Xbox One SDK Update allows developers more eSRAM Control.

[WYP]

Xbox One SDK Update allows developers more eSRAM Control, Improving Xbox One Performance.

Read more on the Xbox One SDK Update which gives developers more eSRAM Control here.

 

[SPS]

Hmm interesting.. remember there's only 32MB of ESRAM

 

[NeverBackDown]

Then Sony releases a new SDK that further improves control etc.. It's just a cycle
But more control of only 32MB has more impact than more control over the other 5Gb used for the games in the xbone.. so it's interesting to see it's already being immediately impactful to devs.

 

[Ramon Zarat]

No amount black magic voodoo will EVER make up for 50% less shader power compared to the PS4. The XB1 will hit 1% better performance with DX12 and another 1% with this SRAM update to counter the negative effect of slow DDR3.

All this is meaningless anyway as BOTH console are already completely obsolete. If in 2015 you can't run DX11 code at 1080P with all the bells and whistles maxed out (drawing distance, level of details, anti aliasing etc...) at a steady 60FPS, then as a gaming device, you are worthless. DX11 API was lauched in 2008, ***7 YEARS AGO***.

 

[jaspervandervalk]

No amount black magic voodoo will EVER make up for 50% less shader power compared to the PS4. The XB1 will hit 1% better performance with DX12 and another 1% with this SRAM update to counter the negative effect of slow DDR3.

All this is meaningless anyway as BOTH console are already completely obsolete. If in 2015 you can't run DX11 code at 1080P with all the bells and whistles maxed out (drawing distance, level of details, anti aliasing etc...) at a steady 60FPS, then as a gaming device, you are worthless. DX11 API was lauched in 2008, ***7 YEARS AGO***.

I have to agree with Ramon on this one, gaming consoles have become so obsolete its a shame that they under perform so massively compared to PC.

People do not want to splash out anything higher than 400GBP on a console so in a way console companies are forced to work around this set back by providing perhaps not even second grade hardware.

Console companies definitely need to rethink their business model for the coming generations, they may have gotten away with it by the skin of their teeth this time but next time I think they may not be so lucky.


Jasper

 

[NeverBackDown]

No amount black magic voodoo will EVER make up for 50% less shader power compared to the PS4. The XB1 will hit 1% better performance with DX12 and another 1% with this SRAM update to counter the negative effect of slow DDR3.

All this is meaningless anyway as BOTH console are already completely obsolete. If in 2015 you can't run DX11 code at 1080P with all the bells and whistles maxed out (drawing distance, level of details, anti aliasing etc...) at a steady 60FPS, then as a gaming device, you are worthless. DX11 API was lauched in 2008, ***7 YEARS AGO***.

Ignorant to say that when you are not creating games running with these seriously low level SDKs and know how they work. Yes an amount of Voodoo can make up for 50% performance if the more powerful hardware has worse optimization.. but because Sony optimize there stuff it won't always will. Saying 1% better performance with DX12/eSRAM combined is an exaggeration yes, but honestly it's a horrible one. Also they never said it was Dx12 in the article...

"Techland, the developer behind Dying Light has said that the reason they were able to run both the PS4 and Xbox One versions at the same framerates and resolution is because they have been able to utilise the eSRAM in the Xbox One more effectively. This additional eSRAM control, and the performance gains that have come with it, is due to Decembers Xbox One SDK Update."

Did you not read the article? That's pretty impressive i must say. Made up for 50% performance did it not? DDR3 will always be a factor when it comes down to how much raw data needs to be tossed around, that i won't disagree with. Still however means that the eSRAM can now be used to greater lengths to help aide in serious data restricted moments. Key word here is moments though since there is only 32MB of it. Hard to contend with 5GB of GDDR5 at all times.

This reply is no way to be offensive. Just trying to point things out here.

 

[Meladath]

Ignorant to say that when you are not creating games running with these seriously low level SDKs and know how they work. Yes an amount of Voodoo can make up for 50% performance if the more powerful hardware has worse optimization.. but because Sony optimize there stuff it won't always will. Saying 1% better performance with DX12/eSRAM combined is an exaggeration yes, but honestly it's a horrible one. Also they never said it was Dx12 in the article...



Did you not read the article? That's pretty impressive i must say. Made up for 50% performance did it not? DDR3 will always be a factor when it comes down to how much raw data needs to be tossed around, that i won't disagree with. Still however means that the eSRAM can now be used to greater lengths to help aide in serious data restricted moments. Key word here is moments though since there is only 32MB of it. Hard to contend with 5GB of GDDR5 at all times.

This reply is no way to be offensive. Just trying to point things out here.

The problem is that there is only 32MB of it. It won't be useful for very many things at all. It will face the same problems the Xbox 360 did with any post process techniques, almost impossible to actually implement due to low memory count, or slow to implement because of slow memory if you don't use the eSRAM.

While it can be used to alleviate the DDR3 problem a bit, it is a very poor solution, why the hell they didn't go with GDDR5 I don't think anyone knows, its a far better solution. The amount of usage the eSRAM will have to have to get anywhere near close to making the DDR3 as good as GDDR5 is pretty damn high, and as well all know, unless you are talking about GPGPU, memory transfers are always a bottleneck.

 

[NeverBackDown]

The problem is that there is only 32MB of it. It won't be useful for very many things at all. It will face the same problems the Xbox 360 did with any post process techniques, almost impossible to actually implement due to low memory count, or slow to implement because of slow memory if you don't use the eSRAM.

While it can be used to alleviate the DDR3 problem a bit, it is a very poor solution, why the hell they didn't go with GDDR5 I don't think anyone knows, its a far better solution. The amount of usage the eSRAM will have to have to get anywhere near close to making the DDR3 as good as GDDR5 is pretty damn high, and as well all know, unless you are talking about GPGPU, memory transfers are always a bottleneck.

I know? I addressed and even said that. The 32MB is not a lot but it is extremely fast so properly using it will help. There's no need to use post processing on consoles. It's too GPU intensive and will take away the quality more than enhance hit because something else will have to compensate for it. Full GDDR5 is the better solution yes. I'm sure MS went down the cheaper route and threw in eSRAM to help compensate for DDR3. Now if there was more like 64-128MB it wouldn't have been a bad idea. 32MB is only because they did the same with the 360.

 

[SPS]

I know? I addressed and even said that. The 32MB is not a lot but it is extremely fast so properly using it will help. There's no need to use post processing on consoles. It's too GPU intensive and will take away the quality more than enhance hit because something else will have to compensate for it. Full GDDR5 is the better solution yes. I'm sure MS went down the cheaper route and threw in eSRAM to help compensate for DDR3. Now if there was more like 64-128MB it wouldn't have been a bad idea. 32MB is only because they did the same with the 360.

All next gen console games use post processing. It used to be thought of as some very expensive technique that was applied right at the end of the frame such as AA, however most if not all games use a deferred renderer now so screen space techniques are actually very effective for their cost. Imagine a next gen game with no post processing... You'd want your pennies back.

 

[NeverBackDown]

All next gen console games use post processing. It used to be thought of as some very expensive technique that was applied right at the end of the frame such as AA, however most if not all games use a deferred renderer now so screen space techniques are actually very effective for their cost. Imagine a next gen game with no post processing... You'd want your pennies back.

What's a deferred render? If it does it before each frame then "post" processing wouldn't be an accurate way to describe it. If it's before each frame then i stand correct, there is no need for it.

 

[SPS]

What's a deferred render? If it does it before each frame then "post" processing wouldn't be an accurate way to describe it. If it's before each frame then i stand correct, there is no need for it.

It's where all mesh data is rendered to the G-Buffer. Data such as world space positions, normals, diffuse, material IDs, etc are stored in various render targets, the G-Buffer. Then lighting is performed per pixel. It also lends itself very well to screen space techniques such as ambient occlusion, reflections, DoF, subsurface scattering, etc because you have all the data in screen space. Instead of your renderer being bound by the amount of polys, you are now only limited by fill rate, because all intensive GPU effects are applied per pixel of your render target/frame buffer/back buffer, not per vertex per mesh. This is how lots of lights are able to be rendered.

Post process means apply effects to the overall image, I.e. Your render target after you have drawn your meshes. Post, means after.

 

[NeverBackDown]

It's where all mesh data is rendered to the G-Buffer. Data such as world space positions, normals, diffuse, material IDs, etc are stored in various render targets, the G-Buffer. Then lighting is performed per pixel. It also lends itself very well to screen space techniques such as ambient occlusion, reflections, DoF, subsurface scattering, etc because you have all the data in screen space. Instead of your renderer being bound by the amount of polys, you are now only limited by fill rate, because all intensive GPU effects are applied per pixel of your render target/frame buffer/back buffer, not per vertex per mesh. This is how lots of lights are able to be rendered.

Post process means apply effects to the overall image, I.e. Your render target after you have drawn your meshes. Post, means after.

I know what it means. No need for an english lesson. I was talking about the renderer being before each frame so therefore the post processing wouldn't be accurate way of describing it. But now that you have explained it, all the details are applied after meshes are created which makes sense. Earlier in regards to post processing, i thought he meant AA/AF was what he was talking about. Nothing like a deferred renderer That was why i said it wasn't needed.. AF could be useful since it does hardly anything to performance but AA is a no go as far as i am aware. TO much cost for a console's limited power

 

[SPS]

I know what it means. No need for an english lesson. I was talking about the renderer being before each frame so therefore the post processing wouldn't be accurate way of describing it. But now that you have explained it, all the details are applied after meshes are created which makes sense. Earlier in regards to post processing, i thought he meant AA/AF was what he was talking about. Nothing like a deferred renderer That was why i said it wasn't needed.. AF could be useful since it does hardly anything to performance but AA is a no go as far as i am aware. TO much cost for a console's limited power

I didn't really get what you meant by before lol. UE4's temporal AA runs quite well on console and I'd expect most games have some sort of AA running (current gen).

 

[NeverBackDown]

I didn't really get what you meant by before lol. UE4's temporal AA runs quite well on console and I'd expect most games have some sort of AA running (current gen).

Wouldn't AA in any form be bad? They could use that processing power for better lighting or textures? That would make the game look better imo.

Just looked up Temporal AA.. many games on next gen consoles are actually using this. From what i have read, it's similar in cost to FXAA but its end result is much better and more versatile because it can be added to any pass. Saw pictures of different AA on Ac4.. it definitely needed it! Without it, looked all blurry and lines were not defined at all.