Slashdot Mirror

Why don't you ask Carmack? Back in 2000 he talked about how, although voxels are interesting, hardware support for polys is just too good to ignore. In fact, the hardware support is so good that even voxel engines usually go through the rather expensive process of converting their voxels into polys in real-time because they'll make up that speed penalty by leveraging the poly drawing power of todays graphics cards.

Although everybody focuses on how easy it is to get correct shadows, reflections, and refractions with a ray tracer, another interesting fact is that ray tracers don't really care how much geometry is in your scene. That means that once you get to a certain level of detail, ray tracers can become more efficient than rasterizers!

There's one big flaw, though. Ray tracers need a fast way to intersect each ray with the world geometry. This is a problem, because creating some way to index all your geometry is really expensive, especially if you have to do it in real-time. We need some clever data structures and algorithms to solve this problem.

So, how is Carmack solving it? Well, you'll be happy to hear that he's bringing back voxels!

Ray tracing may be the future of rendering--it's definitely part of the future at least. There is an old joke that goes "Ray tracing is the technology of the future and it always will be!". I don't think that's completely true, but I do believe that ray tracing is not the answer--it's part of an answer. Furthermore, I believe that the C/C++ language programming interfaces for GPU computing are useful for programming ray tracing to run on GPUs. Over time, I expect the graphics APIs will evolve to embrace ray tracing as part of the 3D graphics "bag of tricks" for making images.

Stanford and Nvidia are working on GPU folding also GPU folding has so far been much more buggy than cpu folding. Q6600 are great folders, you should give it a try before calling it a waste of potential!

Yeah, he came out against Intel's version of raytracing, but he has his own raytracing ideas going on as well. Here is an article:
http://www.pcper.com/article.php?aid=532
A relevant quote:
"I have my own personal hobby horse in this race and have some fairly firm opinions on the way things are going right now....I do think that there is a very strong possibility as we move towards next generation technologies for a ray tracing architecture that uses a specific data structure; rather than just taking triangles like everybody uses and tracing rays against them and being really, really expensive."

Despite what some major 3D game engine creators have to say if real-time ray tracing comes sooner than later, at about the time an eight core CPU is common, I think we might be able to do away with the graphics card especially considering the improved floating point units going in next gen. cores. Consider Intel's QuakeIV Raytraced running at 99fps at 720P on a dual quad-core Intel rig at IDF 2007. This set-up did not use any graphic card processing power and scales up and down. So, if you think 1280x720 is a decent resolution AND 50fps is fine you can play this now with a single quad-core processor. Now imagine it with dual octo-cores which should be available when? Next year? I hazard 120fps at 1080P on your (granted) above average rig doing real time ray tracing some time next year IF developers went that route AND still playable resolutions and decent fps with "old" (by that time) quad-cores.

Pretty good read; interesting that this guy is talking to press a lot more:

http://www.pcper.com/article.php?aid=530

Must be part of the "attack Intel" strategy?

A couple more reviews that aren't as, um, positive:

http://www.pcper.com/article.php?type=expert&aid=550&pid=2
http://www.techreport.com/articles.x/14606
http://www.bit-tech.net/hardware/2008/04/23/amd_phenom_x3_8750/1

Some of the other reviews to check out:

http://www.pcper.com/article.php?type=expert&aid=548&pid=2
http://www.techreport.com/articles.x/14583
http://www.maximumpc.com/article/the_new_fastest_hard_drive_ever

"Eg; writing a game engine with a video thread, audio thread and an input thread still leaves 13 cores idle. You really cant thread those much farther (the ridiculously parallel problem of rendering is handled by the GPU)."

Woopsie. I think you presume that games don't need more processing before the GPU so much.

What if you could thread out, and preprocess the video? We don't know, cause it's not yet practical. The tools to write that software don't exist.

Actually, if we get enough cores as CPU, when do we start to need less GPU? The real question will be if the multi-cored CPU feeding the GPU frames is the answer, or is Intel's http://www.pcper.com/article.php?aid=534&type=expert&pid=3 GPU vision the answer?

I suspect we could turn video card development on its head by processing the raster on a 16-32 core CPU array, leaving 1 core for audio, 1 core for game engine, 1 core for AI, and 1 core+ for physics.

But the bottom line is that games will drive multicore development, with imaging/graphics/video (same thing as games, really) also on board.

That may sound like a good idea, but what about the fact that Carmack says the stuff is a waste (and isn't planning to program it in)....let alone we're talking Intel integrated graphics, which is not stuff that gamers use. So where is this magically supposed to make a difference other than break compliance?

Personally, I'd have people pay more attention to textures and make things more efficient than concentrate on a new shiny method to improve shadows. Shadows are the first thing to go from non high end systems. Does any part of intel graphics sound like high end systems? No, and it never will. So who is going to really use this stuff? About scalability: all graphics have scalability. Ray tracing is low on that priority level. With 32 GPUs, things like textures will come first, not ray tracing, not rasterization, and not anything involving shadows.

The wiki link talks a lot more about Rasterization (current method) than Ray Tracing. Ray tracing has its potential uses, but we are WAY way off from it being necessary or something.

David Kirk, NVIDIA: I'm not sure which specific advantages you are referring to, but I can cover some common misconceptions that are promulgated by the CPU ray tracing community. Some folks make the argument that rasterization is inherently slower because you must process and attempt to draw every triangle (even invisible ones)--thus, at best the execution time scales linearly with the number of triangles. Ray tracing advocates boast that a ray tracer with some sort of hierarchical acceleration data structure can run faster, because not every triangle must be drawn and that ray tracing will always be faster for complex scenes with lots of triangles, but this is provably false.

There are several fallacies in this line of thinking, but I will cover only two. First, the argument that the hierarchy allows the ray tracer to not visit all of the triangles ignores the fact that all triangles must be visited to build the hierarchy in the first place. Second, most rendering engines in games and professional applications that use rasterization also use hierarchy and culling to avoid visiting and drawing invisible triangles. Backface culling has long been used to avoid drawing triangles that are facing away from the viewer (the backsides of objects, hidden behind the front sides), and hierarchical culling can be used to avoid drawing entire chunks of the scene. Thus there is no inherent advantage in ray tracing vs. rasterization with respect to hierarchy and culling. /blockquote

PC Perspective also features an article written in January on the impact of ray tracing in games. It provides many pictures of what it will look like and what the benefits are vs rasterization. It's written by Daniel Pohl, a research scientist at Intel.

http://www.pcper.com/article.php?aid=532&type=overview

Intel has this article about the hardware needed to run at 50fps at 1920x1080p. They're claiming you need 8 cores. In a couple of years, that could well be within reach for most gamers.

There's also this John Carmack Interview. Carmack isn't too optimistic about ray tracing replacing rasterized graphics.

An interesting read on this very subject here. Quote:

"I have my own personal hobby horse in this race and have some fairly firm opinions on the way things are going right now. I think that ray tracing in the classical sense, of analytically intersecting rays with conventionally defined geometry, whether they be triangle meshes or higher order primitives, I'm not really bullish on that taking over for primary rendering tasks which is essentially what Intel is pushing."

Carmack admits he has his own personal preference, but generally he's pretty sensible about these things. He's usually called it correctly in the past when people have pushed various technologies that were supposed to take over the world, and they've fallen by the wayside.

Hopefully he'll chime into this latest article with some further thoughts.

On average the quad-gpu Radeon solution is definitely quite a bit slower. Xbit used a rather bizarre assortment of games to reach those results. If you take a look at the two reviews from hardocp and pcperspective, the title of "fastest quad-gpu setup" goes to nVidia. This comes as no surprise given that nVidia dominates the single and dual gpu configurations of the same underlying technology (i.e. G92 is quite a bit faster than RV670).

From HardOCP.com's conclusion:

From PCPerspective.com's conclusion:

There are some other good looks at RS780 performance:

http://www.pcper.com/article.php?aid=527 - looks at Hybrid CrossFire with several games in real world testing as well as GPU overclocking; also features the new AMD X2 4850e processor
http://www.techwarelabs.com/reviews/processors/780g-and-4850e/ - looks at both the chipset and CPU
http://techreport.com/articles.x/14261 - good motherboard review
http://www.bit-tech.net/hardware/2008/03/04/amd_780g_integrated_graphics_chipset/1 - tests HQV and HD audio systems

Ooh I love this game:

http://www.pcper.com/images/reviews/511/isaiah_arch.jpg

How about this? Or this?

These are just proof of concept things, keep in mind. A couple researchers who don't necessarily have time to add in a bunch of special effects and make it look good, and on hardware that will be available to consumers in only a couple of years. And as explained in the article (although, honestly, I take it with a grain of salt) ray tracing only becomes logarithmically more difficult to render as you add more to the scene rather than linearly more difficult, plus you don't have overhead of portal calculations or precalculation of visibility occlusion, and physics detection and AI object detection can both be done using rays simultaneously to rendering (read: at almost no additional cost).

Seriously, you took a ray tracing screenshot from FOUR YEARS AGO and tried to compare it with a game that came out less than 6 months ago. And not only that, but a doctored publicity screenshot of crysis that doesn't even represent how the game looks. Trust me I worked at EA, I know how those publicity screenshots are made, and it has nothing to do with real time rendering.

The upper screenshot looks quite boring, because it has no lights and no shadows in it. In the lower example each pixel is being lit by an average of 3-5 light sources simultaneously.

Screenshot

If we use ray tracing to render all the shadows accurately, then each pixel would ordinarily require one visibility ray and 4 shadow rays. If we were to use rasterization to achieve the same quality image, it would only save us the 1 visibility ray per pixel. Rasterization techniques simply are not robust enough to achieve the same kind of accuracy and fidelity as ray tracing for shadows, reflections and refractions. If you imagine a trend toward higher and higher quality, the savings of using rasterization for the visibility determination trends towards insignificance.

The quote is referring to Quake 4. So they already can raytrace a semi-modern game at 90 FPS, and they have a graph that very clearly shows raytracing at a performance advantage as complexity increases. Just look at the damn graph (page three),

I don't have to look at the damn graph to tell you that what people are going to want is this

And what they are going to get is this

And, they should just be happy with this (which, is pretty awesome)

My point is that real time photorealistic rendering will never catch up with what people expect from their games. It will always be behind. If all you want is mirrors, then find a faster way to implement them at the expense of a bit of quality.

According to the article, incorrect. Read the 3rd page: http://www.pcper.com/article.php?aid=506&type=expert&pid=3

FTA:
"...The Intel X48 chipset is a refresh of the X38 chipset aimed at the high end desktop market. It will be the first chipset to support 1600 MHz FSB parts (though current boards do as well in some cases) and will have unlocked bus ratios for improved overclocking ability. So there really isn't much change from the X38 chipset -- and in fact most X38 motherboards aimed at the enthusiast will probably support 1600 MHz FSB processors anyway. For my testing I used the Asus P5E3 Deluxe motherboard based on the X38 chipset to run the QX9770 and it ran without an issue.... http://www.pcper.com/article.php?aid=484

Sounds like many existing Intel X38 chipset mainboards will work with the QX9770, and I'd bet Intel's DX38BT can run it, (but probably at FSB 1,333MHz) http://www.intel.com/products/motherboard/DX38BT/index.htm

But ... but ... check out these graphs!

I don't have a clue what "Bioshock" or "0xAA 8xAF" are, but look at the pretty bars! Instead of using a normal box plot, they're using a triple bargraph, with each bar a different color, *and* fancy gradient shading. And it's got a watermark. Let's see OpenGL do that, bitches.

That's how cool DX10 is. It doesn't need performance. Nobody's even looking at the numbers. They're looking at the shading on the bars on our graph.

One of the benefits of raytracing is that it can render real curved surfaces, including NURBS and volumetric surfaces: without taking advantage of this aspect of the technology, would we really see much visual benefit?

Sure. Even without the obvious things you're pooh-poohing like reflection and refraction (and raytracing can produce good results for these on curved surfaces), you get universal shadowing (including self-shadowing, the lack of which really stands out for me in even the best rasterized images), far better shading and bump-mapping, and better illumination... and reduced development costs because so many things that are handcrafted now fall out of the technology.

Look at the second image. Look at the guy in the background. Look how his body beneath the gun is shadowed. You're never going to get that kind of effect from rasterizing. And that makes things look a lot more "plasticky" to me than sharp edged shadows do.

http://www.pcper.com/article.php?type=expert&aid=4 32&pid=2
http://www.hothardware.com/Articles/Intel_Core_2_E xtreme_QX6850
http://www.legitreviews.com/article/538/1/
http://www.tomshardware.com/2007/07/16/cpu_charts_ 2007/

PC Perpsective

http://www.pcper.com/article.php?type=expert&aid=4 26&pid=2

Anyone find others?

http://www.pcper.com/article.php?type=expert&aid=4 02
http://enthusiast.hardocp.com/article.html?art=MTM zNiwxLCxoZW50aHVzaWFzdA==
http://www.anandtech.com/video/showdoc.aspx?i=2979
http://www.hexus.net/content/item.php?item=8593

Big AMD Fan here too:
http://www.pcper.com/article.php?aid=326

Here are some links to other interesting reviews of these products:

http://www.pcper.com/article.php?type=expert&aid=3 92 tested under Vista 64-bit and shows the 8600 GTS behind the aging ATI X1950 Pro
http://enthusiast.hardocp.com/article.html?art=MTM yNCwxLCxoZW50aHVzaWFzdA== tested under XP and shows better performance on the 8600 GTS
http://www.hexus.net/content/item.php?item=8409 tested under XP but not a lot of newer games

Some links about this ASUS bit:
writeup&discussion in french
another writeup, this one's in english
siteadvisor mention
dynamoo blog mention
ithome-tw blog mention

Obviously, don't go to the URLs of the unsafe sites (which are mentioned on a few of these pages) from a vulnerable browser/platform. Be warned.
-os

Look at one of the references in the linked article: http://www.techarp.com/showarticle.aspx?artno=393& pgno=1

A "Vista Certified" device that:

A)Is incredibly difficult to get to install, and
B)Results in repeatable on-boot BSODs, and
C)Is incredibly difficult to get to uninstall, *and*
D)Leaves packages on your HD after uninstall that cause repeatable on-boot BSODs.

Either the Vista (display) driver development process is as much of an after-thought as Linux driver development, or Vista's "NEW AND INNOVATIVE" hardware environment is so incredibly buggy that wrestling with all the necessary work arounds is a very difficult task.

My guess? The new Vista driver model is so overly complex that developers will have a hard time working with it indefinitely. Either development budgets will have to go up (unlikely, for ATI and Nvidia, at least), or hardware release cycles will have to slow. Given that Vista has been in *public* development for such a long time (Betas & Release candidates), I'm guessing there is a systematic problem to driver development that most hardware companies cannot adapt to.

Take a look at this: http://www.pcper.com/article.php?aid=357
"Finally, the complexity of these drivers is simply astounding. Diercks claimed that each of the six drivers that NVIDIA has to develop for Windows Vista is roughly 20 million lines of code long; about as much code as Windows NT 4! While I am sure there is some significant driver overlap between the six separate modules and the 20 million lines on each, projects of that magnitude are something most normal people couldnt even begin to wrap their heads around. "

Consider that Vista contains approximately 50 million lines of code, and took 5+ years to develop. Consider that Linux Kernel 2.6.0 was 6 million lines of code, and contains *thousands* of drivers.

Now, does this mean that Vista driver programmers are simply going to give up, Vista will collapse, and we'll all switch to another OS? Of course not; these companies *will* manage to overcome the overly complex development environment, and will create working drivers. In Time.

What we may see, however, is that Linux drivers will start improving faster than Windows drivers; and I can even potentially forsee a day when the Linux binary video drivers beat Vista drivers to the punch, in terms of properly supporting newer hardware. Architectural problems don't necessarily cause development to fail, but serious organizational difficulties impact release cycle, and result in more annoyance and security bugs.

Yes it does. Scroll down in the PC Perspective article to the FNA=Flexible Network Architecture section - it's about a page down. It talks about how the card is basically a miniature computer, running Linux.

How about reading this.. http://www.pcper.com/article.php?aid=357

Which contains a much more authoritative response from Dwight Diercks - Vice President, Software Engineering at NVIDIA

for BlueGene 367 TFlops double precision http://www.netlib.org/utk/people/JackDongarra/faq- linpack.html#_Toc27885741

no thanks

Yep, that's one of the things that got me excited about it as well. Did you also read this article on ray tracing on the same pcper.com site by a German guy that made a Quake 4 ray tracing engine?

http://www.pcper.com/article.php?aid=334

Blame the US. Nvidia is an American company. ATI's drivers are way more stable even according to Microsoft, this review site and numerous others. Plus, ATI is a Canadian company :) (or at least they were until they got bought by AMD).

Just thought I drop a link to this article that actually looks at current gaming performance on Vista for both NVIDIA and ATI:

http://www.pcper.com/article.php?type=expert&aid=3 54&pid=2

With the knowledge that you can actually transfer the OEM license from machine to machine, just buy the OEM version. http://www.pcper.com/article.php?aid=352&type=expe rt&pid=4

This article did the same thing, but without the need to hack anything or install any drivers on the XP system:

http://www.pcper.com/article.php?aid=325

http://www.pcper.com/article.php?type=expert&aid=3 19

This review looks at gaming and such too, but also touches on the NVIDIA CUDA (Compute Unified Device Architecture), that NVIDIA is hoping will get super computing into mainstream pricing. What thermal dynamics programmer would love to access 128 1.35 GHz processors for $600?

http://www.pcper.com/article.php?aid=319&type=expe rt&pid=5

http://www.pcper.com/article.php?type=expert&aid=3 19

This review looks at gaming and such too, but also touches on the NVIDIA CUDA (Compute Unified Device Architecture), that NVIDIA is hoping will get super computing into mainstream pricing. What thermal dynamics programmer would love to access 128 1.35 GHz processors for $600?

http://www.pcper.com/article.php?aid=319&type=expe rt&pid=5

Why is the submitter only pimping the HotHardware review? Here's more (in no particular order):

HardOCP
Guru3D
Anandtech
Bjorn3D
PCPerspective
nV News

- http://www.madshrimps.be/gotoartik.php?articID=482
- http://www.hothardware.com/viewarticle.aspx?articl eid=861&cid=1
- http://www.hexus.net/content/item.php?item=6538
- http://www.mvktech.net/content/view/3357/48/
- http://pcper.com/article.php?aid=287
- http://uk.theinquirer.net/?article=33872
- http://www.reghardware.co.uk/2006/08/23/review_ati _radeon_x1950_xtx/
- http://www.techpowerup.com/reviews/ATI/X1950XTX
- http://www.bjorn3d.com/read.php?cID=954
- http://techreport.com/reviews/2006q3/radeon-x1950x tx/index.x?pg=1
- http://www.extremetech.com/article2/0,1697,2007324 ,00.asp
- http://www.tgdaily.com/2006/08/23/ati_releases_rad eon_x1950/
- http://www.guru3d.com/article/Videocards/375/
- http://www.hardwaresecrets.com/article/131
- http://www.hardwarezone.com/articles/view.php?id=2 020&cid=3&pg=1
- http://www.firingsquad.com/hardware/ati_radeon_x19 50_xtx_performance/
- http://www.driverheaven.net/reviews/X1950XTXreview /

up to date list: http://www.madshrimps.be/forums/showthread.php?s=& threadid=26526

http://www.pcper.com/article.php?type=expert&aid=2 87

Here the review talks up the signle X1950 XTX card but finds the CrossFire platfrom from ATI still very under-developed.

Looking for another article that isn't down?

http://www.pcper.com/article.php?aid=283

The E6700 we tested here, that bests the FX-62 that is currently selling for over $1000, has a predicted price of $530; nearly half the price!! If Intel sticks with that price, and AMD doesn't drastically lower theirs, the Core 2 Duo line up is going to tear AMD apart.

Since Japan has already hit Intel for anti-competetive moves, can AMD prove illegal dumping?