Stock options have value and should be expensed somehow, but to "only" expense the gains when the employee exercises and benefits leads to all sorts of counter-productive results.
As CEO, I work hard to increase share price to benefit the shareholders. I somehow achieve my goal, then all my employees (including me) exercise/sell to reap the benefits. Suddenly my earnings take a huge hit. Boom, my stock price crashes. Sure, I could call it a one-time charge, but option exercise/sell is basically out of my control and could happen every quarter. In the end, shareholders and financial analysts would have to ignore this aspect of my earnings, which brings us back to the situation we have today.
The speed of the high-level language is getting less and less important. Java3D is an API that is implmented as native code (JNI?) on the target system. The application can construct the data using Java prior to runtime, but display lists and scene graphs take the high-level language out of the critical path.
It's true that SciVis, CFD and the like aren't going to be happy running in Java, but I believe you can do a good flight sim or game using Java3D.
(Disclaimer: I haven't used Java3D... OpenGL and Performer are/were my preference, but Java3D was designed with the help of the SGI Performer team)
I didn't watch any of the episodes on Fox because I hadn't even discovered Buffy yet. 6 seasons of DVD Buffy later (waiting for the 7th), we buy Firefly's DVD set and were really impressed.
Firefly, the series, has the usual Joss Whedan good characters interacting in interesting ways. Apparently they didn't show the pilot first on Fox and I can't imagine how that could have worked. The 2 hour pilot (also called Serenity) is a must see and its CGI special effects are surprisingly good.
My experience with console development systems is that you have almost no rights to transfer/sell. Everything must be handled through Sony/Microsoft/Nintendo. I predict that, if they want to, they can confiscate all of his development systems. Maybe the rules are different in Japan, but I doubt it.
When I saw this on slashdot, I thought that I had been doing SETI@Home for much longer than that, but apparently I registered May 16th, 1999, early in the UTC. Their news release puts the anniversay as May 17th.
Racksaver was selling dual-machine 1U racks for several years and I owned a few of them. Think deep, not tall. Racksaver seems to have renamed itself Verari and only has dual-Opteron in a 1U now. Most dense configs seem to be blade-based these days. Verari advertises 132 processors in a single rack, but I suspect they are not king in this area.
If Google is innovating in this area, it could either be on price or in density.
The biggest problem right now is that there are new graphics cards coming out every 6 months with architecture changes every 18-24 months. Games have been behind the development curve for a while. Finally the two big game engines come out with new versions, trying to aim at what should be reasonable at the time.
Unfortunately, the rest of the PC hardware has turned into complete commodity and its unclear whether its worth spending another $500 on the rest of the computer to hold the FX6800 when it comes out (things are relatively quiet in the land of CPU and memory, where spending 3x the money may get you a 20% increase).
If you're a 3D software developer trying to pick which features to use to get decent market penetration (yet still take advantage of the new programmability), you're pretty-well hosed right now with the various flavors of pixel/vertex shader instructions and program lengths available on the various cards.
ATI 9600, NV FX5600 - these are the cards/capabilities I would depend on to be widespread in the installed base by Xmas 2005.
All DirecTiVo (including HD), record the digital bit stream directly from the satellite onto the hard drive, so there is no degradation at all. The HD-TiVo added OTA (over-the-air) tuners for the local digital TV broadcasts and those bits are also sent directly to the hard drive. 19Mbits/sec is the maximum HD rate for OTA, while satellite/cable encodings of HD tend to be 13Mbits/sec or less.
In reading the initial comments about the HD-TiVo, there is one complaint that could be a problem for those who are currently using a DirecTV HD receiver like the DTC-100 and a non-HD DirecTiVo.
Apparently, when the HD-TiVo gets a non-HD signal, it doesn't automatically switch its output to 480i/480p. It also doesn't stretch/zoom the image to fill a 16x9 screen. This means you need to manually switch the output if you want your TV's de-interlacer/scaler to adjust the image. Depending on who you ask, this is a no-op, annoying, or fatal. (I'm probably in the annoying camp)
At least according to the reports I could find, Alias and Autodesk are not members/founders of 3DIF. I would be very surprised if a "universal" 3D format is created without their help.
The problem is that there are many diverse needs/users for 3D data and the data is so large and/or hard-to-render that simplifying it to a single format that meets the needs of CAD, film animation, sci-vis, game animation, 3D web content, and GIS is not reasonable.
On the other hand, VRML may have been too early, but a poly+texture+simplebehaviours format that was well supported by all applications would be a good thing to (re)create.
Clipper was the first RISC CPU. It replaced Intergraph's N32032 based-workstations with a 33Mhz, 5 MIPS (what part of that is "RISC"??) processor. As compared to the 32032, though, it rocked.
Compilers were tricky, Green Hills was the only choice.
E911 lets the service provider know where you are by your tower connection (and perhaps by other tower signal strengths). It's expensive to keep track of, but is required by law. They don't have enough financial interest in returning that location information to you yet.
AGPS (assisted GPS) also depends on the service provider to calculate your location, but it actually uses real GPS satellite signals to do it. The signal strengths are uploaded to a server which does the heavy-lifting of figuring out the location. Again, this depends on the SP servers and they aren't going to be terribly interested in returning lots of location points back to you.
Real GPS does the CPU work on the device. There are rumored to be a few phones with this capability, but I use a Garmin Gecko 301. $228, records 10K points, downloads via serial port (yuck), and burns a set of alkaline batteries in 7 hours (or less). IMHO, it'll be a while before location information is as easy to get to as the cheapest GPS. Battery power is just too valuable.
I predict Xbox2 _definitely_ won't be compatible with Xbox1 based on Jen-Hsun's comment.
Microsoft buys all of its technology for game consoles nearly off-the-shelf. The Xbox1 graphics was a slightly modified (and slightly early) GF3 and the CPU was an OTS Intel P3 733. This allowed them to have vendors bidding against each other and get a very good price (AMD and Gigapixel got screwed last time).
Now that they have chosen ATI and IBM for graphics and CPU, respectively, any technology used in the prior generation is not only hard to emulate but runs into IP land mines.
Sony likes backwards compatibility, owns all of its graphics technology, and works hard to built a cheap version of last-gen into next-gen. The interesting question will be whether PS3 plays PS1 games:-).
Nintendo just doesn't care. They face a similar problem to Microsoft, but have better contracts with their suppliers _and_, in this case, GameCube2 seems to have the same corporate suppliers as Gamecube did (ATI and IBM). Even so, I would guess that the blow off compatibility.
You missed a good point in the original question. Even if you have tons of RAM, cache size is not growing as quickly and you will thrash your data cache far more quickly if all your pointers double in size. I don't know if immediate mode addressing instructions are common for 64bit operands but if they are, it could thrash your icache sooner as well.
Bandwidth from memory to cache will also be used by these larger pointers.
OTOH, other than disk controller caches (?), what kind of embedded systems need more than 4GB online simultaneously ?
I've been using UBB threads (php + MySQL (et al)) for a public forum for over a year and it has been robust and easy to maintain. Its cost is about the same as vBulletin ($229). Using it for intranet, authentication integration with existing systems will be something to worry about.
OTOH, back in early 1990's era SGI, NNTP/trn was the only way to go. sgi.general, sgi.bad-attitude... ahh, those were the days. Corporate YP solved auth.
I saw the article on Tom's Hardware a couple of days ago and couldn't understand why they were making a big deal about it. The 300Gbyte drive has been out since (at least) July. Fry's has already had a sale where it was $249 with another $50 off in rebate, making it $199 (That sale is over). I've bought and used several of these and they work great (performance isn't critical for my app).
Last year in August (?), Maxtor announced the 240G and 320G drives (4 platters @ 80G), but apparently had problems and never shipped them. 11 months later, they finally ship 300G. OTOH, its the biggest thing out there, so they're doing something right.
Sun is just following SGI down the tubes
on
Merrill Lynch Rips Sun
·
· Score: 5, Insightful
Both SGI and Sun were killed (past tense) by commodity hardware that was "good enough" to take away their sales even before they stopped innovating. In SGI's case, they panicked believing Itanium would come out in 1997 and kill them. They tried to switch to commodity hardware but couldn't stomach it and the dithering ate away at them. Itanium still sucks to this day (but MIPS could never break 1Ghz...).
Sun fell down on the workstation side a long time ago, but their servers were hot thru the.com boom of 2000. Now they're getting killed by Dells at the low end and grids at the high end. They have a huge number of employees because the company is feeding off historical service contracts (the same thing that's keeping SGI on life support). Sun needs to shrink, simplify, and focus or they'll be dead in 10 years also.
Immersion seems more likely to have a strong patent in haptic than SCO has any claims on Linux, but...
This seems like a strategic settlement by Microsoft to punish Sony. Microsoft paying the settlement to Immersion gives Immersion fuel to feed its lawyers in their continued attack on Sony. You can be sure that when Sony settles or is taken to court the total will be $260M, not $26M.
My experience with the cable industry has been that they have huge opportunity, but are too risk-averse technologically to make these things happen. They are still, today, ham-strung by the GI/SA encryption duopoly that has made their settop boxes an embarassment in the CE industry.
Wireless doesn't provide the cable industry with any competitive advantage over a telco/satellite alliance. Both can offer the same improved services w/ or w/o wireless. Wireless scares both cable and DSL providers because it will make sharing of broadband that much easier, so they will make efforts to delay its deployment (although I doubt they can have much effect there).
Finally, offering "remote game-play terminals" as suggested in the ppt slides seems like an interesting idea, but ignores the bandwidth/compression costs of 30fps 1920x1080 low-lateny gaming that will be prevalent by the time this idea is executed. HDTV streams are 13-19Mbit/sec, but compressed-on-the-fly game content will be much higher bandwidth. It will also be continuous for the hours of game console usage/day. There's a good reason to put a powerful computer or a game console on the other end of the wire: 3D graphics are a GREAT compression mechanism.
The 3D graphics world is filled with contradictions. Nvidia has not always released WHQL certified drivers in recent months, but their quality has been the best (my ATI 9700 still hangs the machine with current ATI drivers).
The 4403 drivers are not beta drivers, but were released yesterday as the "Detonator FX" (yawn). They claim to be WHQL-certified, but the asterisk points out that this is only for FX-class graphics boards. My 4600 doesn't qualify. All that being said, they seem fast and reliable in my anecdotal benchmarks/testing.
Regarding illegal tunes: I ran strings on the driver and the only weird thing I saw was the nvcpl.dll knows where 3DSMAX.EXE is installed. Since this is the display properties control panel, its probably inocuous. This doesn't prove that they don't tune for 3DMark2003, though. Reading the report, it does seem suspicious, but I believe the tune would have to be specific to that particular benchmark.
Finally, isn't anyone bothered that all the real world benchmarks show little difference between the cards? No one runs 3DMark2003 for fun. Are we all buying cards in the hopes that Doom3 will save us? Do 3D graphics hardware improvements matter when the software is so far behind?
113 flights, 14 dead astronauts, = 1 in 8 chance of dying?
No. 113 flights, 2 lost flights = 1.77% chance of dying.
Try to live in (or get to) space is a risky undertaking. Yes, there should be a new shuttle, but don't ever expect it to be "safe". I am sure _all_ of today's astronauts would accept even _more_ risk if NASA and the US were willing to take some chances on the Moon, or Mars.
Finally, we need a dumb, heavy-lift rocket. We lost the SaturnV plans, the Energia seems to work but no one is using it now.
http://k26.com/buran/Future/Mars/mars.html
I realize its tradition and there's a certain extra bit of personal ownership involved in having a copy of the contents arrive entirely in my mailbox, but...
Why do newsletters waste the bandwidth of sending out copies of the content to everyone on the list? A small email with a single link to the contents of the new newsletter would work just as well and only those people that read it would spend the bandwidth. In addition, the author could then scrutinize the logs and see what links and what sections generated interest and perhaps better serve the (sometimes not fully understood) audience.
I must have gotten lucky. I bought the $99 DirecTiVO, added a $99 120G drive, and then watched DirecTV "take over" the service and reduce the cost to $5/month. Everything about my TiVO experience has been low cost.
I'm suprised people aren't interested in the MP3 playing. An ethernet-enabled MP3 player puts my computer's stored music in the living room. Up until now, only the Audiotron and a couple of limited-distribution hacks have been built to do this (at least in the $500 category).
PCs in the living room are ugly, loud, and/or have bad user-interface. I have high hopes for TiVOs UI to my MP3s.
I'd be a bit surprised if it did _not_ show up was a USB storage device. I'm guessing the reason they require Windows* is because their host app has a lot of synchronization smarts that they advertise (and can't deliver on Mac/Linux).
ATA/133 support is not important for performance. No one or two disk drives can saturate the bus, but...
ATA/133 is currently the _only_ way to connect HDD greater than 137GB to you system and be able to use the extra space.
IBM is @ 180, WD is @ 200, and Maxtor should be shipping their 320 any day now. For my servers at work and my media storage at home, 4x120 is not enough. Sure, I could use the provided PCI ATA/133 card, but... that's lame.
Slashdot Mirror
Stock options have value and should be expensed somehow, but to "only" expense the gains when the employee exercises and benefits leads to all sorts of counter-productive results.
As CEO, I work hard to increase share price to benefit the shareholders. I somehow achieve my goal, then all my employees (including me) exercise/sell to reap the benefits. Suddenly my earnings take a huge hit. Boom, my stock price crashes. Sure, I could call it a one-time charge, but option exercise/sell is basically out of my control and could happen every quarter. In the end, shareholders and financial analysts would have to ignore this aspect of my earnings, which brings us back to the situation we have today.
The speed of the high-level language is getting less and less important. Java3D is an API that is implmented as native code (JNI?) on the target system. The application can construct the data using Java prior to runtime, but display lists and scene graphs take the high-level language out of the critical path.
It's true that SciVis, CFD and the like aren't going to be happy running in Java, but I believe you can do a good flight sim or game using Java3D.
(Disclaimer: I haven't used Java3D... OpenGL and Performer are/were my preference, but Java3D was designed with the help of the SGI Performer team)
I didn't watch any of the episodes on Fox because I hadn't even discovered Buffy yet. 6 seasons of DVD Buffy later (waiting for the 7th), we buy Firefly's DVD set and were really impressed.
Firefly, the series, has the usual Joss Whedan good characters interacting in interesting ways.
Apparently they didn't show the pilot first on Fox and I can't imagine how that could have worked. The 2 hour pilot (also called Serenity) is a must see and its CGI special effects are surprisingly good.
15 episodes (pilot + 13) + extras.
Strongly Recommended!
My experience with console development systems is that you have almost no rights to transfer/sell. Everything must be handled through Sony/Microsoft/Nintendo. I predict that, if they want to, they can confiscate all of his development systems. Maybe the rules are different in Japan, but I doubt it.
When I saw this on slashdot, I thought that I had been doing SETI@Home for much longer than that, but apparently I registered May 16th, 1999, early in the UTC. Their news release puts the anniversay as May 17th.
Was that really their first day?
Racksaver was selling dual-machine 1U racks for several years and I owned a few of them. Think deep, not tall. Racksaver seems to have renamed itself Verari and only has dual-Opteron in a 1U now. Most dense configs seem to be blade-based these days. Verari advertises 132 processors in a single rack, but I suspect they are not king in this area.
If Google is innovating in this area, it could either be on price or in density.
The biggest problem right now is that there are new graphics cards coming out every 6 months with architecture changes every 18-24 months. Games have been behind the development curve for a while. Finally the two big game engines come out with new versions, trying to aim at what should be reasonable at the time.
Unfortunately, the rest of the PC hardware has turned into complete commodity and its unclear whether its worth spending another $500 on the rest of the computer to hold the FX6800 when it comes out (things are relatively quiet in the land of CPU and memory, where spending 3x the money may get you a 20% increase).
If you're a 3D software developer trying to pick which features to use to get decent market penetration (yet still take advantage of the new programmability), you're pretty-well hosed right now with the various flavors of pixel/vertex shader instructions and program lengths available on the various cards.
ATI 9600, NV FX5600 - these are the cards/capabilities I would depend on to be widespread in the installed base by Xmas 2005.
All DirecTiVo (including HD), record the digital bit stream directly from the satellite onto the hard drive, so there is no degradation at all. The HD-TiVo added OTA (over-the-air) tuners for the local digital TV broadcasts and those bits are also sent directly to the hard drive. 19Mbits/sec is the maximum HD rate for OTA, while satellite/cable encodings of HD tend to be 13Mbits/sec or less.
In reading the initial comments about the HD-TiVo, there is one complaint that could be a problem for those who are currently using a DirecTV HD receiver like the DTC-100 and a non-HD DirecTiVo.
Apparently, when the HD-TiVo gets a non-HD signal, it doesn't automatically switch its output to 480i/480p. It also doesn't stretch/zoom the image to fill a 16x9 screen. This means you need to manually switch the output if you want your TV's de-interlacer/scaler to adjust the image. Depending on who you ask, this is a no-op, annoying, or fatal. (I'm probably in the annoying camp)
At least according to the reports I could find, Alias and Autodesk are not members/founders of 3DIF. I would be very surprised if a "universal" 3D format is created without their help.
The problem is that there are many diverse needs/users for 3D data and the data is so large and/or hard-to-render that simplifying it to a single format that meets the needs of CAD, film animation, sci-vis, game animation, 3D web content, and GIS is not reasonable.
On the other hand, VRML may have been too early, but a poly+texture+simplebehaviours format that was well supported by all applications would be a good thing to (re)create.
Clipper was the first RISC CPU. It replaced Intergraph's N32032 based-workstations with a 33Mhz, 5 MIPS (what part of that is "RISC"??) processor. As compared to the 32032, though, it rocked.
Compilers were tricky, Green Hills was the only choice.
Good name for a dog... bushdog, RIP.
E911 lets the service provider know where you are by your tower connection (and perhaps by other tower signal strengths). It's expensive to keep track of, but is required by law. They don't have enough financial interest in returning that location information to you yet.
AGPS (assisted GPS) also depends on the service provider to calculate your location, but it actually uses real GPS satellite signals to do it. The signal strengths are uploaded to a server which does the heavy-lifting of figuring out the location. Again, this depends on the SP servers and they aren't going to be terribly interested in returning lots of location points back to you.
Real GPS does the CPU work on the device. There are rumored to be a few phones with this capability, but I use a Garmin Gecko 301. $228, records 10K points, downloads via serial port (yuck), and burns a set of alkaline batteries in 7 hours (or less). IMHO, it'll be a while before location information is as easy to get to as the cheapest GPS. Battery power is just too valuable.
I predict Xbox2 _definitely_ won't be compatible with Xbox1 based on Jen-Hsun's comment.
:-).
Microsoft buys all of its technology for game consoles nearly off-the-shelf. The Xbox1 graphics was a slightly modified (and slightly early) GF3 and the CPU was an OTS Intel P3 733. This allowed them to have vendors bidding against each other and get a very good price (AMD and Gigapixel got screwed last time).
Now that they have chosen ATI and IBM for graphics and CPU, respectively, any technology used in the prior generation is not only hard to emulate but runs into IP land mines.
Sony likes backwards compatibility, owns all of its graphics technology, and works hard to built a cheap version of last-gen into next-gen. The interesting question will be whether PS3 plays PS1 games
Nintendo just doesn't care. They face a similar problem to Microsoft, but have better contracts with their suppliers _and_, in this case, GameCube2 seems to have the same corporate suppliers as Gamecube did (ATI and IBM). Even so, I would guess that the blow off compatibility.
You missed a good point in the original question. Even if you have tons of RAM, cache size is not growing as quickly and you will thrash your data cache far more quickly if all your pointers double in size. I don't know if immediate mode addressing instructions are common for 64bit operands but if they are, it could thrash your icache sooner as well.
Bandwidth from memory to cache will also be used by these larger pointers.
OTOH, other than disk controller caches (?), what kind of embedded systems need more than 4GB online simultaneously ?
I've been using UBB threads (php + MySQL (et al)) for a public forum for over a year and it has been robust and easy to maintain. Its cost is about the same as vBulletin ($229). Using it for intranet, authentication integration with existing systems will be something to worry about.
OTOH, back in early 1990's era SGI, NNTP/trn was the only way to go. sgi.general, sgi.bad-attitude... ahh, those were the days. Corporate YP solved auth.
I saw the article on Tom's Hardware a couple of days ago and couldn't understand why they were making a big deal about it. The 300Gbyte drive has been out since (at least) July. Fry's has already had a sale where it was $249 with another $50 off in rebate, making it $199 (That sale is over). I've bought and used several of these and they work great (performance isn't critical for my app).
Last year in August (?), Maxtor announced the 240G and 320G drives (4 platters @ 80G), but apparently had problems and never shipped them. 11 months later, they finally ship 300G. OTOH, its the biggest thing out there, so they're doing something right.
Both SGI and Sun were killed (past tense) by commodity hardware that was "good enough" to take away their sales even before they stopped innovating. In SGI's case, they panicked believing Itanium would come out in 1997 and kill them. They tried to switch to commodity hardware but couldn't stomach it and the dithering ate away at them. Itanium still sucks to this day (but MIPS could never break 1Ghz...).
.com boom of 2000. Now they're getting killed by Dells at the low end and grids at the high end. They have a huge number of employees because the company is feeding off historical service contracts (the same thing that's keeping SGI on life support). Sun needs to shrink, simplify, and focus or they'll be dead in 10 years also.
Sun fell down on the workstation side a long time ago, but their servers were hot thru the
Immersion seems more likely to have a strong patent in haptic than SCO has any claims on Linux, but...
This seems like a strategic settlement by Microsoft to punish Sony. Microsoft paying the settlement to Immersion gives Immersion fuel to feed its lawyers in their continued attack on Sony. You can be sure that when Sony settles or is taken to court the total will be $260M, not $26M.
My experience with the cable industry has been that they have huge opportunity, but are too risk-averse technologically to make these things happen. They are still, today, ham-strung by the GI/SA encryption duopoly that has made their settop boxes an embarassment in the CE industry.
Wireless doesn't provide the cable industry with any competitive advantage over a telco/satellite alliance. Both can offer the same improved services w/ or w/o wireless. Wireless scares both cable and DSL providers because it will make sharing of broadband that much easier, so they will make efforts to delay its deployment (although I doubt they can have much effect there).
Finally, offering "remote game-play terminals" as suggested in the ppt slides seems like an interesting idea, but ignores the bandwidth/compression costs of 30fps 1920x1080 low-lateny gaming that will be prevalent by the time this idea is executed. HDTV streams are 13-19Mbit/sec, but compressed-on-the-fly game content will be much higher bandwidth. It will also be continuous for the hours of game console usage/day. There's a good reason to put a powerful computer or a game console on the other end of the wire: 3D graphics are a GREAT compression mechanism.
The 3D graphics world is filled with contradictions. Nvidia has not always released WHQL certified drivers in recent months, but their quality has been the best (my ATI 9700 still hangs the machine with current ATI drivers). The 4403 drivers are not beta drivers, but were released yesterday as the "Detonator FX" (yawn). They claim to be WHQL-certified, but the asterisk points out that this is only for FX-class graphics boards. My 4600 doesn't qualify. All that being said, they seem fast and reliable in my anecdotal benchmarks/testing. Regarding illegal tunes: I ran strings on the driver and the only weird thing I saw was the nvcpl.dll knows where 3DSMAX.EXE is installed. Since this is the display properties control panel, its probably inocuous. This doesn't prove that they don't tune for 3DMark2003, though. Reading the report, it does seem suspicious, but I believe the tune would have to be specific to that particular benchmark. Finally, isn't anyone bothered that all the real world benchmarks show little difference between the cards? No one runs 3DMark2003 for fun. Are we all buying cards in the hopes that Doom3 will save us? Do 3D graphics hardware improvements matter when the software is so far behind?
113 flights, 14 dead astronauts, = 1 in 8 chance of dying? No. 113 flights, 2 lost flights = 1.77% chance of dying. Try to live in (or get to) space is a risky undertaking. Yes, there should be a new shuttle, but don't ever expect it to be "safe". I am sure _all_ of today's astronauts would accept even _more_ risk if NASA and the US were willing to take some chances on the Moon, or Mars. Finally, we need a dumb, heavy-lift rocket. We lost the SaturnV plans, the Energia seems to work but no one is using it now. http://k26.com/buran/Future/Mars/mars.html
I realize its tradition and there's a certain extra bit of personal ownership involved in having a copy of the contents arrive entirely in my mailbox, but... Why do newsletters waste the bandwidth of sending out copies of the content to everyone on the list? A small email with a single link to the contents of the new newsletter would work just as well and only those people that read it would spend the bandwidth. In addition, the author could then scrutinize the logs and see what links and what sections generated interest and perhaps better serve the (sometimes not fully understood) audience.
"studies show"? Please cite them. Your comments ring true, but I would love to read the evidence that backs them up.
I must have gotten lucky. I bought the $99 DirecTiVO, added a $99 120G drive, and then watched DirecTV "take over" the service and reduce the cost to $5/month. Everything about my TiVO experience has been low cost. I'm suprised people aren't interested in the MP3 playing. An ethernet-enabled MP3 player puts my computer's stored music in the living room. Up until now, only the Audiotron and a couple of limited-distribution hacks have been built to do this (at least in the $500 category). PCs in the living room are ugly, loud, and/or have bad user-interface. I have high hopes for TiVOs UI to my MP3s.
I'd be a bit surprised if it did _not_ show up was a USB storage device. I'm guessing the reason they require Windows* is because their host app has a lot of synchronization smarts that they advertise (and can't deliver on Mac/Linux).
ATA/133 support is not important for performance. No one or two disk drives can saturate the bus, but... ATA/133 is currently the _only_ way to connect HDD greater than 137GB to you system and be able to use the extra space. IBM is @ 180, WD is @ 200, and Maxtor should be shipping their 320 any day now. For my servers at work and my media storage at home, 4x120 is not enough. Sure, I could use the provided PCI ATA/133 card, but... that's lame.