the CPU overheads, it isn't a benchmark that I would ever use as a basis for GPU benching. Get something like Heaven/3DMark and some games and then we will have a clearer indication of what we can fanboy over"

And I stick by it, as clearly the difference between mine and Andy's and the fact that my DX11 score was higher indicates that this bench isn't a good indication of GPU grunt.

humbug · 4 Sep 2015 at 16:08

Gregster said:
As I get more frames in DX11 over DX12, I stick by what I said from the very start and that was "Whilst AOTS is a CPU bench and shows how DX12 helps with the CPU overheads, it isn't a benchmark that I would ever use as a basis for GPU benching. Get something like Heaven/3DMark and some games and then we will have a clearer indication of what we can fanboy over"

And I stick by it, as clearly the difference between mine and Andy's and the fact that my DX11 score was higher indicates that this bench isn't a good indication of GPU grunt.

Andy's results are certainly unusual. perhaps Oxide made an unreliable pile of junk here, who knows, maybe more people will get a 30/40% boost in performance out of nowhere.

Nvidia getting higher DX11 performance has been well debated, Nvidia initially blamed Oxide which actually turned out to be a problem with Nvidia's Drivers. it also doesn't help given that Nvidia doesn't actually have Hardware ASync.

Gregster · 4 Sep 2015 at 16:18

Well, there is a thread already for discussing what is what, so best leave this one for people who have the game and posting results

muziqaz · 4 Sep 2015 at 16:41

Why is everyone so surprised by Andybirds results? I have been saying it for a long time: graphics settings are broken in that benchmark. Simple as that.
Now, here is my 1080p run with ~~custom settings Kaap suggested~~:

https://flic.kr/p/xVr3a5

It does look like I am using Crazy settings plus 4xMSAA, yes? Well, no, because in this run I disabled Temporal AA, and unless I am completely blind, it doesn't report in benchmark result screen that I've done it.

I am not saying that Andybird lied to us and disabled TAA, or other settings, but I am suggesting that game somehow disabled some quality settings, when he was trying to fiddle with custom settings. That would explain why Titan xs are being beat so bad by Andy's supercomputer.

Now for real runs with correct settings:

1080p 4xMSAA 1117Mhz Fury X: https://flic.kr/p/ydK6xa

1440p 4xMSAA 1117Mhz Fury X: https://flic.kr/p/yaJPu1

4k 4xMSAA Fury X Stock: https://flic.kr/p/xgaWPX
4k 4xMSAA 1110Mhz Fury X: https://flic.kr/p/xg2WkC
4k 4xMSAA 1117Mhz Fury X: https://flic.kr/p/yd4eDV

4k 4xMSAA 300Mhz Fury X: https://flic.kr/p/yd4eHn

Regarding MSAA, I don't see the point of setting it up, since devs said it is still broken, and there is no difference with or without it in this benchmark, especially in 4K. But since thread was set up by our resident Eagle Eye Kaap, it is his choice

Disabling it in 1080p gave me 6fps more.

P.S. CPU as in sig, and xfire was disabled, since while it is enabled, game just crashes

LtMatt · 4 Sep 2015 at 16:44

I did warn you to wait.

muziqaz · 4 Sep 2015 at 16:46

Streetlight said:
R9 290 1285/1500
AMD 9590 @5GHz
1080p
15.8 drivers

I wonder what your card would post with decently clocked Haswell E

muziqaz · 4 Sep 2015 at 16:47

AMDMatt said:
I did warn you to wait.

Some people are very impatient here

It would be logical to wait till xfire/sli support is enabled in the game, by then we would have more fix benchmark anyway

now there are too many settings to hide and get away with (I'm not suggesting anyone here done it).

fs123 · 4 Sep 2015 at 16:52

muziqaz said:
Some people are very impatient here
It would be logical to wait till xfire/sli support is enabled in the game, by then we would have more fix benchmark anyway now there are too many settings to hide and get away with (I'm not suggesting anyone here done it).

So it's possible to modify stuff in some config file that doesn't show up in the screenshots? Then it would make these benchmark comparisons null and void.