Slashdot Mirror

Your own bullshit wikipedia article has "citation needed" for your claim of a hoax presentation.

Yet you believe it wholeheartedly.

Sony, Blu-Ray, FMD?

Try Constellation 3D

http://www.secinfo.com/$/SignIn.asp?ReturnTo=%2F%24%2FSEC%2FRegistrant.asp%3FCIK%3D1080290
SEC info about the company.

http://www.digit-life.com/articles/digeststorage2k1jan/index.html
Read down below about Lite-On and C3D. Oh, and who's Lite-On's bestest buddy at the time? (SONY)

http://www.thocp.net/hardware/fmd_rom.htm
1999 first published in september
2001 first license given to Sony
2002 august C3D disappeared
2003 D Data Inc aquired the patents of C3D en will develop HD DMD

They licensed to Sony and folded.
Then some other no name company was sold the patents.

I think I meant to write "Bit-Matched Playback". It simply means that the sound card plays back with the same bitness and sampling rate as the original sound. Today's Soundcards resample the sound into 24-bit/48KHz before playback, but even the best resampling algorithm introduces errors. With Bit-Matched Playback the soundcard is unable to play back two sounds with different sampling rate simultaneously but the output is more correct.

Look at http://www.digit-life.com/articles2/multimedia/creative-x-fi.html for a very good rundown of the SoundBlaster X-Fi.

If someone reading this thread simply wants a high quality card for playback and gaming, look at the ASUS D2X (PCIe). Digit-life has a fairly detailed audiophile review of the D2, complete with sampling shots, etc. It looks like a good replacement for Creative, and it cost about 2/3 of the equivalent. I do not know about Linux compatibility - I only have it installed on my windows overclocked SLI gaming rig.

5 Watts saved on an expected power usage of between 10 and 25 Watts is pretty significant.

See the power usage specs here: http://www.digit-life.com/articles2/storage/hddpower.html, a bit older perhaps, but not that much.

This old computer fogey completely agrees.

Then: A dozen different video card manufacturers, twice that many chipsets, equal variety of drivers.

Also: The only drivers you got were the ones that came packaged with the card. Oh sure, you could download them if you had a modem, and it was set up properly, and wasn't the card that you needed the drivers for, and the manufacturer had a BBS, and the line wasn't busy...

Then: IDE=slow. Master? Slave? Cable? WTF is this?

Also: 5 MEGAbyte hard drives, FULL HEIGHT. Cable-connection order matters. And remember the joys of TERMINATING RESISTORS? Kids these days have it so damned easy.

Then: Set up your modem to connect to your ISP and hope you don't get any incoming calls. Firewall? What's that?

Also: Modem init strings. ATH0M0S11=50ETC,ETC. Modem connect strings. Busy signals... ALWAYS busy signals. X-Modem, Y-Modem, Z-Modem (now with recovery!). DIP switches, conflicting IRQs... tons of fun.

Then: Scanner? SCSI (and don't forget your terminators). Printer? Parallel. Video in? Forget it. ...NULL modem cables, Joystick ("game") cards, the only quality video-out card was the Targa 3000 and it cost an arm and a leg, a 300 KB file took hours to download...

Then: Steel case weighing 20kg, built out of razor blades.

Nothing more to add here. They really sucked that much.

This was planned for release at the end of 2000 by a company by the name of C-3d. It was called 'FMD':

"The first generation of disc productions from Constellation 3D will be a family of 120 mm multi-layer FM-discs with capacity up to 140 GBytes and with read speed up to 1 GBytes/s."
http://www.digit-life.com/articles/3ddisk/

The semi-rugged laptops are good machines.

I use a CF-Y2, not too cheap, but it's got decent battery life and is lightweight (about 3.5 pounds), not to mention that it survived a fall off my desk with no problems.

Here is another review of the new socket/ddr2 amd part
It's great to look at fx-62 results - it looks like only that processor (or if you overclock it) can
use the available bandwidth
        "Frankly speaking, it's the main competitor who must be bustling now. AMD is doing great anyway. At least in terms of CPU performance. Durability of the K8 core and its capacity to adapt to new market realia is admirable: having lived without major modifications through two process technologies, dual cores, and now a new memory controller, this core meticulously responds to each improvement with performance gains. We were very skeptic about future chances of the new AMD platform against the new processor core from Intel (Intel designed the new core nearly from scratch, while AMD K8 is rather old), but our tests warmed up our interest. The situation may turn out not that simple"

AMD Catches Up in Technology and Shoots Out in Performance
http://www.digit-life.com/articles2/cpu/amd-athlon -64-fx-62.html/

yep, and Core Duo and Solo are just the latest rev of the P6 core that's been in every IA32 chip except the Pentium 4, from the PPro to the Pentium M. In other words, all this news says is that Netburst is dead, and 32-bit computing lives a little longer.

The real new chip line is coming later in the year, when Intel's new architecture comes out: see these 2 great articles by Oleg Bessonov over at Digit Life on Conroe, the future, and Yonah, the current Intel CPU.

Of course, this is Slashdot, so about 3 people will read these through, and only 2 of those will grok 'em, but their server will get melted anyway...

yep, and Core Duo and Solo are just the latest rev of the P6 core that's been in every IA32 chip except the Pentium 4, from the PPro to the Pentium M. In other words, all this news says is that Netburst is dead, and 32-bit computing lives a little longer.

The real new chip line is coming later in the year, when Intel's new architecture comes out: see these 2 great articles by Oleg Bessonov over at Digit Life on Conroe, the future, and Yonah, the current Intel CPU.

Of course, this is Slashdot, so about 3 people will read these through, and only 2 of those will grok 'em, but their server will get melted anyway...

"IA-32 EL is OS-based and is only available after an OS has booted,"
http://www.digit-life.com/archive.shtml?2006/0125

Betcha money it's not any form of Unix.

an Intel spokesperson confirmed that the Montecito platform, which will premiere the company's next-generation 64-bit Itanium architecture, will dispense with executing all 32-bit instruction set applications on-die, prompting customers to opt instead for software-based emulation which Intel promises will be faster anyway.

Here are a couple of actual "reviews" comparing a broad sweep of video cards:

Digit-Life's 3Digest

Tom's Hardware's VGA Charts

Anyone know of any others? One of the big problems in the hardware review site industry is that they all review the same stuff and duplicate one another's work 100 times over (for various reasons which I won't go into), while you'd be hard pressed to find a single review of many low-mid range cards. Even if the purpose of such reviews would simply be to inform people about how poorly they perform, it's a major oversight. There is still a heavy bias toward high-end stuff in the above linked reviews, but at least there are a few low-end and mid-range cards chucked in.

P.S. Another pity is slashdot's poor editorial standards, accepting the description of the linked article as a "review" being the latest example. I guess I could just stop visiting, but then I'd miss out on all the insightful comments from visitors who actually do produce some worthwhile content. So I just block the ads, so as not to reward the editors' laziness.

No way you can get 100Gb/s, no PC interface can handle 10GB/s. Hell most PC's don't have that kind of bandwidth from the CPU to main memory (DDR2-533 maxs at around 6.4GB/s as seen here). Actually after a little searching the Internet 2 speed record is more like 7.21Gb/s for IPv4 traffic.

How about just get a Pentium M motherboard? Newegg has one, it's a little pricey as $250, but that's not super unreasonable. Power usage for the slowest Pentium M that Newegg carries is 21W nominal. The biggest problem you are going to have is the HDD's. It seems no one makes 5400RPM desktop HDD's anymore. So spinup power requirements are 16-30W per drive! A RAID 5 array with 3 more expensive drives is obviously superior to the 4 smaller, cheaper drives I normally advocate.

Well, there are some tests with A LOT of graphics card, like this (a year old one, don't know about newer tests): http://www.digit-life.com/articles2/over2k4/index. html

Here's my chance to blab about the PVR I built myself. It's not pretty, but it runs great.
Here are the specs: Leadtek WinFast PVR 2000 TV/FM tuner card; P4 2.8E / ASUS P4P800; onboard sound; 512MB RAM; 80GB + 120GB HD; WinXP Pro.
The software I built uses: Windows Media Encoder SDK; Visual Basic 6; PHP; FireBird; Apache.
Using VB, I wrote code that goes to Zap2It and downloads 12 days worth of TV show programming and parses it into my FireBird DB. From there I have a web front end that lets you search/sort though shows. You can choose to record one show or create a rule that would record a certain show every time it's on. It also handles scheduling conflicts by prioritizing rules and doesn't record a show if it's been previously recorded.
The back end is a VB app that runs all the time and checks the FireBird DB for the next show to be recorded. When it finds one and it's time to start recording it issues a command line request to the Windows Media Encoder to start recording on channel x for x number of seconds. The size and audio/video bitrate are set using the encoder's profile editor.
The profile settings I use consist of: Windows Media Audio 9.1/Video 9; VBR quality base of 90 (usually has a video bitrate of just over 1000kbps); Video size 320 x 240. At these settings the CPU uses about 20% and 1hr worth of video is about half a GB.
I play the shows by streaming them to the Xbox running xbmc.
I also have a command line script that runs every night and deletes any shows that are older than 15 days. If I haven't watched it by then, it's not worth watching.
This setup has worked great for me for the last year. The next step would be to replace the whole setup with MythTV. I'd have the back end on my computer and the front end on the Xbox.

Adjacent Cache Line Prefetch (Enabled/Disabled) - to enable/disable the adjacent cache line prefetch mode. When disabled, only one 64 byte line from the 128 byte sector is prefetched (which contains the requested data). When enabled - both lines are prefetched no matter whether they have or have not the requested data.

All but the lowest-end (and the 1U rack mountable) servers from both HP and IBM come with PCIe (don't know anything about Dell servers because the company I work for can't resell them). PCI-X has potential for higher speeds, but cross-talk is becoming a problem for that bus. PCIe will be doubling in speed with the PCIe 2.0 spec which should come out fairly soon, and future speed increases should become possible as technology improves. With ethernet controllers now pushing the 10Gbps mark, and serial attached SCSI hitting 6Gbps mark. PCIe server hardware is rare right now, but that will change. PCIe desktop hardware (aside from PCIe 16x video cards) is still pretty rare, too.

The world is moving back to serial. From USB to SATA/SAS to PCIe, these are all serial technologies replacing traditional parallel ones. If everything goes well, the result could be that desktops and servers become more converged on standards than they are now. Being able to use a large array of SATA drives attached to a SAS controller when you're more concerned about storage density than speed, and getting a single bus interface on both server and desktop so economies of scale kick in to lower prices are goals of each of these groups.

I just got a ASUS DiGiMatrix, it's not only silent but it fits in with my stereo components. I just got it and haven't worked out all the features but it's been working well as a normal component in my network running headless without an issue. All the software is Windoze so I put XP Pro on it and manage it from my linux box w/ Terminal Services. Take a look.

when you can get a silent Athlon 64 in a tiny Shuttle XPC SN85G4. I bought one bundled with an Athlon 64 3000+ and it's quietest, fastest desktop I've ever owned. Suse justs hums along silently on this thing. The proprietary Shuttle cooling system is silent and effective. The DVDRW is the loudest thing on this system. Outpost.com is selling the deal I got for $379 Add memory, hard drive, CDROM, and the 64 bit OS of your choice (Suse is flawless) and you're in business.

my portable linux machine has a 37-key keyboard ... but i don't use it much. easier to just ssh to it when i need to type something... ;)

Umm - not exactly. What you are describing is commonly known as a TEC, or "Thermoelectric Cooler" (and also known as a Peltier cooler). They are not composed of a single Peltier junction, but rather a large multitude of such junctions in what is basically a semiconductor package. A TEC is basically a semiconductor-based heat pump, which uses electricity to move heat from one side of the TEC to the other. In the process, the TEC also generates a fair amount of heat (no such thing as a "free lunch") from the hot side, while also consuming a fair amount of energy (which most of it is converted to heat, the rest moving heat from the cold side to the hot side). Finally, remember that you can't generate cold - coldness is the absence of heat.

Now, as I have noted, TECs work via the use of multiple Peltier junctions. Peltier junctions are fairly simple devices, consisting of a junction made between two dissimilar conductors. TECs are made in this manner, but in the same way as ICs are made (more or less). On a larger scale, though, a Peltier junction is easy to make - get a piece of alluminum wire, and a piece of copper wire, and twist them together: congratulations, you have just made a simple (though very inefficient) Peltier junction (however, this is not what can cause house fires in older homes with alluminum wiring - this problem is caused by a combination of alluminum and copper wire junctions heating up because of the resistance of the junction, thus starting a fire).

Peltier junctions work by using something known as the Peltier Effect, which was observed in 1834 by Jean Peltier 13 years after the Seebeck effect was first discovered. The Seeback effect is basically the inverse of the Peltier effect - in that two disimilar conductors, in the presence of a heat source, generate an electrical current. This effect has been used in industry to detect and measure extremely high temperatures, such as what is generated in various industrial furnaces. It has also been experimented with as a method to use the sun to generate electricity (ie, use a solar furnace to focus the sun on a bundle of the junctions) - there is actually an old Popular Mechanics article from the 1950's or 1960's showing how to build such a device.

As to whether you can use a TEC to exploit the Seebeck effect in a practical manner - probably not, as they aren't designed to work in this manner, and you might destroy the device. However, these devices are cheap enough, and if you supplied an appropriate cooling system for the device you might get a bit of electricity out of it - just don't expect much. It would make an interesting science fair project for the kids, though (grab a TEC, a fresnel lens, and a very large heat sink with a fan, bolt it all together and hook it up to a cheap voltmeter, then set it in the sun).

BTW - where did all the real geeks on this site go to, anyhow?

Show some evidence for "SPEC is subject to all kinds of problems." The only recent spec controversay is Apple's dishonest G5 benchmarks, your paranoid anti-Intel conspiricy theory aside, and the fact that you bring up GCC shows you know nothing, since only 4/14 specfp programs are C, the rest are fortran. Also, have a look at comparison of Intel GCC and PGI compilers. Hand coding?!?! Are you insane? Do you have any idea of the complexity of a modern processor's scheduling? Go talk to a comp arch professor, and get bitchslapped for retardedness. If GCC sucks for the G5, then use something better, nobody's forcing you to use it.

Ah, I see what the problem is. Steve Jobs Reality Distortion Field has given you selective reading comprehension. "LINPACK has been largely superceded by LAPACK which has been designed to run efficiently on shared-memory, vector supercomputers." Note that it was designed to run efficiently on shared-memory, vector supercomputers, therefore providing a better benchmark of said supercomputers superness. The reason Linpack is used is because clusters would completely choke on this test. All of this however, completely ignore the fact that you're trying to use a test of a supercomputer to decide what a processors worth.

Linpack is a one class of algorithms, SPEC is 14 fp and 12 int classes, selected to provide a *wide* variety in algorithms.

Find a CPU, not supercomputer, benchmark that shows the G5 is faster, otherwise it's just your word that your Andersonesque calculations are worth more than the accountable SPEC.

But all this really doesn't matter because of one question, if the top500 results prove the G5 is so much better, why didn't Apple use them?

VLIW really is like another layer of abstraction designed to simplify out-of-order execution from the processor point-of-view. You pass in a block of RISC commands that are compiler optimized to not stall the pipeline and execute them in parallel. Intel basically made a VLIW for CISC called EPIC (Explicitly Parallel Instruction Computing). A fairly detailed analysys of the architectures is here

The benefit, as you've said, is that you can reduce hardware complexity by removing the instruction reordering and branch prediction units typical of RISC processors (like PPC or Alpha) and force those tasks onto the compiler, which, in turn, reduces the heat and power consumption of the CPU. It does make code larger, as the compiler has to duplicate and rearrange code to most efficiently optimize for the instruction words.

but you have to keep in mind, that the first generation of minidiscs came out long before (in 1992 according to this site: http://www.digit-life.com/articles2/sonymzn1netmd/ ) mp3's were that popular, as they are now.
i bought my first md (sony's model R30) came out in 1996 - and was a real innovation regarding it's smaller size compared to portable cd-players!
ok, so you had to reencode to attrac, but: you have to rip a cd alike to get it onto your mp3player of choice! so, IMHO reenconding is not a point to hold against md's...

but i agree that with sony locking the format, they made it unpopular to big (internet!) communities - and therefore killed the chances of being "the new" state-of-the-art in music-formats.

You make it sound like Japan hasn't had blu-ray drives and set-top recorders since April, 2003.

This kind:

http://www.digit-life.com/articles2/pda/t-mobile-m da-iii.html

:-)

Abit: http://techreport.com/etc/2004q2/abit-tour/index.x ?pg=1 ECS/PC-Chips: http://www.anandtech.com/mb/showdoc.html?i=1891&p= 1 AOpen & Asus: http://www.digit-life.com/articles/mobofactories/ Chaintech: http://www.sharkyextreme.com/hardware/articles/cha intech_taiwan/

I think an advantage to the consumer is an external drive that behaves as closely as possible to an internal one. It seems that firewire performs better:
http://www.digit-life.com/articles/usb20vsfirewire /

Given the option I'd get an HD with both interfaces and use firewire as much as i can.

For mp3 to sound good, you need to use a good encoder (LAME), VBR, and joint stereo. And when those settings are used, its universally accepted that the maximum quality attainable by MP3 is higher than that for OGG. (and MPC beats them both). But OGG is definately a more effecient format (destroys mp3 below 160K), so given enough time/effort to fine tune the psychoacoustic models, it will one day surpass mp3 at high bitrates.

Don't know about the "universally accepted" hand-waving, but I had a hunt for your source and found a good analysis here

Note, however, that this is dated from 2001, using oggvorbis 1.0beta4 and LAME 3.88. Anyone know of a more up-to-date comparison of high-bitrate encoding?

The completely different architectures of PowerPC vs x86 make it impossible to calculate and compare the emulated speed of an x86 to the native PowerPC speed. 80%*3Ghz != 2.4Ghz PowerPC

Lets travel back, way back. There was the Adlib audio. Then Creative Labs introduced the 8bit, 11Khz Sound Blaster, then the Sound Blaster Pro which added stereo. Then there was the Sounds Blaster 16, Pro Audio Spectrum 16, and the Gravis Ultrasound (GUS) back in 1991.

The GUS was way ahead of the others. It could mix up to 32 channels in hardware. It always played the sound back at 44Khz via interpolation (unless you had too many channels active at once). It had up to 1meg of on board sound memory so it could be totally independent of your CPU. The Demo scene loved it. It had faked 3d sound via QSound..

It never caught on :( Creative's control was too powerful. Even the GUS PnP which was based on the AMD Interwave sound chip failed. Eventually Gravis was bought and the exited the sound business.

Years later Aureal, attempted to bring good audio to the PC and break Creative's control with its Vortex sound card. They ran into money issues. Creative sued them. They won, but the lawsuit drained their money and they went bankrupt. Creative then bought the remains (patents) of the company.

But rumours are nVidia hired many of the out of work engineers, which developed the Sound Storm for the Xbox. Which then nVidia fortunately brought to the nForce. Which unfortunately won't be in future versions because nobody is willing to pay for it. Even if it is "free". Gamers are more interested in a "free" hardware firewall.

Looking back at how Gravis, AMD, Aureal, and others have failed despite having superior products makes me wounder how a company could successfully introduce better audio to gamers. Maybe if it helped you win at FPS games... Seeing nVidia leave the audio market is sad, but I've been sad about this many times before. I'm kind of numb to the pain of seeing a great new technology with high hopes of making things better fail due to lack of interest.

I have a feeling we'll be stuck using Intel's "Azalea" for a long long time. It's certainly not bad, but it has the CPU do the work instead of a coprocessor. What do you expect from Intel when they made a nice new DX9 graphics core, but didn't use hardware T&L? Gotta try to create a market for those faster CPUs somehow... Sure, it can output some Dolby signals if they are precomputed (i.e. DVDs), but it can't encode them if they are dynamic (i.e. games). Unless you have a really powerful CPU. Oh well, at least Intel High Definition Audio as it is officially known now beats AC'97.

Ahem. Set aside 15 minutes and digest this whitepaper. Computer audio is vastly more complicated than simply increasing sample rate and bits per sample, and decreasing SNR and THD. If Aureal had still been around, Creative would have certainly been encouraged to work harder on their drivers and hardware acceleration of this stuff.

http://www.digit-life.com/articles2/amd-hammer-fam ily/

Basically, 2.0 ghz Opteron SPECfp peak 1170
1 ghz Itanium 2 SPECfp peak 1356.

copying 14gb of content over a USB connection should only take a few minutes

USB 2.0 (Hi-Speed) is supposed to be comparable to FireWire 400 at full data transfer speeds, and it only takes minutes to fill up that much content on an iPod through FireWire 400. Actually, USB 2.0 can work at 480 Megabits per second while FireWire 400 works at 400 Megabits per second, although FireWire performs slightly better than USB 2.0 for some reason.

Whatever the reason is that this thing takes hours to complete, it is an awful design flaw at the software level. People aren't going to wait hours for a portable device to get ready. The whole point of a portable device is so you can move around freely, and not be shackled to one location. But this thing keeps you shackled, then lets you go after a few hours, according to how the article described it being used. You could say that would be just the initial download if it was just a music collection. But since it is for a Media Center device that records from television, chances are the entire video collection will be constantly changing, and people will have to download the entire thing all over again. Maybe this has to do with the conversion, but if MS had any sense, they would make the Media Center do the conversion earlier, while it was recording video, rather than during the transfer process.

I would love this product integrated with a head mounted display. Can you imagine looking around in stereo AND your head controls tracking.

Would that not simply be a virtual reality helmet, such as this one? Some other stuff may be found here and here.

Looking around I can't really find any integrated tracker/display headsets, though it is may be because I am not looking in the right places. Expect anywhere from $500 to $3000 and up for these solutions.

This has already been done for the 6800 series!

Wonder if it's been tested on the 6600 yet?

Pipeline mod app

...although even some cameras that are theoretically "USB2.0", like the Nikon D70, still only transfer at USB1.1 speeds (see the very bottom of the page).

And while USB2.0 is theoretically faster than Firewire, it's well known that it is not faster in sustained transfers - I've got a Maxtor Firewire/USB2.0 external hard drive that bears this out; even the manufacturer quotes a higher speed for the firewire interface. (See here for example stats.)

My Clie NX-70v already has a universal remote, and it's a fully functional PalmOS PDA, with other toys like a still/video camera, an MP3 player, a voice recorder, and a full (but tiny) qwerty keyboard.

And according to Amazon, you can get the things for under $120 now. Don't you just love the radical depreciation on a device you paid four times more for barely 18 months ago? *sigh*

But anyway, yeah, the remote. It's fantastic. Going out to eat with a whole bunch of friends, and the restaraunt puts you all in a private room with not one, not two, but three televisions obnoxiously drowning out conversation with sports & news? No problem! The same thing happened to me last week, and the Clie was able to turn off all the televisions right from the table, even when the TVs were 20 or 30 feet away. This only worked, of course, because I happened to have my Clie with me -- but then, it's a PDA, and I almost always have it with me.

The device described in this article, aside from being several times more expensive, is also several times less likely to have general purpose use outside of your living room (unless you're in the habit of going around town turning off televisions, but that isn't a very common hobby). If you're going to spend that much money, why not get a general purpose device?

The Clie I have isn't the only one that has the remote, either. The PEG-T665C also has one, as did the PEG-T415, and it seems like all the models in the NX/NV series (the folding clamshell ones like the NX-70v) have it, too. Even the fanciest of these should be available for a couple hundred bucks cheaper than the Navitus, and all of them are more capable. Shop around! :-)

Is the Geforce 4 MX supported? I know that origionally Carmak wanted to require programable shaders, is that still the case, or did he relent and support the fixed function pipline that the Geforce 4 MX line inherited from the Geforce 2?

Last I heard, id intended on including at least some GF4 MX cards on the minimum requirements list, which would indicate that a GF2 or GF2-Ultra would be even more playable (which is to say not very).

Maybe John Carmack got the GF4 MX working on the NV20 path (Geforce 3) by using its limited pixel shader support. So maybe a GF4 MX is the minimum GPU that uses vertex shaders and makes DOOM 3 "look good."

I couldn't find any good information on NVIDIA's site, but here's one review that compares the GeForce4 MX with the GeForce2 Ti (including a little info on pixel shader support): MSI GF4MX420, GF4MX440 and GF4MX460 Video Cards Review

Quoting from http://www.digit-life.com/articles/ntfs/:

Each file on NTFS has a rather abstract constitution - it has no data, it has streams. One of the streams has the habitual for us sense - file data. But the majority of file attributes are also streams! Thus we have that the base file nature is only the number in MFT and the rest is optional. The given abstraction can be used for the creation of rather convenient things - for example it is possible to "stick" one more stream to a file, having recorded any data in it - for example information about the author and the file content as it was made in Windows 2000 (the most right bookmark in file properties which is accessible from the explorer).

Does this qualify as provision for metadata in the FS?

Blu-Ray burners are available now, and the discs are bigger than HD-DVD's

HD-DVD will not support cartridges, this is mainly because you can make the drives and discs cheaper.

I would imagine that requiring a cart would make laptop drives easier to produce (no tray), although making them optional would make them DAMN complex.

I had a go some time back. The camera rays were pre-calculated. A bounding box was calculated for each triangle in image space. These were ordered by starting row and column. Lists of the possibly visible triangles were maintained as the image space was scanned. Groups of geometry had bounding spheres. I was getting around 15 seconds/frame.

However, there seem to be many open source real-time ray-tracing projects going on:

OpenRT, with it's own FAQ. This project seems to have several games written for it.

Rearview is another game-engine based on ray-tracing

There's also the Avalon Project. There was also an article discussing the use of SSE Instructions. The Source code to a demo is available at RAVI-Demo. Other projects include RealStorm.

It certainly seems to be an active area.

TrustedReviews

Digit-Life

PC Mag

The a tag is your friend...

TrustedReviews

Digit-Life

PC Mag

The a tag is your friend...

The Light Speed Memory Architecture (LMA) that was present in the GeForce3 has been upgraded as well, with it's major advancements in what nVidia calls Quad Cache. Quad Cache includes a Vertex Cache, Primitive Cache, Texture Cache and Pixel Caches. With similar functions as caches on CPU's, these are specific, they store what exactly they say.

Boy this turned into a bit of a tome.

For a switch I went with an 8 port SMC EZSwitch 8508T. I chose it since:
1. It supports jumbo frames. According to my testing it will pass ethernet packets up to 9212 bytes which should correspond to a 9198 byte MTU.
2. It doesn't have a cooling fan. A definate plus since in my experience the little fans in switches such as this can become quite annoying as they age.
3. It comes with rack mount ears.
4. It's affordable. I purchased it from Securemart.com for $139.31 shipped. Ordered it Thursday or Friday, it arrived Monday or Tuesday.

As to NICs, one of my PCs already had an Intel gigabit port on the motherboard. In addition I purchased 4 more Intel Pro 1000/MT Desktop Adapters. Since:
1. They have good driver support on both Linux and Windows.
2. They support jumbo frames. Supposedly up to around 16000 bytes.
3. They're supposed to be pretty fast/efficient. It's kind of dated but you can find a comparison of some 32-bit gigabit NICs here.
4. They'll do 66Mhz if your motherboard supports it and of my systems does.
5. They have DOS NDIS2 drivers so I can use Ghost to make/restore images over the network.

One I purchased through Intel's evaluation program for $35.31 shipped. As I recall it took over a week to show up. The other three I ordered from OnlineMicro for $28 each plus $11.32 shipping. Be sure to change the shipping option from ground to 2 day air if you order more than 1, it's cheaper. They shipped them out the day of my order and they arrived on time.

One of the Intel NICs died about 4 hours after I installed it. I swapped it with another and the replacement has been working fine for a few weeks now. I ran the diagnostics on it and other all but the link test passed. When the OS is booted up the switch shows no link lights but sometimes when the PC is off the link lights do come on. I've also tried it in another PC where it exhibits similar symptoms. I haven't yet contacted Intel about getting it replaced.

I spent a lot of time tweaking various things. Some findings:
1. With default SO_RCVBUF sizes a MTU in the neighborhood of 4000 or so bytes seems to get about the best network/application wide throughput. Specifically the otherwise fast NF7-S system below would lose almost 50% throughput with 9000 byte MTUs with the default SO_RCVBUF size. Linux to Linux lost around 30% as I recall.

In theory you can change the default SO_RCVBUF size on linux by echoing appropriate values to:
/proc/sys/net/ipv4/tcp_rmem
Other than that you appear to have to change this setting in each individual application. One application of note that allows you to easily make this change is samba. See your: /etc/samba/smb.conf

2. If you crank the SO_RCVBUF size up to 200ish k or more then a 9000ish byte MTU can eek out another 5ish percent more bandwidth. Thus for the moment I've decided to just stick with 4076.

3. MTUs that are not of a size of the form 8x+4 cause Linux to behave oddly when it performs path MTU discovery. Namely for jumbo sizes that don't fit that form the discovery decides that the PMTU is 1492. You can read more detail about it in a Usenet post I made here. I still don't have a good picture of what'