• Hey Guest. Check out your NeoGAF Wrapped 2025 results here!

Xbox One December SDK Update brings better eSRAM performance.

Status
Not open for further replies.
Sucks that deferred rendering is not common this generation because of the Xbone's architecture.

i'm not entirely certain this is the only reason why...

deferred engines are terrible for AA solutions [which is why they're forced to use FXAA most often, no?], and with the hardcore gamer's increasing demand for decent image quality, i can see why devs would move away from this rendering pipeline.

is't not forward+ better in every way anyway?
 
i'm not entirely certain this is the only reason why...

deferred engines are terrible for AA solutions [which is why they're forced to use FXAA most often, no?], and with the hardcore gamer's increasing demand for decent image quality, i can see why devs would move away from this rendering pipeline.

is't not forward+ better in every way anyway?
What benefits do defered rendering bring anyway. Judging with my own eyes GZ and PES don't seem to be doing anything that other similar games can't/aren't.
 
failing to understand how he/she is a sony apologist...

they do have the better hardware... MS can scramble all they want but Sony doesn't have to

Doesn't have to? yeah.. why would they even invest in their platform software architecture gosh... I mean its so powerful and stuff doesn't even need to optimize.
 
Amazing what a difference software can make when its in synergy with the hardware.

Good stuff, hopefully we get even better games for both platforms in the coming years.
 
What benefits do defered rendering bring anyway. Judging with my own eyes GZ and PES don't seem to be doing anything that other similar games can't/aren't.

Short answer: You can run as many lights as you want in only two shader passes.

You still need to calculate each of the lights but the lights will only be calculated where they actually hit the pixels and only pixels that need to be lit by a light will be calculated.
 
failing to understand how he/she is a sony apologist...

they do have the better hardware... MS can scramble all they want but Sony doesn't have to
Presenting updates to the Xbox SDK as "scrambling" or "insecure" while ignoring the fact that sony and every other tech company does tge same thing does point to a bit of bias.
 
I love how this is a thing now from microsoft.

"We've increased the pooooooooowazzzzzzz!!!! to maximum!!!!!"

Quite sad really. But in marketing terms they're onto a winner as people then believe its on par with ps4.

Did MS come out and say something about this? Or is it just this one dev?
 
That makes no sense.

It does if you are getting paid to write bollocks.

The "gap is narrowing" crap assumes Sony's SDK is stood still perpetually which is isn't and never will be. This rush to announce everytime MS introduces a new SDK is a decent way of getting some good news out there, but this would have happened even if MS had the faster hardware.

MS must be concerned because if devs turn to compute and push on with some big advancements they will be absolutely smoked.
 
Short answer: You can run as many lights as you want in only two shader passes.

You still need to calculate each of the lights but the lights will only be calculated where they actually hit the pixels and only pixels that need to be lit by a light will be calculated.
Doesn't compute Lol.

Give me a real world example please. I'm really trying to understand this
 
i'm not entirely certain this is the only reason why...

deferred engines are terrible for AA solutions [which is why they're forced to use FXAA most often, no?], and with the hardcore gamer's increasing demand for decent image quality, i can see why devs would move away from this rendering pipeline.

is't not forward+ better in every way anyway?

You can't use MSAA with deferred rendering, however you can still use good AA solutions like SMAA, which gives you really good IQ, see Ryse.
 
I love how this is a thing now from microsoft.

"We've increased the pooooooooowazzzzzzz!!!! to maximum!!!!!"

Quite sad really. But in marketing terms they're onto a winner as people then believe its on par with ps4.

What thing? It was a developer that said this not MS.

Let me guess MS paid them to said this?
 
GZ not count as general performance in the games? O_O
It's only one game. And not a very good example since it seems to have an inherent bottleneck that has nothing to do with raw power, but the architecture of the system.

Let me ask you this IF future fox engine games are able to run at higher than 720p, what would be the reason for the performance gains?
 
Doesn't compute Lol.

Give me a real world example please. I'm really trying to understand this

OK. Let's say you have a an object casting a shadow. Then you put another solid object in front of part of that shadow.

Forward: Rendering first object. Writing the complete shadow data to the frame buffer. There's another object here, rendering that object on top of the shadow.

Deferred: OK, let's render the objects. Oh I can see that another object is going to be on top here so I don't need to render shadows for all of those pixels. Now let me shade in all of the shadows for the object that is actually visible.

So with forward you've gone and rendered pixels that weren't actually part of the final image because they were fully occluded by another object. That's wasted power that could be used elsewhere or it could put your frame time longer than 16ms at which point your frame rate dips to 30.
 
Short answer: You can run as many lights as you want in only two shader passes.

You still need to calculate each of the lights but the lights will only be calculated where they actually hit the pixels and only pixels that need to be lit by a light will be calculated.

I don't know what that means, but it sounds really impressive.
 
OK. Let's say you have a an object casting a shadow. Then you put another solid object in front of part of that shadow.

Forward: Rendering first object. Writing the complete shadow data to the frame buffer. There's another object here, rendering that object on top of the shadow.

Deferred: OK, let's render the objects. Oh I can see that another object is going to be on top here so I don't need to render shadows for all of those pixels. Now let me shade in all of the shadows for the object that is actually visible.

So with forward you've gone and rendered pixels that weren't actually part of the final image because they were fully occluded by another object. That's wasted power that could be used elsewhere or it could put your frame time longer than 16ms at which point your frame rate dips to 30.

So if I take this correctly, because Microsoft decided to use 32mb they are forcing developers to use an inferior technology?
 
It's only one game. And not a very good example since it seems to have an inherent bottleneck that has nothing to do with raw power, but the architecture of the system.

Let me ask you this IF future fox engine games are able to run at higher than 720p, what would be the reason for the performance gains?
WTF. Deferred light are expensive and prove the raw power of a machine, because cost bandwith. The hell of logic it's your? Because xbone can't handle it, they not count in the raw power?
 
So if I take this correctly, because Microsoft decided to use 32mb they are forcing developers to use an inferior technology?

Not quite but close.

WTF. Deferred light are expensive and show the raw power of a machine, they cost bandwith. The hell of logic it's your? Because xbone can't handle it, they not count in the raw power?

No idea what you are trying to say mate?
 
Every single thread that's about SDK performance improvement will have folks saying it's closing the performance gap. Every single one of them.

I really don't understand this myself when both companies are working to improve things for devs and squeeze more performance out of their hardware.
 
It's not like it's a super GPU or anything. It's a gimped 7870. With the lower clock speeds, it's in between a 7850 and 7870. Games don't rely exclusively on the GPU either. Be interesting to see if DX12 stuff help out as well with efficiency.

Also it's only a 50% GPU advantage on paper and because of the CU count. That extra 10% from the Kinect tweak and increased clock speed on the Xbox GPU over the PS4 GPU lower that percentage a decent amount.
Not really as the 50% gpu advantage is from basic hardware comparisons. Things like the kinect tweak are mearly closing the gap to the 50% base line disadvantage.
 
So if I take this correctly, because Microsoft decided to use 32mb they are forcing developers to use an inferior technology?

I wouldn't say inferior. Both rendering techniques have their upsides and downsides. Plus you can't use deferred rendering easily/at all with transparent objects, multipass materials and certain kinds of AA. Plus it's far more likely the DDR3 is responsible since deferred rendering hammers memory bandwidth like there's no tomorrow.
 
WTF. Deferred light are expensive and show the raw power of a machine, they cost bandwith. The hell of logic it's your? Because xbone can't handle it, they not count in the raw power?

I don't think you have a full grasp on the concepts being discussed here. I know I don't but I'm trying to learn. Something you should try instead of saying inaccurate things.
 
I wouldn't say inferior. Both rendering techniques have their upsides and downsides. Plus you can't use deferred rendering easily/at all with transparent objects, multipass materials and certain kinds of AA. Plus it's far more likely the DDR3 is responsible since deferred rendering hammers memory bandwidth like there's no tomorrow.

What about forward+ rendering? What are its upsides and downsides?
 
OK. Let's say you have a an object casting a shadow. Then you put another solid object in front of part of that shadow.

Forward: Rendering first object. Writing the complete shadow data to the frame buffer. There's another object here, rendering that object on top of the shadow.

Deferred: OK, let's render the objects. Oh I can see that another object is going to be on top here so I don't need to render shadows for all of those pixels. Now let me shade in all of the shadows for the object that is actually visible.

So with forward you've gone and rendered pixels that weren't actually part of the final image because they were fully occluded by another object. That's wasted power that could be used elsewhere or it could put your frame time longer than 16ms at which point your frame rate dips to 30.
Ok that makes sense, but to me this makes it seem deferred rendering would be much less taxing.

Am I wrong in that assessment?
 
I don't think you have a full grasp on the concepts being discussed here. I know I don't but I'm trying to learn. Something you should try instead of saying in accurate things.
Wut? Why deferred light shouldn't prove the best raw power of a machine? Are you really serious? Surely I have best knowledge of your.
 
I wouldn't say inferior. Both rendering techniques have their upsides and downsides. Plus you can't use deferred rendering easily/at all with transparent objects, multipass materials and certain kinds of AA. Plus it's far more likely the DDR3 is responsible since deferred rendering hammers memory bandwidth like there's no tomorrow.

Would you then say that it transparent objects would be rare or non-existent in MGS:V?

This stuff sounds really interesting. Is there any good sources information if someone
wanted to learn about it or is it all bleeding edge stuff?
 
Wut? Why deferred light shouldn't prove the best raw power of a machine? Are you really serious?
You seem too stuck trying to "prove the best raw power machine"

Nobody is disputing that. The conversation is about a bit more than that.

GZ and PES can be played on pretty weak PC hardware, so there's more to defered rendering than needing a lot of raw power.
 
You seem too stuck trying to "prove the best raw power machine"

Nobody is disputing that. The conversation is about a bit more than that.

GZ and PES can be played on pretty weak PC hardware, so there's more to defered rendering than needing a lot of raw power.
Ok so you don't have a single idea of what you are talking about. It's clear now.ps4 has a better architecture of xbone because has more bandwith, more ram, unified ram without count the simple gpu. That's why deferred light work better without expensive cost. Weak pc are generally more powerful of the actual console, so it's a different league. I not want to prove anything, I want just say sdk improvement are welcome but you can't never cover those gap.
 
Does the PIX bit mean it was improved, or it wasn't there for XBO before? Because I know first hand it was one of the reasons the 360 was relatively great to develop for. If every effort before now was going without it...Hmm.
 
Ok so you don't have a single idea of what you are talking about. It's clear now.ps4 has a better architecture of xbone because has more bandwith, more ram, unified ram without count the simple gpu. That's why deferred light work better without expensive cost. Weak pc are generally more powerful of the actual console. I not want to prove anything, I want just say sdk improvement are welcome but you can't never cover those gap.
Ok yes ps4 is more powerful. Now can we move on to what I was actually talking about?
 
Doesn't have to? yeah.. why would they even invest in their platform software architecture gosh... I mean its so powerful and stuff doesn't even need to optimize.

oh, don't get me wrong.

i don't mean to say Sony shouldn't, and i'm quite sure i've read here that they've already brought many SDK improvements, no?

i just mean to say what others have already said, and shouldn't be discounted / disparaged for - the hardware gap is real and can't simply be made up with through SDK improvements alone.

if i'm off base here, i'm fully willing to be properly edumacated.
 
900p is acceptable now?

No, but the Xbone's GPU wasn't really built for 1080p gaming in the first place, especially with deferred rendering engines. It has half the number of ROPs that AMD themselves recommend for 1080p gaming. It wasn't a good GPU in early 2012 and it certainly isn't a good GPU in late 2014. You can thank Microsoft's insistence on having 8 gbs of DDR3 rather than GDDR5 or dedicated high speed VRAM for that.

What about that cpu advantage...

lmfao
 
You can't use MSAA with deferred rendering, however you can still use good AA solutions like SMAA, which gives you really good IQ, see Ryse.

ahhhh, very good to know.

i didn't realize SMAA was an option in deferred engines - for some reason i had it in my head that they were FXAA only [which is ew, obviously].
 
Status
Not open for further replies.
Top Bottom