computerbaseAshes of the Singularity Beta1 DirectX 12 Benchmarks

[railven]

This was the original Ashes unveil very early on, running on Mantle.

https://www.youtube.com/watch?v=t9UACXikdR0&feature=youtu.be&t=101

You notice a lot of dynamic lights on all the projectiles, lots of smoke trails, basically similar to the images we see on the side by side comparison for the 390X.

Notice Oxide specifically mentions "Thanks to Mantle & you can do this on DX12 or Vulkan too... every single shot is casting light." You can clearly see dynamic lighting of the projectiles.

I noticed it was clearly absent in the first alpha benchmark that tech sites did.

On the 970 @ Digital Foundry
http://www.eurogamer.net/articles/digitalfoundry-2015-ashes-of-the-singularity-dx12-benchmark-tested

There is absolutely zero dynamic light casting in that initial release for NV GPUs.

Compared to the 390 @ Digital Foundry

Notice there is some dynamic lights, but much less than the initial Mantle reveal and much less than the recent 390X side by side video.

Then it makes sense when you put that into context with what Oxide have said, they disabled async compute at NV's request. That there is indeed a vendor specific path and it's for NV only.

The Radeons are rendering a scene with many dynamic light sources all that time while the NV GPUs according to these shots from tech sites above are not.

You mean like this?

You used the first image in the slide show, that is image #3.

 

[Dygaza]

That video showed a lot better difference than those pictures.

https://www.youtube.com/watch?v=o2b6Nncu6zY

check scene at 4:45 for example

Quess aircrafts ran out of fuel in nvidia's rendering

 

[Silverforce11]

@railven

From that set too.

They do have light on the green laser projectile and explosions. Rockets are very flat, no lighting.

It's a shame they did not do side-by-side comparison, since Digital Foundry usually do that format for all the benchmarks!!

Should look like this:

 

[railven]

@railven

From that set too.

<snip>

They do have light on the green laser projectile and explosions. Rockets are very flat, no lighting.

I'd have to see the other video. With different shots and onscreen content, I don't really see the issue.

Watching the 390 video, when there is little activity, game looks flat over all (rather dull, woof that ground texture!).

one thing, screen shots do not do this game any justice. Kind of entertaining seeing all the little projectiles flying around. I'm no longer an RTS fan, but I'd play this just to see the swarm effects.

 

[Dygaza]

Ignore that video btw, it's not valid proof. They are running different versions of the game. Pause at 5:15 and check game versions.

 

[PhonakV30]

I watched this :

Ashes of the Singularity: DX11 vs DX12 Benchmarks [Radeon R9 390/Core i7 4790K/Core i3 4130]

there are nothing lights! I think this video is same as Zotac GTX 980 If you look at this :

Field testing : AMD vs NVIDIA : DirectX 12 - Ashes of the Singularity

on 390x side , there are ton of lights/smokes/lasers and....now GTX 980 or First link , those Videos are similar.

 

[Silverforce11]

Ignore that video btw, it's not valid proof. They are running different versions of the game. Pause at 5:15 and check game versions.

Different CPU you mean?

4 core Skylake for 390 and 12 core Xeon for 980.

 

[railven]

Did you guys watch the video for the 390?

Because it looks nothing like those screen shots either:

https://www.youtube.com/watch?v=OuUcIfcMDtc

I can't find the 970 video to make a better comparison, but this is probably as close as possible to that 970 image, and it looks just as flat/bad if you ask.

https://youtu.be/OuUcIfcMDtc?t=85


EDIT:
Some of these videos being linked I can't watch, something abouy PONYCANYON and being blocked in my country.

 

[Dygaza]

Different CPU you mean?

4 core Skylake for 390 and 12 core Xeon for 980.

In this video:

https://www.youtube.com/watch?v=o2b6Nncu6zY

AMD is running game version: 0.62.14002
Nvidia is running game version 0.61.13854


We need video where both are running same game version. I own the game and there are quite many patches coming all the time, can't keep track when they add effects and fixes.

 

[Dygaza]

Did you guys watch the video for the 390?

Because it looks nothing like those screen shots either:

https://www.youtube.com/watch?v=OuUcIfcMDtc

I can't find the 970 video to make a better comparison, but this is probably as close as possible to that 970 image, and it looks just as flat/bad if you ask.

https://youtu.be/OuUcIfcMDtc?t=85

You would also need to have same detail levels in comparision, Digitalfoundry is using Medium settings, while ther other video had high and crazy settings. Also digital's videos are using oldest version of the game.

 

[Elixer]

I just find it funny that, that video is flagged.
"This video contains content from PONYCANYON. It is not available in your country. "

 

[Good_fella]

So when AMD not rendering (NV exclusive) GPU Physx effects nobody cares, when Nvidia not rendering (AMD exclusive) Async effects is big deal.

https://www.youtube.com/watch?v=8-mpm_47k6Q
Look here. Missing effects available in Everspace.

 

[3DVagabond]

So when AMD not rendering (NV exclusive) GPU Physx effects nobody cares, when Nvidia not rendering (AMD exclusive) Async effects is big deal.

https://www.youtube.com/watch?v=8-mpm_47k6Q
Look here. Missing effects available in Everspace.

When nobody is blocking nVidia from rendering the effects but they can't, people care.

 

[PhonakV30]

So when AMD not rendering (NV exclusive) GPU Physx effects nobody cares, when Nvidia not rendering (AMD exclusive) Async effects is big deal.

https://www.youtube.com/watch?v=8-mpm_47k6Q
Look here. Missing effects available in Everspace.

Async effect is part of DX12 as standard API in Game industrial and it's Oxide exclusive not AMD exclusive so Nvidia must have this feature , while Physx effects is belong to nvidia and can be converted to OpenCL if Nvidia allows it, Like this :

Porting cuda application to opencl

 

[TheELF]

Async effect is part of DX12 as standard API in Game industrial and it's Oxide exclusive not AMD exclusive so Nvidia must have this feature , while Physx effects is belong to nvidia and can be converted to OpenCL if Nvidia allows it, Like this :

Porting cuda application to opencl

They are not talking about physx, look at tesselation that's a Dx11 specific feature(one of many just like async compute) but "everybody" (you know who) calls it gimpworks because only cards with very good Dx11 implementation can run it at ok framefrates.

https://msdn.microsoft.com/en-us/library/windows/desktop/ff476340(v=vs.85).aspx

By implementing tessellation in hardware, a graphics pipeline can evaluate lower detail (lower polygon count) models and render in higher detail. While software tessellation can be done,
...
...

Not even microsoft tells you that you have to run all the features in hardware,of course it will give you better results but it's not required to qualify for being compatible.

 

[guskline]

SilverForce11, I ran the DX12 benchmark for Ashes on both of my rigs below. Obviously the GTX980TI is more powerful than the R9 290. Even though the 290s are in CF the game apparently doesn't support multiple cards so the results obviously favor the GTX980TI when not using the cpu test but favors the 4790k when the cpu test option is checked. Note my 5960x is clocked at 4.4 Ghz (44x100) while my 4790k is at 4.7Ghz (47x100),

I did not see as clearly the phenomenon you point out between AMD and Nvidia cards. I will say the R9 290 run seemed to "sparkle" more than the GTX980TI run.

For those who ask that I run the same cpu with the different cards, I would like to accommodate you but both of my rigs are custom watercooled with single loop setups which would require draining refilling etc. Not worth it.

 

[Silverforce11]

@guskline

There's no need to switch CPU I don't think it matters. It's more a graphical setting/effect.

Lookout for something like this:

Glows on missiles (besides the trails only)

FYI: This is from a user with Fury X @ http://imgur.com/a/x4gT6

Glowing missiles:

Bomber bombs:

Missiles glow from launch to impact:

Another user on Fury X also has all the glowing effects:

http://imgur.com/a/of5aS

From here: https://www.reddit.com/r/Amd/comments/45bd6b/need_testers_for_dx12_ashes_of_the_singularity/

 

[guskline]

SilverForce11, I'll run them right now and focus on the "glow factor". Be back in @10 minutes.

 

[guskline]

SilverForce11, I'm back after another DX12 benchmark run on both of my rigs below. I should learn how to capture and post the videos. It would be much more accurate and descriptive. (at nearly 65 need to learn this techniques!).

Anyway, no question the R9 290 produced more afterglow. The GTX980TI also did but it looked "staged" or planned and was NOT nearly as effective or broad based as the results from R9 290 (Sapphire Tri-X at 1000/1300). For instance in a fire fight scene with the 290 each hit on an object appears to produce an afterglow while only certain hits on the 980TI produce the same result.

Interesting phenomenon.

P.S. It's fun having both a R9 290 and GTX980TI to see the differences in DX12 as they evolve.

 

[AtenRa]

Yea, remember this ??

https://youtu.be/H1L4iLIU9xU?t=347

 

[Silverforce11]

Yea, remember this ??

https://youtu.be/H1L4iLIU9xU?t=347

This is more important point:

https://youtu.be/H1L4iLIU9xU?t=15m12s

People have always assumed Async Compute is something to help with shader uptime but it is not just that simple.

Many compute tasks do not actually use the same parts of the GPU shaders, but if they are forced to run in serial mode, they delay overall rendering.

When compute tasks can be run in parallel without synchronous limitations, the GPU can output more rendering work than possible otherwise.

 

[PhonakV30]

@guskline

Your game Version?

 

[AtenRa]

This is more important point:

https://youtu.be/H1L4iLIU9xU?t=15m12s

People have always assumed Async Compute is something to help with shader uptime but it is not just that simple.

Many compute tasks do not actually use the same parts of the GPU shaders, but if they are forced to run in serial mode, they delay overall rendering.

When compute tasks can be run in parallel without synchronous limitations, the GPU can output more rendering work than possible otherwise.

I was referring to the entire video

 

[Azix]

This is more important point:

https://youtu.be/H1L4iLIU9xU?t=15m12s

People have always assumed Async Compute is something to help with shader uptime but it is not just that simple.

Many compute tasks do not actually use the same parts of the GPU shaders, but if they are forced to run in serial mode, they delay overall rendering.

When compute tasks can be run in parallel without synchronous limitations, the GPU can output more rendering work than possible otherwise.

yep. Best thing about dx12 IMO. Its literally free performance. Either get prettier games at the same fps or more fps (probably much better minimums). Compute all the way

 

[Mahigan]

SilverForce11, I'm back after another DX12 benchmark run on both of my rigs below. I should learn how to capture and post the videos. It would be much more accurate and descriptive. (at nearly 65 need to learn this techniques!).

Anyway, no question the R9 290 produced more afterglow. The GTX980TI also did but it looked "staged" or planned and was NOT nearly as effective or broad based as the results from R9 290 (Sapphire Tri-X at 1000/1300). For instance in a fire fight scene with the 290 each hit on an object appears to produce an afterglow while only certain hits on the 980TI produce the same result.

Interesting phenomenon.

P.S. It's fun having both a R9 290 and GTX980TI to see the differences in DX12 as they evolve.

Well then... I'm left speechless. We've been comparing apples to oranges all along.