Imagination update from CES - Ray tracing slated for Q4

[D.P.]

Deferred rending is commonplace in DX11 games in anycase.

nvidia actually have a load of IP in deferred rendering and tile based DR they acquired form 3DFX. They haven't used it because the gains don't out way the transistor cost.
PVR have been going on about their magic tech for well over a decade and promising that immediate mode rendering has 12 months left before it goes extinct. Funny how they are always 12months away from taking over the world.

It is also completely wrong to believe that immediate mode rendering renders every pixel multiple times, not least the depth buffer will ensure only the closer objects are drawn and a well made game engine will often throw objects at an approximate correct order to minimize overdrawn. techniques like display lists, instancing, commandlists are standard on OGL/DX and make PVR's solution entirely redundant.

 

[Pottsey]

Yes immediate mode rendering doesn’t render every pixel but it’s not where near 99% efficient at removing pixels like TBDR and immediate mode rendering doesn’t have the same level of bandwidth savings or memory savings.

“Also please link me up to where people are saying that RT is less computationally and power intensive than traditional rasterization. At the same visual fidelity RT will always be more computationally intensive. Lets see what some bloke who works for Facebook has to say on the matter: http://arstechnica.com/gadgets/2013/...mment-23723213 some interesting comments about overdraw reduction in traditional IMR systems as well. ”

I already given you the link more than once in this thread and other threads. But like all evidence that doesn’t fit into your viewpoint you just ignore it. The link I gave you was GPU engineers from GPU Company’s talking at high end GPU conferences. The audience all full of GPU engineers from all the major company’s so it’s not like the speakers could talk rubbish and make it all up. The audience would call you out on any rubbish if tried that. But apparently that’s not enough evidence for you!

As for the bloke from Facebook did you even bother to read the first line? He edited his post to say he has to partly redact what he saying as he had no details on PowerVR’s particular technology which went against what he said. If I recall correctly later on he posted he changed his mind after reading up on the PowerVR technology

“As for your crazy idea that PVR have some magic technology for not rendering objects that are not visible on screen leagues ahead of anybody else...”

You know what that just proves everything I said about you. Either you are clueless or are just trying to cause any argument. Do you have any idea just how wrong what you said is? So now you think well document technology confirmed by many people is “magic technology”. After you said something at stupid as this it’s not even worth trying to debate with you over everything else that’s wrong in your post. Not 100% sure which it is that you don’t understand or are just arguing even though you know you are wrong. But between this comment and ignoring the GPU conference and GPU engineers as evidence well it says a lot about you.

 

[layte]

Yes immediate mode rendering doesn’t render every pixel but it’s not where near 99% efficient at removing pixels like TBDR and immediate mode rendering doesn’t have the same level of bandwidth savings or memory savings.


I already given you the link more than once in this thread and other threads. But like all evidence that doesn’t fit into your viewpoint you just ignore it. The link I gave you was GPU engineers from GPU Company’s talking at high end GPU conferences. The audience all full of GPU engineers from all the major company’s so it’s not like the speakers could talk rubbish and make it all up. The audience would call you out on any rubbish if tried that. But apparently that’s not enough evidence for you!

As for the bloke from Facebook did you even bother to read the first line? He edited his post to say he has to partly redact what he saying as he had no details on PowerVR’s particular technology which went against what he said. If I recall correctly later on he posted he changed his mind after reading up on the PowerVR technology



You know what that just proves everything I said about you. Either you are clueless or are just trying to cause any argument. Do you have any idea just how wrong what you said is? So now you think well document technology confirmed by many people is “magic technology”. After you said something at stupid as this it’s not even worth trying to debate with you over everything else that’s wrong in your post. Not 100% sure which it is that you don’t understand or are just arguing even though you know you are wrong. But between this comment and ignoring the GPU conference and GPU engineers as evidence well it says a lot about you.

You will excuse me if I take presentations given by Imagination Technologies as a biased source.

Also, Carmack knows his onions, did you bother to take in what he had to say about ray tracing and rasterization? What about what he said about overdraw reduction in existing rasterization technology? Did you read it? No of course not, it doesnt fit your narrative so it is hand waved away.

Lastly, learn to sarcasm fella

 

[Pottsey]

Are you aware speakers have to be vetted by an advisory board made up of seasoned industry professionals? All speakers have to submit a proposal on a subject that is none biased, high-level quality and relevant. Spaces to speak are limited and only a limited amount of submitted proposals get allowed though.

You cannot just turn up on stage and sprout biased PR rubbish. Strictly no vendor-related submissions are permitted and no PR is allowed.

http://www.gdconf.com/conference/c4p/summits.html
http://www.gdconf.com/conference/c4p/
http://www.gdconf.com/aboutgdc/advisoryboard.html

Are you seriously telling me that was biased? Can you even point out one thing that is remotely biased about it? What about all the ray tracing talks not done by PVR are they biased as well, if so how so? Did you notice the Q&A session where those leading industry experts in the audience could question anything that didn’t make sense or looked wrong?

Furthermore those are detailed technical talks. PVR didn’t just go “because I said so” they didn’t go la-la-la we are right. They explained in technical detail why they said what they did. They gave examples and demonstrations with whitepapers. They explained precisely in detail why what you said is flat out wrong. There were no bias, only pure facts and technical details. So what you said is flat out wrong. .

“Also, Carmack knows his onions, did you bother to take in what he had to say about ray tracing and rasterization? What about what he said about overdraw reduction in existing rasterization technology? Did you read it? No of course not, it doesnt fit your narrative so it is hand waved away.”

It fits into my narrative perfectly. Can you explain how it doesn’t? How does anything he said go against what I said? Please point it out in detail. Carmack said they have good overdraw reduction. Good is not the same as perfect overdraw reduction that all TBDR run at. Not just PVR but all the none PVR run at 99% reduction. TBDR are well known in the Industry to have vastly improved overdraw reduction. Its one of the main points of TBDR and its why so many GPU company use TBDR.

As for Carmack are you aware he wrote that ray tracing stuff before he know about PowerVR?. He redacted what he said after finding out about PVR ray tracing. He then went on to say “I am very happy with the advent of the PVR Wizard ray tracing tech. RTRT HW from people with a clue!

https://twitter.com/id_aa_carmack/status/446021820290842624

John Carmack ‏said “In the long term RT will win,as it is winning in offline. The near and mid term is the hard part.”

John Carmack said “yes, that is a very important point, and IMG was very smart about leveraging the existing GPU hardware.”

 

[layte]

No Pottsey, you are taking a tech talk from PVR and treating it as the unbiased total truth. Stop it, you cannot use a vendor's tech talk as proof of their technologies superiority. So yea, you are basically telling everyone that PVR is the best, because they said so.

As for the overdraw, it does not have to be perfect, just good enough, which surprise, surprise everyone states it is. TBDR got crushed on the desktop, and the same thing will happen now the big GPU vendors are turning their attention to mobile, even the other mobile focussed GPU vendors do not touch it. As testing has shown, it's not some secret weapon or amazing feature, just a different methodology, which if it truely offered such advantages as you seem to think would surely be used by another vendor (as another poster pointed out, once again NV own significant IP in this area), yet nothing...

I did note you cherry picked the quotes from that page. People seemed to be very offhand about it's ability to be viable without pickup from NV and AMD, and note the quote "in the long term" so the comment regarding its current validity stand. So we are back to RT being viable years down the road, just like it always has been, so not this PVR wonder chip you are betting the farm on, and leveraging existing GPU hardware, so no secret sauce.

Lets take a look at PVR's hybrid RT solution: https://www.youtube.com/watch?v=LyH4yBm6Z9g Not exactly a looker is it. That's the problem, you either get terrible performance, or poor visuals.

 

[Orangey]

you cannot use a vendor's tech talk as proof of their technologies superiority

Then why is everyone fawning over Nvidia every time Jensen holds up a new GPU?

 

[layte]

b'atches love dem wood screws.

 

[Pottsey]

“Stop it, you cannot use a vendor's tech talk as proof of their technologies superiority.”

First what about all the other vendors who said the same thing? Look a GPU engineer submitting a none vendor technology paper to a board of seasoned industry professionals who vetting it to make sure is about as unbiased as you can get. The paper was cross checked by a large team of industry professionals some from competing company’s and all of them agreed with the paper and allowed it to be given. None of the AMD or NVidia GPU engineers had a problem with the paper. None had a problem in the audience and questions it.

So we have the big name GPU engineers and seasoned industry professionals from the big name vendors saying the same thing. But no they are all wrong because you said so. Where is your proof?

Since it appears the entire professionals industry is wrong and you are right can you show us your proof?

“So yea, you are basically telling everyone that PVR is the best, because they said so.” .

Now who is sticking their fingers in their ears and going la-la-la. You have been given in full technical detail why you are wrong. The technical information was cross checked by seasoned industry professionals who all agree. The talk was not about PVR technology so you cannot use that as an excuse.

For some reason you are so against being wrong or admitting a mistake that you are going against what the entire seasoned industry professionals say!!!. Stop and think about that. You have no evidence, no facts; the professionals say you are wrong. Full facts and technical details on why are wrong have been given. But you ignored the facts and technical details and are now going into a crazy far fetched fantasy to desperately try and prove that you are not wrong.

https://www.youtube.com/watch?v=LyH4yBm6Z9g

That’s about 100x faster then what anyone else has shown. That was done in a slow speed test kit. That’s why it’s so slow it’s a proof of concept. Once you get hardware you get that quality and the speed.

“So we are back to RT being viable years down the road, just like it always has been, so not this PVR wonder chip you are betting the farm on, and leveraging existing GPU hardware, so no secret sauce.”

The first full hardware arrives as a development platform in Q4 2015, products after that. Then you have full speed and high quality.

“which if it truely offered such advantages as you seem to think would surely be used by another vendor (as another poster pointed out, once again NV own significant IP in this area), yet nothing...”

Last time I looked because of its advantages almost all the main vendors did switch to TBDR or at least moved to being tile based. So many switched to TBDR or versions of it there are more using TBDR then not. What does that say to you? Why did what was it 80% of the market switch if it’s not better? (80% is an estimation from market share so might be a little off)

NV do not own significant IP. The IP was never tested in a user end product and more importantly the IP was never finished. The IP is 20+ years outdated and has not been worked on to keep up-to-date. It doesn’t solve any of today’s problems. The IP couldn’t run a modern game as it lacks just about everything needed with it being destined in the mid 1990’s and it was never finished. So yes NV do have some technology in that area and it’s a starting point but it’s a long, long way from useable in today’s world.

 

[Stanners]

Then why is everyone fawning over Nvidia every time Jensen holds up a new GPU?

Dunno, what do you feel when a new AMD card is held up. I would imagine you can apply that.

 

[layte]

Edit:

I'm tired of the constant back and forth, I just can't keep posting why you are wrong to take everything PVR state as the unbiased truth over and over. Let's wait and see if PVR revolutionary technology does indeed shake the GPU industry to its core.

 

[Orangey]

Dunno, what do you feel when a new AMD card is held up. I would imagine you can apply that.

LOOOOL

I don't fit into your paradigm.

 

[Pottsey]

Edit:

I'm tired of the constant back and forth, I just can't keep posting why you are wrong to take everything PVR state as the unbiased truth over and over. Let's wait and see if PVR revolutionary technology does indeed shake the GPU industry to its core.

Well first I understand what is being said so I can see its not biased. Second I am not blindly taking what PVR state an unbiased truth. I am taking what all the GPU engineers from all the different company’s say and confirm amongst each other. We have the leading GPU engineers agreeing then you saying they are all wrong. You are not just disagreeing with PVR but all the other GPU professionals as well. Pretty much the whole world knows when complex stuff is thrown into a rasterizer complexity tend to grow faster than raytracers. I.E. reflections, shadows etc. Yet you insists the world is wrong and you are right all with no evidence.

What about all the talks and whitepapers which are not from PVR saying the same thing? What about all the other companys? You have been wrong time and time again in this thread how many more crazy storys do you have to come up with before you admit you made mistakes? You have been proven 100% wrong on a number of points yet you refuse to acknowledge any mistakes. That’s something you need to take a look at that. There is nothing wrong with getting something wrong. So why do you keep inventing all these crazy excuses and story’s? We all get things wrongs even me like when I thought ray tracing would be closer to 5 years then 7.


EDIT:
Some possibly interesting videos on ray tracing.

https://www.youtube.com/watch?v=jUMPmDn_ZE8
https://www.youtube.com/watch?v=AXsaxagKWUo

 

[Ayahuasca]

company's, story's

Hard for anyone to take anything you say seriously when you can't do basic english.

Your sig says everything we need to know anyway.

Until we see this wonderful ray tracing being useful who the **** cares what some papers and professionals say.

 

[Pottsey]

Lots of interesting Ray Tracing talks due at developer day CD 2015 in March. So far I spotted the following ray tracing ones, apart from the last two which are just interesting in themselves but not Ray Tracing.

http://www.gdconf.com/

Keynote on latest development and future plans (10:00 am – 11:00 am)
This includes the Wizard ray tracing hardware and details on the APIs and extensions that will be supported.



Enhancing Traditional Rasterization Graphics with Ray Tracing (11:00 am – 11:45 am)
“Hybrid Ray Tracing is a novel technique combining traditional rasterization pipelines with ray tracing elements. In this session we will describe how ray tracing can be easily integrated into existing rasterization pipelines to render effects that are difficult to achieve efficiently with modern raster graphics.”




“Low Overhead Probe-based Global Illumination Using Ray Tracing (1:00 pm – 2:00 pm)
This session will show attendees a novel technique for producing real-time dynamic global illumination with a very low performance overhead. The technique uses a relatively limited amount of ray tracing to produce the global illumination, and should therefore be suited for mobile applications.




Hybrid Ray-traced Shadows for DirectX 11 & 12 Senior Developer Technology Engineer, NVIDIA
“Conventional shadow mapping has its pros and cons, this hybrid technique combines the best of shadow mapping with ray-traced shadows using conservative rasterization.



Next Generation Graphics – from PC to mobile (3:30 pm – 4:00 pm) presented by Tencent Interactive Entertainment
“Mobile devices usually have a different graphics architecture and many restrictions on 3D rendering, and many mobile games have a much worse display than their desktop counterpart. In this session we will introduce the porting process of our in-house game engine from PC to mobile without losing most of graphics features. We will talk about some difficulties we met and tricks we use.”



Creating Truly Scalable Game Engines – Console to Mobile (4:30pm – 5:30 pm)
“In this session, we will discuss the challenges of developing a truly scalable game engine – current gen console to embedded devices – with our panel of industry veterans. How does the wide range of device capabilities affect art pipelines? Do the reduced-overhead APIs, such as Apple’s Metal, enable features that would not be possible with traditional APIs? Is the cost of supporting an increasing number of graphics APIs becoming a development bottleneck? Is it feasible for small studios to take on this challenge, or is it only cost effective for large-scale game and middleware development?”

 

[Pottsey]

The PowerVR Presentations from GDC 2015 are up, still waiting on the videos. Lots of ray tracing information.

http://community.imgtec.com/developers/powervr/presentations/

The Wizard ray tracing GPU has made the shortlist for best tech innovation (http://www.theinquirer.net/inquirer/feature/2400814/inquirer-unveils-tech-hero-awards-shortlist)