[rpg.314]

http://www.brightsideofnews.com/news...ee-fiasco.aspx

After we saw roadmaps for introduction of Larrabee pushed back all the way to 2011,

and according to documents we saw, it won't reach the market in the next 12 months.

WTH is going on with lrb and intel? Are they gonna make their 2010H1 deadline or not? If they slip to H2, they'll quite likely have to put up with 32 nm shrinks of r8xx and prolly 32 nm shrink of gt300 too. Fighting them on a 45 nm chip would be hard, but may be, the delay is because they will go straight to 32 nm, in which case they'll have an advantage if they are able to launch in say Q1.

EDIT:

A bit later I find this,

http://www.semiaccurate.com/2009/09/...out-weeks-ago/



If they have indeed quashed a number of bugs, I suppose they could launch in early 2010.

[fellix]

Most probably Intel wants a smaller die foot print for the monster, to be more competitive on the market and consequently this will give them some extra time to polish the software graphics "pipeline" and the driver model.
After all, there are so mush APIs to validate for.

[compres]

I have high hopes for this project but realistically can't see them outperforming nVidia or ATI in the grpahics only applications.

Would be interesting if we could have larrabee and ATI/nVidia + some nice AMD or Intel CPU next year. The masochists like me can have fun programming 3 processors XD

[rpg.314]

Would be interesting if we could have larrabee and ATI/nVidia + some nice AMD or Intel CPU next year. The masochists like me can have fun programming 3 processors XD

OpenCL FTW

[Ailuros]

Will AMD even use 32nm or will they go directly to 28nm?

[rpg.314]

I think we'll prolly see 32 nm "dumb shrinks". Mixed into the same sku's of 40 nm if need be.

[Ailuros]

Originally Posted by rpg.314

I think we'll prolly see 32 nm "dumb shrinks". Mixed into the same sku's of 40 nm if need be.

What for? I mean there's little doubt that AMD will go for 28nm at Globalfoundries. Is the hussle to change to 32nm@TSMC libraries even for a a dumb shrink worth it?

[SiliconAbyss]

Larrabee is chasing a fast moving target, the chip seems more and more like an albatross. By time time it comes out, AMD might be coming to market with their Fusion GPGPU, and by that time Nvidia will have who knows what.

Intel should stick to what they are good at, bribing OEMs um I mean making CPUs.

[Voxilla]

By the time it arrives, they probably will have ditched the pathetic x86 ISA.

[compres]

I sure hope they ditch the abominable x86 ISA. Why is this needed if the chip will be incompatible with their CPUs anyway?

[rpg.314]

By the time it arrives, they probably will have ditched the pathetic x86 ISA.

To me lrb demonstrates convincingly that you'll have to pry x86 isa from intel's cold dead fingers. They do have an ARM license though. bribing ARM to let them slap a vpu onto a simple in order arm core might be the way to go, after all these promises of programmability.

[rpg.314]

@MODS:

Can you please change the thread title to "Larrabee delayed to 2011?". After all, it is not a fact yet, just Theo's piece.

[itaru]

Intel, IDF 2009 with "Larrabee" a demonstration showing the actual.
After 2010, to introduce high-end graphics card market

http://translate.google.co.jp/transl...hl=ja&ie=UTF-8

[spacemonkey]

So, is the raytraced 'Enemy Territory: Quake Wars' clip the first publicly available image from LRB?

http://www.4gamer.net/games/049/G004963/20090923003/

[spacemonkey]

Better quality here...

http://www.youtube.com/watch?v=G-FKBMct21g

[Nick]

Originally Posted by compres

I have high hopes for this project but realistically can't see them outperforming nVidia or ATI in the grpahics only applications.

Raytracing Quake Wars in real-time really looks mighty impressive to me compared to the spinning cubes I've seen so far from other vendors. At least it shows they have some magic sauce that makes their architecture more efficient. The Radeon HD 4890 versus GeForce GTX 285 also proves that it's not all about the TFLOPS.

Whether the things that give it a massive advantage at raytracing will also help it with classic rasterization is a different question though. And while ATI is doing nothing to combat Amdahl's Law, the rumours surrounding GT300 being a total redesign suggest it still has a chance at stealing Intel's thunder...

[Voxilla]

Originally Posted by Nick

Raytracing Quake Wars in real-time really looks mighty impressive to me compared to the spinning cubes I've seen so far from other vendors. At least it shows they have some magic sauce that makes their architecture more efficient. The Radeon HD 4890 versus GeForce GTX 285 also proves that it's not all about the TFLOPS.

I just looked for a representative video and found this:
http://www.youtube.com/watch?v=mtHDSG2wNho

It doesn't look any better than Crysis IMHO.
Apart from the useless reflective floating spheres it would be hard to tell that this is raytracing.

[Nick]

Originally Posted by Voxilla

It doesn't look any better than Crysis IMHO.

That's completely irrelevant. Other chips simply can't raytrace such a scene in real-time.

It indicates that Larrabee is vastly better at adapting to tasks other than the classic rasterization pipeline. But since the rasterization pipeline has also become highly programmable I expect them to also have certain advantages when rendering Crysis. The only question is whether that translates into an advantage in absolute numbers or not.

[Grall]

Originally Posted by Nick

That's completely irrelevant. Other chips simply can't raytrace such a scene in real-time.

I'd say it's completely irrelevant wether other chips can't raytrace such a scene in real-time if larrabee's raytracing doesn't look one iota better (or run faster) than other chips' classic rasterizing - and looking at this video it DOESN'T.

Apart from the surprisingly fluid framerate this realtime rendering completely underwhelms. The lighting is extremely flat, surfaces look very flat and matte even when close up, the water looks incredibly sluggish and unrealistic (kind of what I'd imagine a sea of transparent mercury would look like).

It indicates that Larrabee is vastly better at adapting to tasks other than the classic rasterization pipeline.

Perhaps. But if it gives no real-world visual improvements (other than more accurate reflections) then it's a pointless advantage. A decidedly ho-hum IQ raytraced scene (at not incredibly high framerates) simply don't push any buttons for me.

Geeking out on realtime raytracing is all good and well for some computing nerds. It might even find a decent niche in some professional market segment. But it won't last in the consumer market if it can't compete with traditional rasterizers in traditional rasterizing titles at the same or better cost/framerate as its competitors. Larrabee has yet to demonstrate it can do that, and considering how long ATI and NV has had to develop the traditional rasterizer (and games devs developing software for them), I'd say intel has its work cut out for it...

It's a very interesting tech though, but the more time passes without larrabee delivering anything substantial makes you doubt the viability of the entire concept.

[rpg.314]

I don't give a rat's ass about programmability if the chip cant deliver better IQ @60 fps than another chip that runs 3Dx.y, whichever way you render it (raytracing, photon mapping, radioisity, rasterization, funny rasterization algorithms.........).

LRB/GPUs/swift shader etc. have to win on better IQ/$/Watt @60fps, PERIOD. Nobody cares about features/programmability if they aren't in D3Dx.y. Unless it delivers on the above metric, it wont sell in enough volumes to sustain R&D on it, even if you are intel. Good luck sustaining it on HPC market alone.

[Lux_]

Let's look at it from another angle: if Larrabee derivate is integrated into every CPU starting from 2012 (providing DX11 support), then all game devs suddenly have new "minimum system requirements". It doesn't matter that the games don't look as great as with addin boards, because the installed base is suddenly huge.

[Ailuros]

Originally Posted by Lux_

Let's look at it from another angle: if Larrabee derivate is integrated into every CPU starting from 2012 (providing DX11 support), then all game devs suddenly have new "minimum system requirements". It doesn't matter that the games don't look as great as with addin boards, because the installed base is suddenly huge.

It probably won't matter either that CPUs will grow by magnitudes in die area and power consumption

[Anteru]

Looking at the Larrabee presentation, I wonder how complicated it is for nVidia/ATI to rasterize the rest, and use ray-tracing only on the water. Both have presented hybrid approaches at SIGGRAPH (with selective ray-tracing), and I would be surprised if they don't provide better performance than what we see here. It's not like Larrabee is magic compared to current GPUs, true, it can run non-graphics workloads (read: stuff which AMD/nVidia cannot show at all), but ray-tracing seems to work on GPUs pretty well already (see OptiX, OTOY stuff), so if they present it as a differentiator, I would expect either rock-solid performance or some really complicated stuff (like AO, to showcase the gather performance, instead of coherent reflections ...)

[Nick]

Originally Posted by Grall

I'd say it's completely irrelevant wether other chips can't raytrace such a scene in real-time if larrabee's raytracing doesn't look one iota better...

It's a tech demo. The first demos of pixel shaders also didn't look impressive at all, but nowadays it's unthinkable not to have support for them. You really have to look at what has been technically achieved here.

I'm not saying we need raytracing support per se, but we do need the kind of flexibility that allows raytracing. Mainstream consumers don't care about rendering Crysis at 200 FPS at 3840x2400. Their monitors (and eyes) only do 60 FPS at 1920x1200. For a long time achieving higher framerates and higher resolutions was the main goal, but this 'fillrate-race' of GPUs is about to end. You can no longer double the number of resources and expect the framerate to double. The individual tasks are getting smaller and the number of dependencies increase.

An architecture like Larrabee is far better at coping with this, and no matter how you feel about the actual graphics that's proven by their raytracing demo. So even if raytracing itself is not the future, Larrabee allows the developers to go in any direction they like, unrestricted by rasterization APIs.

Obviously these things don't happen overnight, so other vendors still have plenty of time to follow the same route once it actually starts to matter. But Intel certainly has a head start.

[Nick]

Originally Posted by Anteru

...but ray-tracing seems to work on GPUs pretty well already (see OptiX)...

Please. Ten pool balls doesn't really equal a Quake Wars scene. Writing a shader that checks for intersection with all primitives in the scene is really not that hard, but you need a really low number of primitives to make it run in real-time. By the way ray-sphere intersection is much simpler than ray-triangle intersection.