Does it explain the even more games that load slower? XB1 is known for long load times and long install times.
Xbox One (Durango) Technical hardware investigation
- Thread starter Love_In_Rio
- Start date
- Status
Does it explain the even more games that load slower? XB1 is known for long load times and long install times.
http://www.dualshockers.com/2014/08...-display-thousand-of-dynamic-lights-and-more/
Seems that Forward+ is starting to be implemented in next gen titles. Would be interesting to know more about who they are utilising the ESRAM in conjunction with F+.
So, I get that deferred shading basically provides efficiency in that you'll only shade fragments that are visible. So, what exactly is forward plus? I can't find a lot of info, but it seems like a 3-stage tiled-renderer that uses compute shaders to generate light lists for each tile. I'm not sure how that avoids that shading inefficiency of a forward renderer.
Did you suggest that would be a way forward for ESRAM because ESRAM would be a good match for a tile-based renderer and compute shaders without having a fat g-buffer?
Found a whole thread for Forward+ vs deferred here for you on B3D:
http://beyond3d.com/showthread.php?t=65195
I believe the answer is yes, and sebbbi has made mention that optimizing for X1 would increase the performance on the other two platforms.
The only thing one might need to consider is that the esram is dedicated for the GPU only, which isn't the case with PS4.
Perfect. Not sure why I didn't think of searching for it. I'm dumb.
You're still in NHL shock. Give it a couple of days.
I made a similar point before, specific to Forward+; in general most optimizations will be beneficial to both platforms, but probably at worst neutral for PS4. In this instance optimizing the tiling for ESRAM and keeping as many operations local as possible will help to utilize the overall available bandwidth.
Good news for the Xbox One. Josh Mosqueira, lead designer of Diablo 3 said that the 1080p resolution was achieved using the new SDK tools, which are a lot better than they were, not because of dropping Kinect and the 10% system resources reservation.
In fact the that extra 10% resources didn't make a difference for them.
https://translate.google.com/transl...os-nuevos-sdk-no-por-quitar-kinect/&edit-text=
In fact the that extra 10% resources didn't make a difference for them.
https://translate.google.com/transl...os-nuevos-sdk-no-por-quitar-kinect/&edit-text=
Questions.
How does it it make a difference in Destiny, but not in Diablo 3? They simply chose not to use it? Maybe didn't have time or the will to use it?
Also is MS not getting the new SDK tools to ALL developers in a timely matter?
Just curious.
Well, I suppose the simplest answer would be that Diablo's bottlenecks were different than Destiny's (possibly CPU-bound or ESRAM-related).
Or it could just be bullshit PR from a non-software engineer.
Betanumerical
Veteran
I doubt he meant literally no difference. That's impossible.
We've always known 10% is not mathematically enough to get 900P to 1080P. It will be a combination of things included in this case some lowered framerates.
Global Illumination via Forward+ has already been implemented in-game in Dirt Showdown, and is available in a couple of subsequent Codemasters titles that use the same, or subsequent, version of the engine (i.e. Grid 2, IIRC)Dave
Can Dynamic GI be done on tiled based renderers Forward or Deferred?
http://www.rage3d.com/articles/gaming/codemaster_dirt_showdown_tech_review/index.php?p=2
Dave
Can Dynamic GI be done on tiled based renderers Forward or Deferred?
There's all kinds of ways to implement dynamic GI, and most of them can work in either a deferred or forward setup (or anything in-between). There are a few techniques that rely on generating lots of virtual point lights (VPLs), which is definitely something that's better suited for a deferred renderer that can handle many hundreds or thousands of lights in a single frame.
Lol thanks guys!
So I'm under the impression that Dynamic GI has to have at least 1 bounce but I'm feeling that I'm wrong on that definition given both your answers.
In particular when I look at Fable Legends or Drive Club, and Tomorrow Children: is that just very well done VPLs? or are those games leveraging some sort of path tracing algo? Because there is an inherent difference in quality between something like Dirt 2 and the unreleased titles. The only reason I asked was because to me it seemed like path tracing and tiled based rendering as a concept sounds more taxing than a non-tiled variant.
So I'm under the impression that Dynamic GI has to have at least 1 bounce but I'm feeling that I'm wrong on that definition given both your answers.
In particular when I look at Fable Legends or Drive Club, and Tomorrow Children: is that just very well done VPLs? or are those games leveraging some sort of path tracing algo? Because there is an inherent difference in quality between something like Dirt 2 and the unreleased titles. The only reason I asked was because to me it seemed like path tracing and tiled based rendering as a concept sounds more taxing than a non-tiled variant.
Last edited by a moderator:
Most definitions of "GI" require at least one bounce.
The responses from Dave and MJP do not disagree with this; the example given by MJP of using virtual point lights refers to the use of dynamic virtual point lights in the calculation of the bounced lighting.
Last edited by a moderator:
One step further, the important thing is to generate the same look as bounced light. If a technique is found that can achieve that without having to perform an actual light bounce, it'd still count as GI.
Rather saying GI requires a minimum of one bounce, it's probably better to say it requires a minimum of one level/iteration of secondary illumination.
- Status
Similar threads
