Even if present fossil fuel plants are used to power EVs, they're still far cleaner per mile than ordinary cars. In California, that works out to about 97% cleaner.
Even the cleanest modern car engine is just plain dirty, even when compared with coal-fired power plants.
And as others have pointed out, as cleaner power plants are brought on line, electric cars will use them too.
OS X is a nice operating system overall, but Expose has got to be one of the most over-hyped computer "innovations" in recent years. On a Powerbook it's far too easy to hit that F11 key by mistake, and the F9 "expose all windows" function just isn't that useful. Because it shrinks all the windows to fit together on the screen without overlapping, their titles and contents are often illegible. So you end up having to move the mouse around to each window in turn to pop up the menu title. That's not very convenient.
Expose is no substitute for a much simpler feature that's been in most X11 window managers for years. E.g., right-clicking on the desktop in fvwm brings up a plain text list of all active windows with their titles and lets you select one. What could be simpler, faster or easier?
Maybe it's because I've only used OS X for four months, but I still think the OS X window manager is inferior in many ways to the far simpler and leaner X11 window managers like fvwm. In OS X I can only resize a window by grabbing the lower right corner of the window; in fvwm I can resize by tugging on any edge or corner. In OS X I can't specify my personal preference of "keyboard focus follows mouse", so I frequently type on the wrong window if I forget to click the mouse after moving. Common with Windows, you have to raise a window to type on it, which can be a pain when copying from another window that then becomes obscured. And then there's that old wretched and clumsy cmd-C/cmd-V Mac scheme for mouse cut-and-pasting, when X11 has long shown the right way it should work. (I haven't mentioned Apple's inexplicable adherence to its childish single-button mouse sans scroll wheel. It's too obvious a shortcoming, but one that's easily overcome with an external mouse.)
Funny how a system so heavily hyped for its "ease of use" can be so much harder to use than a system widely castigated for its difficulty of use.
Just for the record, the only computer interface innovation of recent years that justifies the word is the mouse wheel.
I had no trouble getting it, because I surf through a Squid proxy cache that peers with other caches, and apparently those other caches already had copies of the objects I wanted.
Right, I've been to two of the three (Goldstone and Canberra). After several weeks in Australia, I saw my first kangeroos at the DSN site. I haven't been to Madrid, but photos of the place imply it's fairly dry. Or at least a lot drier than the east coast of the US where I grew up.
It does seem strange to have located the Australian site in Canberra when so much dry and isolated desert is available in the rest of the country.
A JPL tech report gives some rainfall statistics on the three sites. Goldstone is certainly the driest of the three, there's no question about that.
An earlier example was Voyager 2. This spectacularly successful mission almost didn't make it even to Jupiter. Its primary command receiver failed, and the AFC (automatic frequency control) in its backup also failed. That meant the receiver was listening only to a single frequency with almost no tolerance for error. And the precise frequency was a function of component drift, which was in turn mainly a function of receiver temperature.
The failed components never recovered, but JPL was able to work around it. They constructed an elaborate thermal model of the spacecraft to predict the precise temperature (and therefore the operating frequency) of the command receiver. Everything but the kitchen sink went into this model: the effect of attitude on solar heating, the self-generated heat from the electronics, the effect of turning various instruments on or off, the time lags due to structural heat capacities, everything. And it has worked fine ever since.
JPL doesn't get nearly the credit they deserve for their track record in rescuing missions from seemingly fatal failures like these. There's still a pervasive public myth (sustained by the human space flight side of NASA) that only humans in space can fix things when they break. But they seriously overestimate the astronauts' abilities, and they greatly underestimate what a bunch of really smart people can often do from the ground.
The debugging took place over the omnidirectional antenna on the rover (that shiny vertical metal cylinder in many of the photos) precisely so the software on the rover wouldn't have to point the high-gain antenna at the earth.
The low gain antenna operates at much lower data rates, which is one of the reasons the debugging took so long.
As challenging as the links are, they are very well modeled; the signal-to-noise ratio can usually be accurately predicted to a fraction of a dB. This allows the telecom team to confidently schedule downlink sessions at the highest data rate that the link can handle without a significant risk of data loss.
Because very strong forward error correction coding is used, the link tends to be "brittle"; as long as you stay just under the maximum allowable data rate, it will work perfectly. So a lot of work goes into making those accurate link predictions.
But data can still be lost if the signal-to-noise ratio takes an unexpected dip. The most likely cause is rain at the earth station site, as the weather is not as easily predicted and water is a strong absorber of X-band radio energy. Most of the DSN sites are in deserts for just this reason. But even if data is lost, it can be retransmitted later as it is stored on the rover until explicitly deleted.
IPv6 works fine in Mac OS X Panther. You don't even have to turn it on; it "just works". If you have one or more IPv6 routers on your network issuing router advertisements, Panther will just pick them up and do all the right things.
My G5 is the machine I most often reach with IPv6 via 6to4 tunneling when I'm away, since my machines with direct external connections both run Linux and provide 6to4 tunnels. I have two separate IPv6 6to4 Linux gateways for redundancy, one on a TWC cable modem and the other on SBC DSL. My G5 picks up the IPv6 router advertisements for both and adds two IPv6 addresses to its interface list. If one service goes down I can still use the other. And because there's no ingress filtering in 6to4 tunneling I can easily set up asymmetric routes, e.g., going upstream on DSL and downstream on cable. (The cable modem is usually faster in both directions, though, so normally this doesn't make much sense.)
Panther also implements 6to4 tunnels. You have to set them up, but you can have a local 6to4 tunnel interface come up automatically whenever you get on the net with any of your interfaces. This comes in very handy on my G4 PowerBook, although its utility is limited by the prevalence of NATs in certain public Internet access services since 6to4 doesn't work through a NAT.
I'm still new to MacOS having only obtained our Macs just after Panther came out, so I don't know if these features are also present in Jaguar or earlier versions of Mac OS X.
None of the usual consumer-grade routers (e.g., by D-link, Netgear, Linksys, etc) currently implement 6to4 tunneling, to my knowledge. It would be truly wonderful if they did.
Linux supports 6to4 tunneling very well. It coexists very nicely on the same box that provides a whole bunch of other services to my home network: IPv4 NAT routing, incoming and outgoing mail relay, IMAP, web proxy caching, etc. So if you already have a Linux box running 7/24, it's trivial to add 6to4 functionality. But for the many who aren't willing to leave PCs on all the time, adding this function to consumer-grade routers would be a big win.
It's a major myth that you have to wait for the carriers and ISPs to adopt IPv6 before it can become useful to you. All that matters is that your own computers support it, and Windows XP, Linux and Mac OS X all do. I already find IPv6 extremely useful for remotely accessing my own machines even though all my ISPs are completely ignorant of IPv6.
I predict that 6to4 tunneling alongside IPv4 NAT will become a very popular configuration for home routers. IPv4 through NATs will continue to work just fine for all of the things you already do on the net, such as running email and web clients. IPv6 will be primarily useful in running multiple servers that you want to make accessible to the outside world when you're otherwise stuck behind a NAT with a single IP address. This includes VoIP and many other peer-to-peer services of the kind that the Internet was originally designed to support.
In short, IPv6, implemented in the near term with 6to4 tunneling, is the best hope we have of restoring the transparent end-to-end architecture that originally made the Internet great.
The only real drawback to 6to4 is that the 6to4 gateway must have at least one globally routable IPv4 address, not a RFC1918 address (e.g. 192.168.xxx.yy or 10.xxx.yyy.zzz). This is not a problem with most cable modem and DSL services, but if you're already behind somebody else's NAT you can't use 6to4. This is a problem with that DSL service in hotel rooms as well as many 802.11 hotspots.
There is an alternative IPv6 tunneling scheme called Teredo that's designed to run IPv6 through a NAT, but I don't know much about it. If at all possible, I would try to get whoever runs your router to implement 6to4.
That's why it would be a truly wonderful thing if the commodity router vendors were to implement 6to4 tunnels.
It's really great to see people finally enhancing these boxes. These routers have ideal form factors compared to, say, a dedicated router PC running Linux, but their default firmware has always been very poor.
I didn't see one feature mentioned that I'd really, really like to see added to these boxes: an IPv6 6to4 tunnel. This is an ideal way to penetrate a NAT so you can establish direct TCP connections (and speak UDP) to any servers on your LAN from the outside. IPv6 support has been in all of the major operating systems for some time now, including Windows XP, Linux and Mac OS X, and while not every application is IPv6 ready, the important ones (like SSH) already are.
If 6to4 tunneling could be added to these consumer routers alongside IPv4 NAT, IPv6 stands to really take off without any help whatsoever from the ISPs. In fact, I almost prefer that my ISP not implement native IPv6. I like the fact that they now carry my encapsulated IPv6 packets without any ingress filtering, port blocking or other end-to-end-wrecking nonsense, and that they are oblivious to (much less control) the IPv6 address space. If or when the ISPs do implement native IPv6, you can bet that they'll exercise the same degree of arbitrary control that they now do over IPv4.
Another good example of directly relevant prior art that should completely bust this patent is the ampr.org domain, which has existed since the mid 1980s. AMPR stands for AMateur Packet Radio, and domain names in that zone are all of the form "personal_callsign.AMPR.ORG", e.g., ka9q.ampr.org. Until I stopped using that name because of spam, one would send me mail as karn@ka9q.ampr.org.
Except that I seriously doubt anyone who licenses the FairPlay scheme would be permitted to release an open source player. The accessibility of a file format to open source software is very important to me, and that's why I prefer Ogg Vorbis.
Yes, the iTunes application is distinct from the iTMS, but until it's possible to access the iTMS with some other application, the interchanging of the names will probably continue.
I know about the Ogg Vorbis plug-in for iTunes on the Mac. I'm using it, and it works fine. The only problem I've encountered is that it doesn't work if you configure iTunes to leave the music on a network server when you add it to your library. To play Ogg tunes, you have to copy everything to the Mac that's running iTunes.
Spec or no spec, if I can't listen to my purchased iTunes with open source software on the platform of my choice, then the format might as well be completely proprietary. That's all I'm saying.
DVD Jon may have just solved this problem. We'll see what happens.
I understand your point, but my CD collection is large enough that I really don't want to have to rip it more than once, but my ears are now old enough (I'm 47) that I can't really hear any significant degradation when I convert AAC to Ogg. Sure, I would also prefer to avoid transcoding from one lossy format to another, but that will have to wait for iTunes or a similar service to provide native Ogg files. That seems even less likely than Apple adding Ogg support to the iPod.
If Apple really wanted to lock people into iTunes, why did they provide that escape hatch that lets you burn uncrippled songs out to CD? Could it be the reason iTunes took off while the more restrictive services have barely left the starting gate?
This business is supposed to be all about giving people what they want, so why shouldn't I say what I want? I feel strongly about the importance of open, nonproprietary file formats, so I very much want Ogg Vorbis support in an iPod. I'd buy one in a heartbeat if Apple added it.
Sure, I like the iPod's user interface and its slick physical design, and for me those features can overcome the high price and the limited life of the built-in Li-ion battery. But they're not enough to overcome a very basic functional shortcoming: no support for any practical, non-proprietary file format. And because the iPod isn't the only game in town, I will gladly patronize their competition (iRiver) while I wait.
I don't think "bundling" per se was ever the real problem with Microsoft. If you didn't want to use IE, you could just get rid of the icon. Who cares about the files and registry settings that are left? Disk space is cheap.
No, the real issue was and still is all about proprietary control of widely used file formats. Abusive software monopolies can't gain nearly as much traction when they have to adhere to open, accessible standards that they can't arbitrarily change to their own benefit.
And that's exactly the issue here. iTunes may be "based on" a standard called MPEG-4, but with the addition of DRM it might as well be completely proprietary. Even without DRM, there's still the problem of the patents on MPEG-4.
At least Apple provided an all-important escape hatch, the ability to burn unrestricted CDs. So the only real problem with iTunes is that high $0.99 price. When I do buy some tunes, I immediately burn them to an audio CD, rip it back in and re-encode it in Ogg Vorbis, my format of choice. Although I own several Macs, I have no plans to buy an iPod unless or until Apple officially supports Ogg Vorbis, a format I can already play on all of my computers with unencumbered software. Until then, I'll use the iRiver flash player I bought the day after they released Ogg support.
Hint #1: the camera physically turns as it takes each picture in the panorama. Hint #2: some objects in the field of view are much closer to the camera than others.
I don't know this market segment very well, but don't most of the big media houses build their own in-house tools? I thought that's why so many of them are building big Linux clusters.
I have written code for both Intel and PPC SIMD instruction sets: MMX/SSE/SSE2 on Intel and Altivec on the PPC. I'd say Altivec is roughly comparable to SSE2, only significantly cleaner, like the rest of the PPC architecture. I was able to make an Altivec version of my k=7 r=1/2 Viterbi decoder execute at about 33 megabits/sec on a single 2.0 GHz G5, while my SSE2 version of the same decoder runs at only about 21 megabits/sec on a 2.6 GHz P4.
That said, bang for the buck is still what counts when you need a lot of cycles for some specialized computation, and Apple machines still command quite a price premium. Mac OS X, which is vastly superior to Windows on a desktop, would seem to have little going for it over Linux or BSD in a custom "supercomputer" farm. So I guess the question is whether that premium is still worth it for the hardware. I don't know.
Probably because the XServe is intended for the high-end server market, and won't be built in the same quantities as the desktop PowerMac dual G5. Rackmount servers are almost always more expensive than their desktop counterparts.
For a supercomputer array, I'd think that plain BSD or PPC Linux would serve their needs as well as OS X Server or OS X, but they might not get Apple to negotiate to take it out and drop the price since Apple's marginal cost is zero.
I am wondering just what the real market is for the XServe. Mac OS X shines as a desktop system; if you just want an ordinary server, I don't see that Mac OS X has any big advantage over Linux or BSD, even when serving Macs. So you might as well get any of the much cheaper x86 server boxes and run Linux or BSD. And you won't have to pay for those OS upgrades.
Because of its nice, fast PPC chips, the XServe would certainly appeal to anyone building a high-end parallel supercomputer like that at Va Tech, and as soon as I saw the XServe announcement I wondered if the guys at Va Tech were wishing they'd waited. But if the unit price is higher and they can't handle the extra heat load anyway, then it's not a big deal.
I have a dual-CPU G5. It pours out a surprising amount of heat. So I can only imagine what it's like to get the heat of 1,100 G5s out of a room.
Indeed, total system performance is what matters. But the best way to determine that isn't necessarily to run two entirely different packaged binaries, even if they have the same name from the same vendor. You don't know what optimizations, if any, have been made for the two different CPUs, so you don't really know if it's a fair comparison. You don't know if the version that seems slow today has a complete rewrite coming tomorrow that will make it much faster.
While code fragments may not seem like relevant benchmarks to end users who just want to play games, at least those running the benchmarks can have a better idea of just what they're comparing.
Even then you are often seeing the effects of two different compilers that may not be at the same level of maturity.
So the bottom line is very simple: take all benchmarks with a big grain of salt. What's faster today may not be faster tomorrow.
I have never noticed much of a positive correlation between religious zeal and moral behavior toward other people. If anything, I often notice a negative correlation.
Linus cites this defintion from the US copyright act:
"The term 'financial gain' includes receipt, or expectation of receipt, of anything of value, including the receipt of other copyrighted works."
This text sounds awfully familiar. Wasn't it placed there by the No Electronic Theft Act just a few years ago in response to the Brian LaMacchia case? He was accused of exchanging copyrighted software not for money, but in expectation of receiving other pirated software. At the time, I believe you had to take money or tangible property for piracy to constitute a criminal offense. Non-commercial file swapping didn't qualify; it was merely grounds for a civil suit by the copyright holders. NETA plugged up this "loophole".
That Linus could take a lemon like the NETA and turn it into lemonade like this is just wonderful.
A few years ago while going to one of the IETF meetings in Minneapolis I noticed a Windows error message spread across two adjacent flight status monitors at the airport. It complained of a duplicate IP address on the local network. I saw no mouse with which to click "OK".
Almost a week later, when I returned home, the message was still there.
Slashdot Mirror
Even if present fossil fuel plants are used to power EVs, they're still far cleaner per mile than ordinary cars. In California, that works out to about 97% cleaner.
Even the cleanest modern car engine is just plain dirty, even when compared with coal-fired power plants.
And as others have pointed out, as cleaner power plants are brought on line, electric cars will use them too.
Expose is no substitute for a much simpler feature that's been in most X11 window managers for years. E.g., right-clicking on the desktop in fvwm brings up a plain text list of all active windows with their titles and lets you select one. What could be simpler, faster or easier?
Maybe it's because I've only used OS X for four months, but I still think the OS X window manager is inferior in many ways to the far simpler and leaner X11 window managers like fvwm. In OS X I can only resize a window by grabbing the lower right corner of the window; in fvwm I can resize by tugging on any edge or corner. In OS X I can't specify my personal preference of "keyboard focus follows mouse", so I frequently type on the wrong window if I forget to click the mouse after moving. Common with Windows, you have to raise a window to type on it, which can be a pain when copying from another window that then becomes obscured. And then there's that old wretched and clumsy cmd-C/cmd-V Mac scheme for mouse cut-and-pasting, when X11 has long shown the right way it should work. (I haven't mentioned Apple's inexplicable adherence to its childish single-button mouse sans scroll wheel. It's too obvious a shortcoming, but one that's easily overcome with an external mouse.)
Funny how a system so heavily hyped for its "ease of use" can be so much harder to use than a system widely castigated for its difficulty of use.
Just for the record, the only computer interface innovation of recent years that justifies the word is the mouse wheel.
Squid is a Good Thing. Use it.
It does seem strange to have located the Australian site in Canberra when so much dry and isolated desert is available in the rest of the country.
A JPL tech report gives some rainfall statistics on the three sites. Goldstone is certainly the driest of the three, there's no question about that.
The failed components never recovered, but JPL was able to work around it. They constructed an elaborate thermal model of the spacecraft to predict the precise temperature (and therefore the operating frequency) of the command receiver. Everything but the kitchen sink went into this model: the effect of attitude on solar heating, the self-generated heat from the electronics, the effect of turning various instruments on or off, the time lags due to structural heat capacities, everything. And it has worked fine ever since.
JPL doesn't get nearly the credit they deserve for their track record in rescuing missions from seemingly fatal failures like these. There's still a pervasive public myth (sustained by the human space flight side of NASA) that only humans in space can fix things when they break. But they seriously overestimate the astronauts' abilities, and they greatly underestimate what a bunch of really smart people can often do from the ground.
The low gain antenna operates at much lower data rates, which is one of the reasons the debugging took so long.
Because very strong forward error correction coding is used, the link tends to be "brittle"; as long as you stay just under the maximum allowable data rate, it will work perfectly. So a lot of work goes into making those accurate link predictions.
But data can still be lost if the signal-to-noise ratio takes an unexpected dip. The most likely cause is rain at the earth station site, as the weather is not as easily predicted and water is a strong absorber of X-band radio energy. Most of the DSN sites are in deserts for just this reason. But even if data is lost, it can be retransmitted later as it is stored on the rover until explicitly deleted.
My G5 is the machine I most often reach with IPv6 via 6to4 tunneling when I'm away, since my machines with direct external connections both run Linux and provide 6to4 tunnels. I have two separate IPv6 6to4 Linux gateways for redundancy, one on a TWC cable modem and the other on SBC DSL. My G5 picks up the IPv6 router advertisements for both and adds two IPv6 addresses to its interface list. If one service goes down I can still use the other. And because there's no ingress filtering in 6to4 tunneling I can easily set up asymmetric routes, e.g., going upstream on DSL and downstream on cable. (The cable modem is usually faster in both directions, though, so normally this doesn't make much sense.)
Panther also implements 6to4 tunnels. You have to set them up, but you can have a local 6to4 tunnel interface come up automatically whenever you get on the net with any of your interfaces. This comes in very handy on my G4 PowerBook, although its utility is limited by the prevalence of NATs in certain public Internet access services since 6to4 doesn't work through a NAT.
I'm still new to MacOS having only obtained our Macs just after Panther came out, so I don't know if these features are also present in Jaguar or earlier versions of Mac OS X.
Linux supports 6to4 tunneling very well. It coexists very nicely on the same box that provides a whole bunch of other services to my home network: IPv4 NAT routing, incoming and outgoing mail relay, IMAP, web proxy caching, etc. So if you already have a Linux box running 7/24, it's trivial to add 6to4 functionality. But for the many who aren't willing to leave PCs on all the time, adding this function to consumer-grade routers would be a big win.
It's a major myth that you have to wait for the carriers and ISPs to adopt IPv6 before it can become useful to you. All that matters is that your own computers support it, and Windows XP, Linux and Mac OS X all do. I already find IPv6 extremely useful for remotely accessing my own machines even though all my ISPs are completely ignorant of IPv6.
I predict that 6to4 tunneling alongside IPv4 NAT will become a very popular configuration for home routers. IPv4 through NATs will continue to work just fine for all of the things you already do on the net, such as running email and web clients. IPv6 will be primarily useful in running multiple servers that you want to make accessible to the outside world when you're otherwise stuck behind a NAT with a single IP address. This includes VoIP and many other peer-to-peer services of the kind that the Internet was originally designed to support.
In short, IPv6, implemented in the near term with 6to4 tunneling, is the best hope we have of restoring the transparent end-to-end architecture that originally made the Internet great.
There is an alternative IPv6 tunneling scheme called Teredo that's designed to run IPv6 through a NAT, but I don't know much about it. If at all possible, I would try to get whoever runs your router to implement 6to4.
That's why it would be a truly wonderful thing if the commodity router vendors were to implement 6to4 tunnels.
I didn't see one feature mentioned that I'd really, really like to see added to these boxes: an IPv6 6to4 tunnel. This is an ideal way to penetrate a NAT so you can establish direct TCP connections (and speak UDP) to any servers on your LAN from the outside. IPv6 support has been in all of the major operating systems for some time now, including Windows XP, Linux and Mac OS X, and while not every application is IPv6 ready, the important ones (like SSH) already are.
If 6to4 tunneling could be added to these consumer routers alongside IPv4 NAT, IPv6 stands to really take off without any help whatsoever from the ISPs. In fact, I almost prefer that my ISP not implement native IPv6. I like the fact that they now carry my encapsulated IPv6 packets without any ingress filtering, port blocking or other end-to-end-wrecking nonsense, and that they are oblivious to (much less control) the IPv6 address space. If or when the ISPs do implement native IPv6, you can bet that they'll exercise the same degree of arbitrary control that they now do over IPv4.
Another good example of directly relevant prior art that should completely bust this patent is the ampr.org domain, which has existed since the mid 1980s. AMPR stands for AMateur Packet Radio, and domain names in that zone are all of the form "personal_callsign.AMPR.ORG", e.g., ka9q.ampr.org. Until I stopped using that name because of spam, one would send me mail as karn@ka9q.ampr.org.
This patent would appear to be a perfect example.
Yes, the iTunes application is distinct from the iTMS, but until it's possible to access the iTMS with some other application, the interchanging of the names will probably continue.
I know about the Ogg Vorbis plug-in for iTunes on the Mac. I'm using it, and it works fine. The only problem I've encountered is that it doesn't work if you configure iTunes to leave the music on a network server when you add it to your library. To play Ogg tunes, you have to copy everything to the Mac that's running iTunes.
DVD Jon may have just solved this problem. We'll see what happens.
I understand your point, but my CD collection is large enough that I really don't want to have to rip it more than once, but my ears are now old enough (I'm 47) that I can't really hear any significant degradation when I convert AAC to Ogg. Sure, I would also prefer to avoid transcoding from one lossy format to another, but that will have to wait for iTunes or a similar service to provide native Ogg files. That seems even less likely than Apple adding Ogg support to the iPod.
This business is supposed to be all about giving people what they want, so why shouldn't I say what I want? I feel strongly about the importance of open, nonproprietary file formats, so I very much want Ogg Vorbis support in an iPod. I'd buy one in a heartbeat if Apple added it.
Sure, I like the iPod's user interface and its slick physical design, and for me those features can overcome the high price and the limited life of the built-in Li-ion battery. But they're not enough to overcome a very basic functional shortcoming: no support for any practical, non-proprietary file format. And because the iPod isn't the only game in town, I will gladly patronize their competition (iRiver) while I wait.
No, the real issue was and still is all about proprietary control of widely used file formats. Abusive software monopolies can't gain nearly as much traction when they have to adhere to open, accessible standards that they can't arbitrarily change to their own benefit.
And that's exactly the issue here. iTunes may be "based on" a standard called MPEG-4, but with the addition of DRM it might as well be completely proprietary. Even without DRM, there's still the problem of the patents on MPEG-4.
At least Apple provided an all-important escape hatch, the ability to burn unrestricted CDs. So the only real problem with iTunes is that high $0.99 price. When I do buy some tunes, I immediately burn them to an audio CD, rip it back in and re-encode it in Ogg Vorbis, my format of choice. Although I own several Macs, I have no plans to buy an iPod unless or until Apple officially supports Ogg Vorbis, a format I can already play on all of my computers with unencumbered software. Until then, I'll use the iRiver flash player I bought the day after they released Ogg support.
You try it and see if you can do better.
I have written code for both Intel and PPC SIMD instruction sets: MMX/SSE/SSE2 on Intel and Altivec on the PPC. I'd say Altivec is roughly comparable to SSE2, only significantly cleaner, like the rest of the PPC architecture. I was able to make an Altivec version of my k=7 r=1/2 Viterbi decoder execute at about 33 megabits/sec on a single 2.0 GHz G5, while my SSE2 version of the same decoder runs at only about 21 megabits/sec on a 2.6 GHz P4.
That said, bang for the buck is still what counts when you need a lot of cycles for some specialized computation, and Apple machines still command quite a price premium. Mac OS X, which is vastly superior to Windows on a desktop, would seem to have little going for it over Linux or BSD in a custom "supercomputer" farm. So I guess the question is whether that premium is still worth it for the hardware. I don't know.
For a supercomputer array, I'd think that plain BSD or PPC Linux would serve their needs as well as OS X Server or OS X, but they might not get Apple to negotiate to take it out and drop the price since Apple's marginal cost is zero.
I am wondering just what the real market is for the XServe. Mac OS X shines as a desktop system; if you just want an ordinary server, I don't see that Mac OS X has any big advantage over Linux or BSD, even when serving Macs. So you might as well get any of the much cheaper x86 server boxes and run Linux or BSD. And you won't have to pay for those OS upgrades.
Because of its nice, fast PPC chips, the XServe would certainly appeal to anyone building a high-end parallel supercomputer like that at Va Tech, and as soon as I saw the XServe announcement I wondered if the guys at Va Tech were wishing they'd waited. But if the unit price is higher and they can't handle the extra heat load anyway, then it's not a big deal.
I have a dual-CPU G5. It pours out a surprising amount of heat. So I can only imagine what it's like to get the heat of 1,100 G5s out of a room.
While code fragments may not seem like relevant benchmarks to end users who just want to play games, at least those running the benchmarks can have a better idea of just what they're comparing.
Even then you are often seeing the effects of two different compilers that may not be at the same level of maturity.
So the bottom line is very simple: take all benchmarks with a big grain of salt. What's faster today may not be faster tomorrow.
I have never noticed much of a positive correlation between religious zeal and moral behavior toward other people. If anything, I often notice a negative correlation.
This text sounds awfully familiar. Wasn't it placed there by the No Electronic Theft Act just a few years ago in response to the Brian LaMacchia case? He was accused of exchanging copyrighted software not for money, but in expectation of receiving other pirated software. At the time, I believe you had to take money or tangible property for piracy to constitute a criminal offense. Non-commercial file swapping didn't qualify; it was merely grounds for a civil suit by the copyright holders. NETA plugged up this "loophole".
That Linus could take a lemon like the NETA and turn it into lemonade like this is just wonderful.
Almost a week later, when I returned home, the message was still there.