Ashes of the Singularity Coming, with DX12 Benchmark in thread.

[Dygaza]

Didn't oxide said themselves that they don't recommend MSAA 4x to be run in bench, because it's not optimised yet. Then again it's same for everyone, optimised or not.

Doh, this went to wrong thread.

 

[Orangey]

but I still find it very difficult to believe that NVIDIA outright lied about Maxwell 2 and it's async capabilities.

 

[Silent_Scone]

Snip

Best sell that TITAN X quick.

I've seen the comparison made a few times before Silent, that's all. I wasn't sure if you were joking or not, but replied anyway ;P

lol no I was kidding, just the prospect of the finger pointing was amusing

 

[pmc25]

Let them, I can't see how that would be the case. The capabilities are likely very much there. They also tested async compute scenarios in other games and the utilisation makes much more sense. Fair play to them, it's not something I would have really bothered to look into until such a time performance was degraded in a game that I could play in front of me - and not a canned benchmark

Several developers have stated that Maxwell (v1 or v2) flat out can't do asynchrous compute or shaders, the architecture simply doesn't allow for it. You can force it through software, but per Oxide's statement, it's so slow that it actually decellerates things. The capabilities aren't there, and NVIDIA did lie (or at least obfuscate the truth - again).

 

[Gregster]

Well from test after test in this and seeing the results in the bench thread, clearly this game is broken and Nvidia were indeed correct and this shouldn't be taken as a "DX12 generalisation and wait for other benches"

Several developers have stated that Maxwell (v1 or v2) flat out can't do asynchrous compute or shaders, the architecture simply doesn't allow for it. You can force it through software, but per Oxide's statement, it's so slow that it actually decellerates things. The capabilities aren't there, and NVIDIA did lie (or at least obfuscate the truth - again).

Going from what the state of this game is, I wouldn't put too much stock in anything Oxide have said. The game is clearly broken.

 

[muziqaz]

Greg, don't jump the gun so soon Sweet J!!! The game is in pre-beta state. We found out that menu is broken, I am yet to see any indication of performance being broken. And setting menu being broken does not confirm what nvidia PR is yapping about. What is performance indication according to nVidia? overtesselation? Watchdogs? AC Unity, Batman Arkham Knight (that latest disaster one)? Love this list so far with amazing nvidia collaboration.
Oxide devs keep saying they use async shaders, and since GCN async shaders are miles ahead of nvidia ones even on paper, I don't see why their statements would seems as lie.
And we are yet to see any substantial proof that nvidia async shaders are even functional.
I wonder if there is gonna be any benchmark developer with big enough balls to create a benchmark/feature test to measure GPU async shader performance
I wonder...

 

[Gregster]

No jumping the gun and I appreciate that this is a very early Alpha, but the fact remains that I get better performance from DX11 than DX12, 290's are so close to a Fury X and Titan X, MSAA doesn't seem to work properly.

Just my few observations in this.

 

[Howling]

Not understanding the results is not the same as a broken test. Though I'll admit I don't get all the results either, there seems to be a lot more going on than simple async.

 

[Harlequin]

the cmd line programme which demonstrates compute earlier in the thread already shows async isn't working properly on nv cards - something like 1/4 rate (hence the stair case graph)

 

[Harlequin]

No jumping the gun and I appreciate that this is a very early Alpha, but the fact remains that I get better performance from DX11 than DX12, 290's are so close to a Fury X and Titan X, MSAA doesn't seem to work properly.

Just my few observations in this.

what if MSAA is parallel and running async.....

 

[Gregster]

what if MSAA is parallel and running async.....

LMAO What does this even mean? MSAA has nothing to do with ACE.

Not understanding the results is not the same as a broken test. Though I'll admit I don't get all the results either, there seems to be a lot more going on than simple async.

I have been around computers long enough to understand what is what and I can understand that this pre-beta game isn't up to a standard that can be deemed as reliable.

 

[Harlequin]

LMAO What does this even mean? MSAA has nothing to do with ACE.

erm you do know how a frame is rendered don't you??

 

[fs123]

Whether the test is reliable or not does not take away from the general 'accusation' that current Nvidia cards don't support proper Async Shaders. If that is true then I reckon in 1 years time Maxwell cards could be in for a rough ride.

Nvidia customers don't have to worry though, I'm sure Nvidia will release a new card next year that will support all of DX12 features.

 

[Gregster]

erm you do know how a frame is rendered don't you??

No, you tell me how MSAA is linked with ACE?

Whether the test is reliable or not does not take away from the general 'accusation' that current Nvidia cards don't support proper Async Shaders. If that is true then I reckon in 1 years time Maxwell cards could be in for a rough ride.

Nvidia customers don't have to worry though, I'm sure Nvidia will release a new card next year that will support all of DX12 features.

I did ask you earlier if Async was needed and what will it do for Nvidia or what will Nvidia miss out on with potentially not even having it?

I am looking forward to learning stuff today

 

[Gregster]

Ohhh and just to keep my editing skills up, here is DX12 Vs DX11 in AOTS.

Glitchy DX12 recording but game played fine

 

[r7slayer]

Hahaha this had me laughing so hard! Works so well!

 

[r7slayer]

Ohhh and just to keep my editing skills up, here is DX12 Vs DX11 in AOTS.

Glitchy DX12 recording but game played fine

Wow, was that just recording or did you see the same on your screen in game/bench? It got incomparable after so long

 

[Gregster]

Wow, was that just recording or did you see the same on your screen in game/bench? It got incomparable after so long

Glitchy DX12 recording but the game played fine (same as what I said under the video and on the youtube description).

 

[r7slayer]

Glitchy DX12 recording but the game played fine (same as what I said under the video and on the youtube description).

ahh didn't see that bit mate. Ahh okays

If im honest i looked the video without taking notice which one was DX11 or DX12 and looked which one was better! The one on the right looked better visually and seemed to run better. Then i saw the glitchy stuff which threw me haha.
So your video shows for your comparison DX11 runs much better than DX12 on your hardware mate from my observation any ways.

edit -

Take that back the one on the left looks slightly better visually. the one on the right runs sooo much better which is why it looked better.

I wasn't watching in 1080p i was watching in 480 and not full screen.

 

[Gregster]

ahh didn't see that bit mate. Ahh okays

If im honest i looked the video without taking notice which one was DX11 or DX12 and looked which one was better! The one on the right looked better visually and seemed to run better. Then i saw the glitchy stuff which threw me haha.
So your video shows for your comparison DX11 runs much better than DX12 on your hardware mate from my observation any ways.

DX12 recording is off if I use ShadowPlay or my dedicated capture card at the mo and will probs need a fix at software level but very early days and I expect some teething problems with DX12 and compatible hardware.

I thought I would do the vid just to show if there was any visual differences but I couldn't actually see any (looking for why DX11 gave me more frames over DX12) and trying to keep on top of my Youtube/Vegas skills