Larrabee Demo

Water rendered by raytracing? *ROFL*
You mean the REFLECTIONS on the water surface don't you...the water surface is not the point...and they made if "flat" for a reason...you wouldn't be able to see the raytraycing if the surface was filled with hydrodynamics...erhmm scratch that YOU are still not able to the the light ;)


Yes, the reflections :p I somehow deleted the rest of my post mentioning the more "accurate" light interaction with the "water" either way, you can look at the water and see the "ray tracing" and since water is really made up of "reflections" otherwise it's just tinting what's underneath, by looking at the "water" you can see the "ray tracing"

anything else?
 
Yes, the reflections :p I somehow deleted the rest of my post mentioning the more "accurate" light interaction with the "water" either way, you can look at the water and see the "ray tracing" and since water is really made up of "reflections" otherwise it's just tinting what's underneath, by looking at the "water" you can see the "ray tracing"

anything else?

I somehow deleted my reply to you :p
 
I somehow deleted my reply to you :p

I do remember how I deleted mine :D I added some non-relevant info and removed it before posting, and didn't bother re-writing the "more accurate light interactions" part that was relevant.

Either way, essentially the water is where you are going to look in that demo to see the effects of ray tracing. and technically, you can use ray tracing to light the entire scene, since all we see are light reflections. I just wonder if it's possible to do with Larrabee
 
Just that Intel decided to not go custom and instead of cram an x86 (PII?)-derived design into it. Or so we have been told.

More like a 64-bit P54C Pentium core with a 512-bit wide vector unit strapped on to it.
 
Pretty much. Still seems more complex than the SPs in other GPUs, but hey, who's counting? :)

Exactly. :) Paul Maritz among others have pointed out that x86 has a lot of excess baggage, and when designing a GPU I think you would want the ALU/FPUs to be as lean and mean as possible to maximize performance and die space. However you can't beat IA flexibility when doing other things than just drawing pretty pictures. Time will ultimately tell which is the best approach...
 
and they made if "flat" for a reason...you wouldn't be able to see the raytraycing if the surface was filled with hydrodynamics...erhmm scratch that YOU are still not able to the the light ;)

Yes, they made it flat for a reason, which is that if you plugged more sophisticated deformation and normal mapping into intel's water sim, it would collapse into an enormous, aliased mess.

I'm all for ray tracing, and I appreciate the technology, but as phide said this was a pretty poor choice of demo because it struggles to cope with effects which are faster and prettier using existing techniques, and doesn't show much which genuinely requires a ray traced solution.
 
Bottomline is that what they have shown wouldn't have even been impressive 2 years ago. Today it is downright laughable. And if I remember correctly they bought those Project Offset guys so they should have no shortage of decent engine programmers. At the very least show us some tech demo that actually uses ray-tracing for something useful like a Cornell box with radiosity/indirect lighting on translucent objects. Something like this in real-time would have been sufficient:

392210_1252839065_medium.jpg


Instead they show us some busted looking water thats barely on par with any game released since like HL2. Fail.
 
As time goes on, there will be an Intel flavor added to the graphics section.
 
Bottomline is that what they have shown wouldn't have even been impressive 2 years ago. Today it is downright laughable. And if I remember correctly they bought those Project Offset guys so they should have no shortage of decent engine programmers. At the very least show us some tech demo that actually uses ray-tracing for something useful like a Cornell box with radiosity/indirect lighting on translucent objects. Something like this in real-time would have been sufficient:
http://features.cgsociety.org/newgallerycrits/g10/392210/392210_1252839065_medium.jpg

Instead they show us some busted looking water thats barely on par with any game released since like HL2. Fail.

Agreed. Also, something like that in real-time with dynamic lighting and everything would be totally amazing :) I think my company's game engine can come pretty close, but I'm not sure what kind of system it'd require to get 60 FPS :p
 
Bottomline is that what they have shown wouldn't have even been impressive 2 years ago. Today it is downright laughable. And if I remember correctly they bought those Project Offset guys so they should have no shortage of decent engine programmers. At the very least show us some tech demo that actually uses ray-tracing for something useful like a Cornell box with radiosity/indirect lighting on translucent objects. Something like this in real-time would have been sufficient:



Instead they show us some busted looking water thats barely on par with any game released since like HL2. Fail.
Is something like that even possible with any commercial hardware at the moment?
 
Dont get me wrong, this looks beautiful. But, its bullshit. There just rendering ETQW maps with there engine. They just did i simple loop of the map through the engine. Now I would be impressed if i saw standard 3D rotations and manipulations(and all of teh other graphics candy incoperated,but im not sure that if its really something to be used with Ray tracing), AI, Networking, and general game play at a respectable frame rate of 60 fps or higher. Until then, i think this should sit on the back burner.
 
isnt proper ray tracing making the light bounce correctly and hundreds of times based of the surface and its reflectivity is determined by how much light is bouncing off it ? this just looks like a texture that draws everything around it onto itself.
 
Intel's point is to show that they have working silicon. It was not meant as a blow-your-socks off Nvidia/ATI style tech demo.

Think of it as a raytracing "Hello World" program to show that the silicon works.
 
This is entirely unimpressive in every aspect within itself. However if RT can be CORRECTLY implemented in games then it will add that extra lil bit. Still, I'm underwhelmed MAJORLY by this demo. When Intel demos RT on a much more graphically-demanding platform and at a full 60FPS, then I'll be impressed.
 
This is entirely unimpressive in every aspect within itself. However if RT can be CORRECTLY implemented in games then it will add that extra lil bit. Still, I'm underwhelmed MAJORLY by this demo. When Intel demos RT on a much more graphically-demanding platform and at a full 60FPS, then I'll be impressed.

Technically, RTRT has been used since the first 3D engines. Heck, Wolfenstein 3D's engine used a 2D form of RTRT. Rasterization can't work without RT either. In the end it's all rays :p
 
Technically, RTRT has been used since the first 3D engines. Heck, Wolfenstein 3D's engine used a 2D form of RTRT. Rasterization can't work without RT either. In the end it's all rays :p

Noted. I'm a network guy, so the history and particulars on these sorts of things can escape me. So... why couldn't this weak demo of Intel's be done on a GPU already and why is it such a big deal if its already, to a point, implemented in what we're playing with now? This information just makes the demo MORE underwhelming than it already was.
 
Noted. I'm a network guy, so the history and particulars on these sorts of things can escape me. So... why couldn't this weak demo of Intel's be done on a GPU already and why is it such a big deal if its already, to a point, implemented in what we're playing with now? This information just makes the demo MORE underwhelming than it already was.

They were showing people that they have a working chip. That's it.

Their point is not "Hey look at this engine! Doesn't it kick ass?"

Their point is "Hey look at our working chip!"

The CEO of Dreamworks Animation Studio said "we are well on the way of upgrading our software to really take advantage of Larrabee and in terms of speed, flexibility, capacity, it just raises the bar of what we can do by not 2 or 3x, but 20x."

I doubt Dreamworks would be rewriting their rendering software for Larrabee if it was a dud.

Personally, I doubt that Intel will ever write a commercial quality engine of any kind for Larrabee. That'll be up to the game devs. Intel will do a DX11 rasterizer and that's about it.

I think we'll see some innovative stuff for Larabbee come out of academia/research though. Remember those realtime PS3 raytracing demos people were doing a couple years back?

http://www.youtube.com/watch?v=oLte5f34ya8

http://www.youtube.com/watch?v=u6yoz4iL8kQ&feature=related

You'll see stuff like that.

Then there's the chip from Caustic that promises to enable raytracing on traditional gpu's.

http://www.geek.com/articles/games/...ce-card-is-anything-but-conventional-2009076/

It's due in 2010 as well. It should be an interesting year for graphics!
 
Back
Top