Re:How can you compare Afghanistan to WW2?
on
The Drone War
·
· Score: 2
My sequence of events was off, but my point was that it required an extreme effort on behalf of the Allies to defeat Germany, and before the US became involved, the outcome was certainly in doubt. Large numbers of civilians were threatened, the British Home Guard was mobilized & expanded, war production was pushed to its maximum and sacrifices WERE required by the general population to support it. This is quite different from the situation in the US today, where the Afghan campaign is relatively minor.
US civilians were threatened by *al Qaeda terrorists*, yes. But I don't automatically equate al Qaeda terrorists with the Taliban (though the evidence is that the Taliban have supported the al Qaeda operations within Afghanistan). I don't recall any evidence that the WTC and Pentagon attacks were made by Taliban or Afghan miltary forces - they were made by al Qaeda terrorist cells, for the most part operating out of the US. What's more, I doubt that crushing the Taliban's military capability has had any significant effect in the al Qaeda's ability to commit another atrocity in the US or elsewhere.
You seem eager to condemn the lot together, blaming them for actions you have slim or no evidence for. I'd prefer to see the Taliban regime tried in an international UN court before its peers, rather than condoning the vigilante military actions of the US and its supporters. They may well prove to be fully justified - but that should be for the courts to decide. However, that's another issue...
Right, that's the "special" digital link to their own brand of speakers. It'd likely have at least two digital signals, front & rear, in the miniDIN connector. No idea if it follows the S/PDIF standard or not, but it'd require bodging up some kind of adapter at least. It certainly isn't intended to be standard S/PDIF, or they'd have an RCA or BNC connector as well.
The front optical out is standard Toslink S/PDIF, but that's generally only good for short runs. I suppose you could buy a $50 converter to RCA S/PDIF & run a longer cable that way, but that adds a lot to the cost of the unit.
How can you compare Afghanistan to WW2?
on
The Drone War
·
· Score: 3, Insightful
Winston Churchill repeatedly asked his countrymen for brutal sacrifices in World War II. In the new kind of American war, political leaders ask citizens only to keep shopping and traveling.
How are these even remotely comparable? In WW2, before America joined the war, the English were in real danger of losing! Germany was bombing London (remotely, I might add, using the V2 rocket), civilians were dying, and every last bit of effort was required just to hold off the German forces. Churchill was trying to mobilize the entire country in the face of the very real threat of invasion.
In Afghanistan, it couldn't be more different. At no time were US citizens EVER threatened by the Taliban or other Afghan military forces. The overwhelmingly superior US military + allies simply waltzed in and bombed the crap out of them. The cost of the campaign was small change compared to the US GDP. THAT's why no sacrifices were required by US citizens! It had absolutely nothing to do with the technology involved.
All they're doing is quoting the theoretical SNR for a 24 bit device. The question is, what kind of SNR does it REALLY get? I didn't see any figures that looked like real-world results.
SBLive! cards were not exactly known for their clean sound, but as this is outside the noisy box on a separate power supply, it stands a better chance of actually sounding decent.
Sure there's no balanced connectors, but this isn't exactly a professional-level device. But one connector I looked for immediately & failed to see is an RCA S/PDIF out. How am I supposed to run a digital connection to my 5.1 amp downstairs - find a 40 ft optical cable? Stupid to leave off such a cheap & useful connector.
I've always said that our governemt could not get away with, or even propose, the things they do here in any other country. The people wouldn't stand for it.
Sounds a lot like the situation in Australia. Or most Scandinavian countries, AFAIK. Probably quite a few places.
I've heard that e.g. Denmark taxes its citizens up to ~70% (correct me if I'm wrong), but you get boatloads of excellent civil services in return. And monopolies in many areas aren't uncommon - Australia only recently deregulated the auto insurance industry (well, 10-15 years ago anyway).
The number one question (but perhaps not in this forum) on most potential Sledgehammer-owner's minds is, what OS are they going to run on their 64 bit? Apparently not Windows. No announcement has been made by MS or AMD. Yet.
Win64 has been ported to Itanium for some time now. We've already ported our memory-hungry special-FX app to it. But few people outside the server space are going to be interested in getting an Itanium because the performance with legacy IA32 apps is dog-slow. I mean really slow, like P90 speeds. So we don't expect too many sales of that version, just a few for hardcore dedicated seats.
Sledgehammer is really interesting to us. Combine the best available x86-32 performance for running 3DS Max, Lightwave, Photoshop etc etc, along with the serious memory & address space of a 64 bit CPU for our app, not to mention quite a bit more speed when doing 64 bit calculations (pretty common with 64 bit pixels), makes for a powerful and still flexible beast.
But without Windows support, the IA32 performance advantage is largely meaningless. In our market, that relegates Hammer to Linux-64 render farms - which is fine, but it's not where our money is, and it's not where the CPU would shine. You can use Win32 or Linux-32, of course (unlike Itanium), but that's kinda missing the point.
AMD better get MS & Win64 on their side soon, if they want to capture the workstation market. A lot of server apps still require Windows too. The reality of the market is that mainstream OS support is required, or you get niched PDQ.
That's because each thread (under WinNT/2K/XP) reserves 1 MB of address space for its stack, by default. We ran into that one:-) Wondered why we couldn't allocate any more memory, when we were only using half of what was there. Still had physical RAM, but no address space to map it to... 2 or 3 GB isn't enough. We need 64 bit CPUs.
It's possible to reduce the reserved stack, but only for all the threads in a process. We switched to using only a few threads & assigning jobs to them.
nVidia have a popular mid-range line of "professional" 3D chips, the Quadro series (sold by Elsa in its Gloria series). These are basically GeForce chips with a couple of extra features enabled, like hardware anti-aliased lines & two-sided lighting. They're quite a bit more expensive than a consumer GeForce, but a LOT cheaper than most workstation cards.
There's a fewplaces you can look for benchmarks on GeForce, Quadro and mid- to high-end workstation gfx cards. Currently the Wildcat 5110 pretty much rules the roost (at around $3k), with the Quadro2 Pro (under $1k) & FireGL4 (over $1k) competing hotly below that. Lesser cards (FireGL 1 & 2, Quadro, Quadro2 MXR & EX, and the older Oxygen models) can be had for well under $1k. Prices are only from memory, and are probably wildly inaccurate.
Even a standard GF2/GF3 or Radeon does pretty well, impressively so for the price. Rendering quality has been compared (for the GeForce at least), and is roughly equivalent - no major texture or polygon errors, all cards generating the occasional off pixel.
Bottom line: The majority of my customers (2D/3D FX) are switching to GeForce or sometimes Quadro cards - sought-after features include decent (not necessarily superlative) 3D app performance, dual monitor support (WITH hardware accel on both monitors!), and bang for the buck. Good DirectX support doesn't hurt either (very few cards from 3Dlabs support DirectDraw well, and some serious apps do need this).
Any standalone soundcard based on the nForce APU would require some large & reasonably fast local memory. The bandwidth required by the APU for sound rendering can exceed 500 MB/s, according to nVidia - one reason for the 800 MB/s Hypertransport link between the north & southbridge chips, and a significant user of the chipset's "spare" 2.1 GB/s of main memory bandwidth, even when an external gfx card is used.
I've heard different figures for the latency introduced by realtime DD encoding - between 10ms and 70ms. 10ms wouldn't be perceivable in the context of a game, and even 70ms isn't much - a lot for a musical performance, but still difficult to perceive - especially when the frames themselves will also be delayed by up to 33-50ms (when double- or triple-buffering).
I've spent many hours playing games with DD-encoded sound on my Xbox, and I've tried listening specifically for delayed sound, but I haven't noticed any examples yet. The sound, BTW is superb, and is one of the main reasons I bought the Xbox.
As for the SB Live! & Audigy products, how does AC-3 passthrough (for DVD-playback, presumably) help in any way with games? If you're willing to run four separate wires to your amp, you hardly even need an AC-3 S/PDIF connection - software decoding of AC-3 to the soundcard's 4-channel output would probably be sufficient.
Even Zmodem used a 32 bit CRC. Hell, why not just zip the thing?
Only if your Palm is sitting in its cradle
on
Review: SliMP3
·
· Score: 2
That requires the Palm to be in its cradle. You'd have to run a long serial/USB extension cable to put your cradle in your lounge room, as well as the long audio cables.
Not a bad idea for a two-way remote though, even if it isn't wireless. I have a $25 RF remote to control Winamp on my PC, and a 10m S/PDIF cable returning for the sound, but I don't get any other info, just the music.
It's spelled "fluorescent", it will hold "20 - 100 Gigabytes of pre-recorded data" and you can't put one in your DVD player.
Still, 100 GB per disc ain't bad at all. They do have a WORM version, no rewritable yet it seems. However, I'm still waiting (and waiting) for them to produce a real product...
... which will not only record DV (via 1394) and analog video (via S-Video or composite), but also includes a TV tuner, EPG, PVR functions, S-video/composite out, RF remote control, Ulead Video Studio (not too bad) and a pretty darn good 3D card too:-) Price: $399, available RSN.
Or you can just go for a $50-100 analog capture card (with no DV support).
It's not so bad. Since DVD+RW drives can read DVD-R and DVD-RW discs (as well as vice versa), it doesn't matter much which you go for. They can't write to each other's media, but reading is fine.
The main things to consider are the drive itself (speed etc), price (drive & media), and perhaps availability. DVD+RW is supposedly a little more compatible that DVD-RW, but the difference isn't apparently that large in users' experiences.
That said, I'm personally holding off until DVD+RW drives can also write to DVD+R media. That'll be cheaper than rewritable media, and more compatible (rewritable discs - of either standard - have different reflective properties, which confuses some older players into thinking it's a double-layer disc).
HP don't have a clear position on whether current dvd100i drives will be firmware-upgradeable to support DVD+R (Ricoh make the drive unit itself for HP, and they won't say either). When I know that the drive I buy will do this too, I'll be first in line:-)
That's not so easy. The HD is locked by a password, as provided for by the ATA spec. It's not the filesystem that's the difficulty.
What you can do is power the HD with a cable from a PC, turn on the PC then the Xbox, wait till the Xbox firmware unlocks the drive, does what it has to do, then goes idle.
Then you turn off the Xbox (drive stays powered), carefully unplug the IDE cable & attach it to your PC's IDE controller instead. Rescan the drives & you should see it AND be able to read from it. Some info here, and elsewhere on those forums.
Reportedly, the filesystem is a variant on the good old FAT, and not even encrypted. There are some differences, but apparently nothing too challenging (but it won't mount as an ordinary FAT partition, of course).
Executables are signed, and modifying the exe invalidates the signature, so that won't be easy to get around. Perhaps replacing the BIOS would help, but you'd want to find out the HD password first...
And yes, the MAME port was done on an SDK kit, which is much more open of course).
No, "there are in fact four identical 256k copies of the ROM image in the 1 MB flash ROM." - not multiple different versions of the BIOS.
It's entirely likely MS may revise the BIOS at a later stage, for fixing bugs - if it turns out to be necessary. All console makers do. So long as you don't change the external functionality at all, that's fine.
But what the article was saying is that this is a reasonable move by MS, to increase robustness in the possible case of field flash rom upgrades, or even if multiple versions of the BIOS might be required. I'm sure MS are very aware that the latter case is not a desirable one.
Come to that, I doubt that "flash ugrades in the field" are even possible, let alone planned, or the author probably wouldn't have had to replace the ROM at all. It'd be a major security hole, if you could do that!:-)
The article had some great coloured-mipmap shots of the two cards. The GeForce shots showed lovely trilinear filtering of the mipmaps, true per-pixel range-based transitions with nice soft blending. The older Radeon drivers did pseudo-approximate-range-based transitions with soft (but not as nice) blending between the mipmap levels.
But the new Radeon drivers don't bother with soft trilinear blending at all. There is only one 50% blend level between mipmap levels, when trilinear is turned on. That's not trilinear - that's a "dual-bilinear" hack of some kind. And it's still not properly range-based.
Worse, when anisotropic filtering is enabled, you don't get trilinear at all. The mipmap level transitions are bilinear, hard edged. Looks awful. And they're still not properly range-based. THIS is the reason anisotropic filtering doesn't cause the same performance hit on an 8500.
I don't understand why people keep insisting that ATI cards have superior quality images. Certainly not in 3D - they're taking all kinds of quality-reduction shortcuts to try & boost their benchmarks. Fine so long as it's optional, but as before, it isn't.
Their 2D output is fine, better than some brands of nVidia chip-based cards, but you can certainly find other GeForce-based brands which look great in 2D. My reference QuadroDCC looks superb, better than my Matrox G400.
I really wish ATI would stop forcing these compromises on us just to squeeze a few more fps from Q3A. If I want faster performance on a game, I'll lower the resolution, or turn down the texture size or something - in the game, or the driver. If I ask the card for max quality, trilinear & all the nice stuff, I want to get max quality, not some half-assed performance hack.
That said, I'm keen to see what Smoothvision looks like these days. Sounds nice:-)
First time I've heard "immature drivers" cited as a reason to get a product.
You're being inconsistent. First you say you don't care about an extra 10 fps in 3DMark2k1 (though it's actually the 8500 that gets the higher 2k1 score), and you like the cheaper price of the 8500. Then you say you'd rather spend the extra money on an 8500 instead of a Ti200 because of a possible future 10 fps...
It's Saturday, and I'm bored, so I'll reply... Incidentally, I work in the special effects software field for a living, so I'll claim some incidental knowledge of the subject;-)
The solution was obviously trilinear filtering. It does indeed remove moire, but at the cost of bluring the texture beyond recognition in some implementations.
Over-filtering causes blur. Correct filtering reduces aliasing (and therefore moire). This doesn't make up for lack of detail, of course - only more samples can give that - but for a given number of samples, good filtering preserves as much detail as possible while removing the high-frequency artifacting inherently caused by representing a continuous analog signal or picture with a limited number of point samples.
The obvious correct way would be some form of tricubic filtering, as would the obvious way to fix two dimensional scaling would be bicubic filtering.
Trilinear filtering is not the "obvious correct" solution - it's a logical extension to bilinear and it's easy to do, but generally Gaussian filtering is considered to do the best job (but it's quite hard to do fast). It does depend on the case, though.
The banding that is present in 16-bit of depth is not precievable by the faulty human eye.
Incorrect. The sensitivity of the eye to gradations of colour or luminance depends on the range - slight variations in green or light grey are much more easily detectable than slight variations in blue or near-black - but 16 bits total is nowhere near enough to represent continuous tones in almost any range. Many people can easily distinguish differences in certain colour & luminance ranges even when using 24 bits. Your example is easy to spot - have you tried it?
The problem with banding that you see in most games of 16-bit color is caused when the game requests transparencies greater than 1 bit.
Huh? Perhaps you're trying to say, the problem with banding in 16 bit colour is *exacerbated* by repeated overlaying of semi-transparent images, which is true, especially if it's done badly (like early 3dfx hardware tended to).
The use of 32 bits I think is a bit wasteful at this point, but in the future, we would all like to see floating point color implemented in hardware as well(already in place in OpenGL).
32 bits is an absolute minimum for credible graphics work. Film effects typically use 64 bits (48 bit colour, 16 bit tranparency) to avoid banding, or at the very least, a logarithmically-encoded 32 bit scheme.
Floating point image processing is sometimes required, more for representing out-of-range colours than for the extra precision, but is always done in software. The OpenGL API provides for the use of floating point colours, but I know of no OpenGL hardware, consumer or professional, that uses floating point colours in the hardware.
True, of course. But as that's still quite a ways off, both in terms of building such a display and dealing with the sheer amount of data required to represent such an image, we must fall back on techniques to reduce aliasing instead. 2 years is hopelessly optimistic - I would say quite a bit more than 20 years. Thus AA & filtering will be required for some time to come.
The nVidia demo was not realtime nor at a resolution that many gamers would accept.
Did you actually see it, live? I did. I don't know what you define realtime as, but I define it to be "a pace that gives the illusion of motion". Most people accept this to be a few frames per second, or more - and it was definitely that (I judged it to average around 10 fps). Any interactive change, such as a camera move, gave feedback within around a tenth of a second, which is more than enough to work with. Not fast enough for a twitch game like Quake, but good enough for a cutscene, and excellent for a 3D artist to check their work with.
As for the resolution, 1920x1080 is considerably more than 1024x768, which is what most gamers would accept!
I can't help but to see where ray-tracing would make things at least appear more organic.[...] These require polygons(or more CSG for raytracers).
Raytracing typically makes things look sharp & shiny, not organic. While raytracing is excellent for certain effects, generally you need a more advanced lighting solution (such as radiosity or global illumination) to get the more realistic look provided by soft lighting & shadows.
More polygons do help in defining more detailed or smoother organic shapes, but this has nothing to do with raytracing, as such. Incidentally, very few ray tracers use CSG shapes these days - only Real4D comes to mind. Polygons are far easier to use for representing arbitrary objects than a collection of geometrical shapes.
I'm not quite so optimistic about raytracing on a chip.
Actually, people have been doing realtime raytracing in software for years, even on a 486! Admittedly the resolution was low & the scene was simple, but when you think of the sheer floating point grunt of modern graphics hardware (nVidia claim their GeForce3 is capable of 76 gigaflops - a maxed-out, 256-CPU Cray T3D could do 50 gigaflops) and the ever-increasing parallelism being added to these chips as well as the growing clockspeed, I think something will be put together a lot sooner than you think...
I agree that better textures and higher resolutions are not a substitute for more detailed scenes & better physics, but fortunately one does not exclude the other. Two years ago the focus shifted to more polygons, and now game detail is soaring by an order of magnitude.
This year we added programmable vertex & pixel pipelines, and already we're seeing the results (Xbox games feature better bumpmapping, more natural surface lighting & realistically distorted reflections & refraction, in addition to smooth characters & increased detail). OpenGL is being redesigned from scratch to encompass the new paradigm. What will next year bring?
No PC sound card supports Dolby Digital 5.1 (or DTS) for anything except when passing through a pre-encoded signal, e.g. from a DVD. Any sound generated by a PC game (e.g. via DirectSound3D) has to be sent out multiple analog outputs, or is limited to stereo only (though potentially the CPU could encode a matrixed ProLogic signal into this).
Gamecube does not support Dolby 5.1 sound at all (no digital out). Some Cube games do support Dolby Prologic II, which offers pseudo-5.1 sound, matrixed into a stereo signal (and yes, there's still a latency).
PS2 supports only pre-encoded DD5.1 sound (for cutscenes & background music), though I have heard that a new game or two manage realtime DTS 5.1 encoding, by dedicating half the graphics engine to the task.
Xbox has the only Dolby 5.1 realtime encoder, done by a dedicated DSP in its MCPX chip (nVidia's nForce PC chipset does too). This adds only ~10ms latency, less than one frame's worth. There's no special effort required by the developer to use it (a single line of code instructs the sound engine to enable the encoding, and any DirectSound3D source is automatically positioned in 3D by the realtime Dolby encoding and Sensaura HRTFs).
Using the Dolby encoder on the Xbox does not take resources from the game, whereas any realtime multichannel encoding must be done by the CPU (on the Cube), or possibly one of the vector processors (on the PS2).
And yes, from personal experience of Halo on my 5.1 system, the sound is stunning - the best feature of the Xbox, IMHO.
I've heard people say the Xbox loading times are enormously long - I'd have to disagree. I own an Xbox, and I've also played on a Cube, and to me they seem pretty similar.
I did notice that some Xbox games (e.g. DOA3, Project Gotham) have a really long intro before the menu pops up. Perhaps that made people think they had to sit through all of it? I personally hit the Start button imnmediately, and I get the game menu straight away, within seconds of putting in the game.
With the Cube (looking at Wave Race and Rogue Leader), the intro was certainly shorter, but not interruptible. Total time taken before you could get to the menu was more or less equivalent.
I doubt the mini-DVDs will seek any faster than a full-size DVD. Perhaps average seek time would be less for a full 1.3 GB mini-DVD than for a full 9.4 GB disk, but if you stored 1 GB of data on each type of disc, it'd actually seek quicker on a full-size disc (since on the Xbox all data is stored at the outside rim, where it spins faster and you can fit more data per track, thus resulting in higher transfer rates, fewer seeks required, and a smaller average seek distance).
Are the Cube's mini-DVDs dual-layer, like the Xbox? A dual-layer disc would have twice as much data in the same area. Though I don't know what the refocussing delay is to switch layers, or whether an Xbox game always uses both layers or only when one layer is full.
Slashdot Mirror
US civilians were threatened by *al Qaeda terrorists*, yes. But I don't automatically equate al Qaeda terrorists with the Taliban (though the evidence is that the Taliban have supported the al Qaeda operations within Afghanistan). I don't recall any evidence that the WTC and Pentagon attacks were made by Taliban or Afghan miltary forces - they were made by al Qaeda terrorist cells, for the most part operating out of the US. What's more, I doubt that crushing the Taliban's military capability has had any significant effect in the al Qaeda's ability to commit another atrocity in the US or elsewhere.
You seem eager to condemn the lot together, blaming them for actions you have slim or no evidence for. I'd prefer to see the Taliban regime tried in an international UN court before its peers, rather than condoning the vigilante military actions of the US and its supporters. They may well prove to be fully justified - but that should be for the courts to decide. However, that's another issue...
The front optical out is standard Toslink S/PDIF, but that's generally only good for short runs. I suppose you could buy a $50 converter to RCA S/PDIF & run a longer cable that way, but that adds a lot to the cost of the unit.
How are these even remotely comparable? In WW2, before America joined the war, the English were in real danger of losing! Germany was bombing London (remotely, I might add, using the V2 rocket), civilians were dying, and every last bit of effort was required just to hold off the German forces. Churchill was trying to mobilize the entire country in the face of the very real threat of invasion.
In Afghanistan, it couldn't be more different. At no time were US citizens EVER threatened by the Taliban or other Afghan military forces. The overwhelmingly superior US military + allies simply waltzed in and bombed the crap out of them. The cost of the campaign was small change compared to the US GDP. THAT's why no sacrifices were required by US citizens! It had absolutely nothing to do with the technology involved.
SBLive! cards were not exactly known for their clean sound, but as this is outside the noisy box on a separate power supply, it stands a better chance of actually sounding decent.
Sure there's no balanced connectors, but this isn't exactly a professional-level device. But one connector I looked for immediately & failed to see is an RCA S/PDIF out. How am I supposed to run a digital connection to my 5.1 amp downstairs - find a 40 ft optical cable? Stupid to leave off such a cheap & useful connector.
And no, I don't want to upgrade it later, I want it now!
Sounds a lot like the situation in Australia. Or most Scandinavian countries, AFAIK. Probably quite a few places.
I've heard that e.g. Denmark taxes its citizens up to ~70% (correct me if I'm wrong), but you get boatloads of excellent civil services in return. And monopolies in many areas aren't uncommon - Australia only recently deregulated the auto insurance industry (well, 10-15 years ago anyway).
Win64 has been ported to Itanium for some time now. We've already ported our memory-hungry special-FX app to it. But few people outside the server space are going to be interested in getting an Itanium because the performance with legacy IA32 apps is dog-slow. I mean really slow, like P90 speeds. So we don't expect too many sales of that version, just a few for hardcore dedicated seats.
Sledgehammer is really interesting to us. Combine the best available x86-32 performance for running 3DS Max, Lightwave, Photoshop etc etc, along with the serious memory & address space of a 64 bit CPU for our app, not to mention quite a bit more speed when doing 64 bit calculations (pretty common with 64 bit pixels), makes for a powerful and still flexible beast.
But without Windows support, the IA32 performance advantage is largely meaningless. In our market, that relegates Hammer to Linux-64 render farms - which is fine, but it's not where our money is, and it's not where the CPU would shine. You can use Win32 or Linux-32, of course (unlike Itanium), but that's kinda missing the point.
AMD better get MS & Win64 on their side soon, if they want to capture the workstation market. A lot of server apps still require Windows too. The reality of the market is that mainstream OS support is required, or you get niched PDQ.
It's possible to reduce the reserved stack, but only for all the threads in a process. We switched to using only a few threads & assigning jobs to them.
Damn. And I was so looking forward to bringing up my own kids...
There's a few places you can look for benchmarks on GeForce, Quadro and mid- to high-end workstation gfx cards. Currently the Wildcat 5110 pretty much rules the roost (at around $3k), with the Quadro2 Pro (under $1k) & FireGL4 (over $1k) competing hotly below that. Lesser cards (FireGL 1 & 2, Quadro, Quadro2 MXR & EX, and the older Oxygen models) can be had for well under $1k. Prices are only from memory, and are probably wildly inaccurate.
Even a standard GF2/GF3 or Radeon does pretty well, impressively so for the price. Rendering quality has been compared (for the GeForce at least), and is roughly equivalent - no major texture or polygon errors, all cards generating the occasional off pixel.
Bottom line: The majority of my customers (2D/3D FX) are switching to GeForce or sometimes Quadro cards - sought-after features include decent (not necessarily superlative) 3D app performance, dual monitor support (WITH hardware accel on both monitors!), and bang for the buck. Good DirectX support doesn't hurt either (very few cards from 3Dlabs support DirectDraw well, and some serious apps do need this).
I've heard different figures for the latency introduced by realtime DD encoding - between 10ms and 70ms. 10ms wouldn't be perceivable in the context of a game, and even 70ms isn't much - a lot for a musical performance, but still difficult to perceive - especially when the frames themselves will also be delayed by up to 33-50ms (when double- or triple-buffering).
I've spent many hours playing games with DD-encoded sound on my Xbox, and I've tried listening specifically for delayed sound, but I haven't noticed any examples yet. The sound, BTW is superb, and is one of the main reasons I bought the Xbox.
As for the SB Live! & Audigy products, how does AC-3 passthrough (for DVD-playback, presumably) help in any way with games? If you're willing to run four separate wires to your amp, you hardly even need an AC-3 S/PDIF connection - software decoding of AC-3 to the soundcard's 4-channel output would probably be sufficient.
Even Zmodem used a 32 bit CRC. Hell, why not just zip the thing?
Not a bad idea for a two-way remote though, even if it isn't wireless. I have a $25 RF remote to control Winamp on my PC, and a 10m S/PDIF cable returning for the sound, but I don't get any other info, just the music.
Still, 100 GB per disc ain't bad at all. They do have a WORM version, no rewritable yet it seems. However, I'm still waiting (and waiting) for them to produce a real product...
Or you can just go for a $50-100 analog capture card (with no DV support).
The main things to consider are the drive itself (speed etc), price (drive & media), and perhaps availability. DVD+RW is supposedly a little more compatible that DVD-RW, but the difference isn't apparently that large in users' experiences.
That said, I'm personally holding off until DVD+RW drives can also write to DVD+R media. That'll be cheaper than rewritable media, and more compatible (rewritable discs - of either standard - have different reflective properties, which confuses some older players into thinking it's a double-layer disc).
HP don't have a clear position on whether current dvd100i drives will be firmware-upgradeable to support DVD+R (Ricoh make the drive unit itself for HP, and they won't say either). When I know that the drive I buy will do this too, I'll be first in line :-)
...here.
What you can do is power the HD with a cable from a PC, turn on the PC then the Xbox, wait till the Xbox firmware unlocks the drive, does what it has to do, then goes idle.
Then you turn off the Xbox (drive stays powered), carefully unplug the IDE cable & attach it to your PC's IDE controller instead. Rescan the drives & you should see it AND be able to read from it. Some info here, and elsewhere on those forums.
Reportedly, the filesystem is a variant on the good old FAT, and not even encrypted. There are some differences, but apparently nothing too challenging (but it won't mount as an ordinary FAT partition, of course).
Executables are signed, and modifying the exe invalidates the signature, so that won't be easy to get around. Perhaps replacing the BIOS would help, but you'd want to find out the HD password first...
And yes, the MAME port was done on an SDK kit, which is much more open of course).
It's entirely likely MS may revise the BIOS at a later stage, for fixing bugs - if it turns out to be necessary. All console makers do. So long as you don't change the external functionality at all, that's fine.
But what the article was saying is that this is a reasonable move by MS, to increase robustness in the possible case of field flash rom upgrades, or even if multiple versions of the BIOS might be required. I'm sure MS are very aware that the latter case is not a desirable one.
Come to that, I doubt that "flash ugrades in the field" are even possible, let alone planned, or the author probably wouldn't have had to replace the ROM at all. It'd be a major security hole, if you could do that! :-)
The article had some great coloured-mipmap shots of the two cards. The GeForce shots showed lovely trilinear filtering of the mipmaps, true per-pixel range-based transitions with nice soft blending. The older Radeon drivers did pseudo-approximate-range-based transitions with soft (but not as nice) blending between the mipmap levels.
But the new Radeon drivers don't bother with soft trilinear blending at all. There is only one 50% blend level between mipmap levels, when trilinear is turned on. That's not trilinear - that's a "dual-bilinear" hack of some kind. And it's still not properly range-based.
Worse, when anisotropic filtering is enabled, you don't get trilinear at all. The mipmap level transitions are bilinear, hard edged. Looks awful. And they're still not properly range-based. THIS is the reason anisotropic filtering doesn't cause the same performance hit on an 8500.
I don't understand why people keep insisting that ATI cards have superior quality images. Certainly not in 3D - they're taking all kinds of quality-reduction shortcuts to try & boost their benchmarks. Fine so long as it's optional, but as before, it isn't.
Their 2D output is fine, better than some brands of nVidia chip-based cards, but you can certainly find other GeForce-based brands which look great in 2D. My reference QuadroDCC looks superb, better than my Matrox G400.
I really wish ATI would stop forcing these compromises on us just to squeeze a few more fps from Q3A. If I want faster performance on a game, I'll lower the resolution, or turn down the texture size or something - in the game, or the driver. If I ask the card for max quality, trilinear & all the nice stuff, I want to get max quality, not some half-assed performance hack.
That said, I'm keen to see what Smoothvision looks like these days. Sounds nice :-)
You're being inconsistent. First you say you don't care about an extra 10 fps in 3DMark2k1 (though it's actually the 8500 that gets the higher 2k1 score), and you like the cheaper price of the 8500. Then you say you'd rather spend the extra money on an 8500 instead of a Ti200 because of a possible future 10 fps...
The solution was obviously trilinear filtering. It does indeed remove moire, but at the cost of bluring the texture beyond recognition in some implementations.
Over-filtering causes blur. Correct filtering reduces aliasing (and therefore moire). This doesn't make up for lack of detail, of course - only more samples can give that - but for a given number of samples, good filtering preserves as much detail as possible while removing the high-frequency artifacting inherently caused by representing a continuous analog signal or picture with a limited number of point samples.
The obvious correct way would be some form of tricubic filtering, as would the obvious way to fix two dimensional scaling would be bicubic filtering.
Trilinear filtering is not the "obvious correct" solution - it's a logical extension to bilinear and it's easy to do, but generally Gaussian filtering is considered to do the best job (but it's quite hard to do fast). It does depend on the case, though.
The banding that is present in 16-bit of depth is not precievable by the faulty human eye.
Incorrect. The sensitivity of the eye to gradations of colour or luminance depends on the range - slight variations in green or light grey are much more easily detectable than slight variations in blue or near-black - but 16 bits total is nowhere near enough to represent continuous tones in almost any range. Many people can easily distinguish differences in certain colour & luminance ranges even when using 24 bits. Your example is easy to spot - have you tried it?
The problem with banding that you see in most games of 16-bit color is caused when the game requests transparencies greater than 1 bit.
Huh? Perhaps you're trying to say, the problem with banding in 16 bit colour is *exacerbated* by repeated overlaying of semi-transparent images, which is true, especially if it's done badly (like early 3dfx hardware tended to).
The use of 32 bits I think is a bit wasteful at this point, but in the future, we would all like to see floating point color implemented in hardware as well(already in place in OpenGL).
32 bits is an absolute minimum for credible graphics work. Film effects typically use 64 bits (48 bit colour, 16 bit tranparency) to avoid banding, or at the very least, a logarithmically-encoded 32 bit scheme.
Floating point image processing is sometimes required, more for representing out-of-range colours than for the extra precision, but is always done in software. The OpenGL API provides for the use of floating point colours, but I know of no OpenGL hardware, consumer or professional, that uses floating point colours in the hardware.
True, of course. But as that's still quite a ways off, both in terms of building such a display and dealing with the sheer amount of data required to represent such an image, we must fall back on techniques to reduce aliasing instead. 2 years is hopelessly optimistic - I would say quite a bit more than 20 years. Thus AA & filtering will be required for some time to come.
The nVidia demo was not realtime nor at a resolution that many gamers would accept.
Did you actually see it, live? I did. I don't know what you define realtime as, but I define it to be "a pace that gives the illusion of motion". Most people accept this to be a few frames per second, or more - and it was definitely that (I judged it to average around 10 fps). Any interactive change, such as a camera move, gave feedback within around a tenth of a second, which is more than enough to work with. Not fast enough for a twitch game like Quake, but good enough for a cutscene, and excellent for a 3D artist to check their work with.
As for the resolution, 1920x1080 is considerably more than 1024x768, which is what most gamers would accept!
I can't help but to see where ray-tracing would make things at least appear more organic.[...] These require polygons(or more CSG for raytracers).
Raytracing typically makes things look sharp & shiny, not organic. While raytracing is excellent for certain effects, generally you need a more advanced lighting solution (such as radiosity or global illumination) to get the more realistic look provided by soft lighting & shadows.
More polygons do help in defining more detailed or smoother organic shapes, but this has nothing to do with raytracing, as such. Incidentally, very few ray tracers use CSG shapes these days - only Real4D comes to mind. Polygons are far easier to use for representing arbitrary objects than a collection of geometrical shapes.
I'm not quite so optimistic about raytracing on a chip.
Actually, people have been doing realtime raytracing in software for years, even on a 486! Admittedly the resolution was low & the scene was simple, but when you think of the sheer floating point grunt of modern graphics hardware (nVidia claim their GeForce3 is capable of 76 gigaflops - a maxed-out, 256-CPU Cray T3D could do 50 gigaflops) and the ever-increasing parallelism being added to these chips as well as the growing clockspeed, I think something will be put together a lot sooner than you think...
I agree that better textures and higher resolutions are not a substitute for more detailed scenes & better physics, but fortunately one does not exclude the other. Two years ago the focus shifted to more polygons, and now game detail is soaring by an order of magnitude.
This year we added programmable vertex & pixel pipelines, and already we're seeing the results (Xbox games feature better bumpmapping, more natural surface lighting & realistically distorted reflections & refraction, in addition to smooth characters & increased detail). OpenGL is being redesigned from scratch to encompass the new paradigm. What will next year bring?
Gamecube does not support Dolby 5.1 sound at all (no digital out). Some Cube games do support Dolby Prologic II, which offers pseudo-5.1 sound, matrixed into a stereo signal (and yes, there's still a latency).
PS2 supports only pre-encoded DD5.1 sound (for cutscenes & background music), though I have heard that a new game or two manage realtime DTS 5.1 encoding, by dedicating half the graphics engine to the task.
Xbox has the only Dolby 5.1 realtime encoder, done by a dedicated DSP in its MCPX chip (nVidia's nForce PC chipset does too). This adds only ~10ms latency, less than one frame's worth. There's no special effort required by the developer to use it (a single line of code instructs the sound engine to enable the encoding, and any DirectSound3D source is automatically positioned in 3D by the realtime Dolby encoding and Sensaura HRTFs).
Using the Dolby encoder on the Xbox does not take resources from the game, whereas any realtime multichannel encoding must be done by the CPU (on the Cube), or possibly one of the vector processors (on the PS2).
And yes, from personal experience of Halo on my 5.1 system, the sound is stunning - the best feature of the Xbox, IMHO.
Why is it that, everytime something new is announced, someone always complains that this thing is no good because they don't have what they want yet?
I did notice that some Xbox games (e.g. DOA3, Project Gotham) have a really long intro before the menu pops up. Perhaps that made people think they had to sit through all of it? I personally hit the Start button imnmediately, and I get the game menu straight away, within seconds of putting in the game.
With the Cube (looking at Wave Race and Rogue Leader), the intro was certainly shorter, but not interruptible. Total time taken before you could get to the menu was more or less equivalent.
I doubt the mini-DVDs will seek any faster than a full-size DVD. Perhaps average seek time would be less for a full 1.3 GB mini-DVD than for a full 9.4 GB disk, but if you stored 1 GB of data on each type of disc, it'd actually seek quicker on a full-size disc (since on the Xbox all data is stored at the outside rim, where it spins faster and you can fit more data per track, thus resulting in higher transfer rates, fewer seeks required, and a smaller average seek distance).
Are the Cube's mini-DVDs dual-layer, like the Xbox? A dual-layer disc would have twice as much data in the same area. Though I don't know what the refocussing delay is to switch layers, or whether an Xbox game always uses both layers or only when one layer is full.