Actually, mobile Athlon XP's are also multiplier unlocked, work in any motherboard that can take a desktop Barton, and only cost a couple of bucks more than the equivalently rated desktop model.
I think mobile Athlon 64's let you change the multiplier also, but doesn't let you set it higher than the stock setting. Not sure if that helps any for overclocking purposes.
Tim Sweeney Yukihiro "Matz" Matsumoto (Ruby) Alan Kay Ken Arnold (jini) Doug Lea (Java guru) Paul Graham (spam fighter, Lisp advocate)
I just realized I would be hard pressed to name six professional baseball players. Let's see... Ken Griffey Jr., Nomar Garciaparra, Pedro Martinez, Derek Jeter, Barry Bonds, and... uhhh... ummm, Babe Ruth.
I would like to take it a little further, and point out that you can't even always know what is costly for your language--you can only know what's costly for the specific VM's you're using.
Over the years, many operations that were extremely slow in Java have become very fast, and (to a lesser extent) vice versa. The HotSpot JIT compiler and generational garbage collection, for two examples, turned the rules upside down from one version of the JVM to the next. In some versions of JRE 1.4, reusing StringBuffers is incredibly expensive, for other versions it's cheaper than creating new ones. Entering a synchronized block was once expensive, even without contention, and now is cheap. It used to improve performance to mark your methods "final", now it supposedly makes no difference at all. The examples go on and on.
Then there are differences between operating systems, or even different versions of operating systems--isn't Linux's new threads implementation like a thousand times faster than the old one?
I agree with your mantra "write simple code and optimize when needed", but would also like to add some wise person's first rule of optimization: "Forget everything you know."
Too bad external drives are so much more expensive; for example, sonystyle.com lists their latest internal DVD+/-RW for $139.99, and the corresponding external model for $259.99.
Seems like a high price for a user to pay to not have to crack the box (and fiddle with IDE cables/jumpers, etc) at install time. It would also annoy me to have to come up with another power outlet for each drive. And don't most of those external drives have their own noisy little fans?
Did the Dell look like this machine on the inside? And if so, was the noise mostly the whooshing of air, or was it low frequency vibration, or a very high pitched buzzing? If #2 or #3, it was likely the hard drive (though 3 could also be due to bad fans).
If it was whooshing air, then I happily stand corrected, and hurrah for Apple for being better and more innovative than Dell in yet another way.
Yeah, I thought that was weird, but, if you take "Mac OS X" off the sentence "Mac OS X dynamically adjusts the flow of the fluid and the speed of the fans based on temperature" then you'd be describing any heatpiped heatsink with a thermal monitoring fan. The rate at which heat is transferred through a heat pipe is inherently a direct result of the temperature being applied--more heat, faster transfer. I wouldn't put it past Apple's marketing dept to call that "dynamically adjusting".
It is weird about the Mac OS X bit, though. But then, I try not to read too much into marketing people's rendering of things.:)
I'm not at all saying that the G5 is a loud computer... far from it. Certainly it's quiet enough for most purposes. But I think that Dell desktops are quieter. (I've not compared them in the same room so I don't know for sure.)
I'm also not saying it's not impressive that the case has 9 fans and achieves its level of quietness. But when I buy or build a PC, I'm interested in the total dB, or the dB/performance ratio, and not the dB/fan ratio.
I might not have brought it up if OP hadn't poo-pooed Dell, just assuming that Dells are loud and G5's are quiet. Dells are quiet, too. Sure, unlike Apple, Dell may put cost efficiency before design--but that doesn't mean it's just a bunch of monkeys running around their engineering department.
I was surprised to see that a G5 at 2GHz dissipates less heat than even a non-Xeon P4 at, say, 3GHz. That being the case, I think maybe Apple just really decided to swim upstream with those nine fans, considering many PCs get along fine with far fewer fans, no "air channels", and no perforated case (and achieve the same noise level).
It looks like the "liquid cooling" may be a heat pipe, in which case I take back what I said about it being a necessary evil. Heat pipe equipped heatsinks are good; they don't cost much more than traditional heatsinks and provide much better performance, and they don't leak.
It's not hard to believe your modern desktop is quieter than your modern laptop--you can't fit big, slow fans in laptops. And I was only referring to Dell desktops.
Does your Dell Latitude burn your lap like a G4 PowerBook would?
I'm not anti-Apple, really. But they're pretty much subject to the same engineering constraints as everyone else, albeit with a higher budget (I imagine). Given the size and variety of the PC industry, is it really so hard to believe that Apple has not built a machine that is the fastest AND quietest AND coolest-running desktop in the world?
The Dells I'm talking about are of the dark grey variety. If you pop the hood you should see a green plastic hood over the processor, which covers a big slow 92mm fan (mounted on the back panel) and a big heatsink on the processor.
My parents have one of these and the two fans are inaudible from more than 12 inches away. They happen to have a Western Digital hard drive so you can still hear a slight idle whine and you definitely hear seeks, but I would assume most Dells today would have drives with fluid bearings. In any case, if the hard drive is the noisiest component, you can slap it in a $25 NoiseControl NoVibes frame, problem solved (though unfortunately I don't think you can fit it in a PowerMac).
I read somewhere that when Dell started selling a lot of PCs in Singapore that they began getting serious about quietness; apparently Singaporeans care a lot about office ergonomics. So Dell created an acoustics lab, and the results were so good that they started quieting all(?) their computers.
(I don't know if this applies to their Precision workstations or their top of the line Dimension XPS, as I haven't heard those. But it seems to be the case for all mainstream Dimensions.)
I personally disagree that the Mac is amazingly quiet for having any fans. But after spending considerable effort silencing my own computer, and lurking around the SilentPCReview forums, my scale of what's quiet and what's not has been recalibrated.
Those big air channels don't come without a cost. There is only one external drive bay and two hard drive bays, and three PCI slots.
A PC case of similar size has maybe four external drive bays and five hard drive bays, and five PCI slots plus an AGP slot.
A PC case with similar expandability, like the Antec Aria, is much smaller. (Though the Aria doesn't have two CPUs and can't hold 8 sticks of memory.)
While the insides of most PC cases may not look neat and tidy, it's child's play to buy or build a quiet, powerful PC that never overheats no matter what the load. So while PC cases may not (in general) have air channels, I'd say they work as well as they need to (though Intel apparently disagrees, since the BTX form factor is coming and is supposed to be designed for airflow).
I don't think you could reasonably expect a mainstream manufacturer to give up the expandability for an academic increase in cooling efficiency, although Shuttle has clearly demonstrated that at least some people are willing to give up expandibility for a slick form factor.
The G5's are beautiful though; I wouldn't mind owning one.
It does look like a heat pipe. All the newest bestest heatsinks in the PC aftermarket are using heat pipes, they've been around for a while now.
I wish they had just said "heat pipe" instead of "innovative liquid cooling", which to us obviously sounds like a traditional liquid cooling setup a la Koolance. For me, that would be a big negative; I would really rather avoid the complexity and risks of liquid cooling.
Nine fans and 21 sensors, generating half as many decibels. Now I'm not an Apple fan-boy but that's the level of attention to detail that seperates Apple from Dell, etc.
Most Dell desktops are a fair sight quieter than the G5, from what I've seen (and heard). It's impressive that Apple got a computer that needs nine fans to be fairly quiet, but with most Dells, well, you only need two fans (one in the PSU and one over the CPU, with a plastic hood).
The PowerMacs are impressive for all sorts of other reasons, but saying "amazingly quiet for having nine fans" is progress seems backward to me. Same with the liquid cooling; it's only there because the G5 runs incredibly hot. I'm sure people will be touting liquid cooling as an advantage, when really I think it should be viewed as a necessary evil. Wouldn't it be better to use a processor that doesn't need to be liquid cooled??
It looks like all netTunes does is remote the iTunes UI, pixel for pixel. If so, and if it works with iTunes 4.6 at all, the AirTunes feature should work fine.
That's not what "defensive patenting" means. It means your competitor may come at you with their patent on invention X, so you should be prepared to strike back with your patent on inventions Y and Z. Like a nuclear arms race, the desired outcome is not so much annihilation of your enemy as it is a cautious stalemate (or cross-licensing deals).
Not that I think this is a good state of affairs, but "defensive patenting" is not an empty argument.
The words "admit", "regret", "apologize", or "sorry".
I want to give the JBoss folks the benefit of the doubt, and I'm sure many others in the Java/J2EE community want to, too... but they just keep making it so damn hard.
Seems like it's about high time someone wrote a free/open-source version. The tools are all out there: Lucene or Lucene.NET as the search engine, IFilter for tokenizing Office and PDF (the Office IFilter comes with Windows, and the PDF one is downloadable from Adobe's site).
All you'd really need to do is put together a UI and an indexing service... both easy to do with.NET and not terribly difficult to do with Java.
Oh, I think some of those products also index the messages stored in your e-mail client. Not sure how you'd go about that... but at least you could leave an API so somebody else who knows could do it.
Anyway, yeah, I understand that. My question is whether, for the specific purpose of spam filtering, it results in improved performance, and if so whether it's been documented anywhere.
The clustering stuff is certainly interesting for other purposes, and I'm glad there are people out there not only writing the software, but integrating it into the OS. The graphic and industrial designers aren't the only smart people at Apple.
That actually tends not to happen. Most Bayesian filtering packages are weighted very conservatively, so that one or two highly non-spam tokens (like your grandma's e-mail address, or the name of the uncle who is on the little blue pill) will more than counterbalance the spam tokens.
Again, what's intuitive doesn't play out in practice... this seems to be a common theme in the world of statistical spam filtering. For example, you'd think the word "free" would be pretty spammy... in my corpus, it only gets a score of.406 (where 0 is least spammy and 1 is most spammy, and an e-mail must have an aggregate score of.9 to be classified as spam). On the other hand, "sir" gets.945 and "madam" gets.987.
Is there any hard data out there that shows the cluster analysis actually improves on the better Bayesian algos out there? After all, most of the good ones also achieve the 98%+ that this article cites.
According to the FAQ of SpamBayes (I think), they're always getting suggestions of ways to tweak their algos that would "obviously" improve the result, but in almost every case it either makes no difference or hurts accuracy, when actually tested on real data.
These aren't clear cut engineering discussions. Check out some of the threads; they are mostly political and personal in nature, or else just generally trumpeting how good JBoss is.
There's a big difference between an opinion of that nature coming from an employee of the company in question, vs. the opinion of a more or less objective third party. I for one was definitely duped by some of these fake posts into thinking JBoss was much more popular and much better received than it actually is.
I work for a software startup that has been getting a fair bit of buzz in the media and in the blogsphere. Anytime I join a discussion, the first thing I say is that I work for the company. To omit this information is simply disingenuous, and it's something I believe many of JBoss's most outspoken detractors would never do.
It's true that a lot of prominent people in the Java community and esp. on TSS.com have very deep grudges against JBoss. No doubt there's a lot of bias on both sides of the fence. But the JBoss crew always seems to fight a little dirtier, or in this case, a lot dirtier.
Slashdot Mirror
The converter cost $150+ last time I checked, and wasn't offered by Apple, but a company called Gefen.
Submitter is Colgate fanboy?
Actually, mobile Athlon XP's are also multiplier unlocked, work in any motherboard that can take a desktop Barton, and only cost a couple of bucks more than the equivalently rated desktop model.
I think mobile Athlon 64's let you change the multiplier also, but doesn't let you set it higher than the stock setting. Not sure if that helps any for overclocking purposes.
In no particular order:
Tim Sweeney
Yukihiro "Matz" Matsumoto (Ruby)
Alan Kay
Ken Arnold (jini)
Doug Lea (Java guru)
Paul Graham (spam fighter, Lisp advocate)
I just realized I would be hard pressed to name six professional baseball players. Let's see... Ken Griffey Jr., Nomar Garciaparra, Pedro Martinez, Derek Jeter, Barry Bonds, and... uhhh... ummm, Babe Ruth.
I would like to take it a little further, and point out that you can't even always know what is costly for your language--you can only know what's costly for the specific VM's you're using.
Over the years, many operations that were extremely slow in Java have become very fast, and (to a lesser extent) vice versa. The HotSpot JIT compiler and generational garbage collection, for two examples, turned the rules upside down from one version of the JVM to the next. In some versions of JRE 1.4, reusing StringBuffers is incredibly expensive, for other versions it's cheaper than creating new ones. Entering a synchronized block was once expensive, even without contention, and now is cheap. It used to improve performance to mark your methods "final", now it supposedly makes no difference at all. The examples go on and on.
Then there are differences between operating systems, or even different versions of operating systems--isn't Linux's new threads implementation like a thousand times faster than the old one?
I agree with your mantra "write simple code and optimize when needed", but would also like to add some wise person's first rule of optimization: "Forget everything you know."
Seems like a high price for a user to pay to not have to crack the box (and fiddle with IDE cables/jumpers, etc) at install time. It would also annoy me to have to come up with another power outlet for each drive. And don't most of those external drives have their own noisy little fans?
If it was whooshing air, then I happily stand corrected, and hurrah for Apple for being better and more innovative than Dell in yet another way.
It is weird about the Mac OS X bit, though. But then, I try not to read too much into marketing people's rendering of things. :)
I'm also not saying it's not impressive that the case has 9 fans and achieves its level of quietness. But when I buy or build a PC, I'm interested in the total dB, or the dB/performance ratio, and not the dB/fan ratio.
I might not have brought it up if OP hadn't poo-pooed Dell, just assuming that Dells are loud and G5's are quiet. Dells are quiet, too. Sure, unlike Apple, Dell may put cost efficiency before design--but that doesn't mean it's just a bunch of monkeys running around their engineering department.
I was surprised to see that a G5 at 2GHz dissipates less heat than even a non-Xeon P4 at, say, 3GHz. That being the case, I think maybe Apple just really decided to swim upstream with those nine fans, considering many PCs get along fine with far fewer fans, no "air channels", and no perforated case (and achieve the same noise level).
It looks like the "liquid cooling" may be a heat pipe, in which case I take back what I said about it being a necessary evil. Heat pipe equipped heatsinks are good; they don't cost much more than traditional heatsinks and provide much better performance, and they don't leak.
Does your Dell Latitude burn your lap like a G4 PowerBook would?
I'm not anti-Apple, really. But they're pretty much subject to the same engineering constraints as everyone else, albeit with a higher budget (I imagine). Given the size and variety of the PC industry, is it really so hard to believe that Apple has not built a machine that is the fastest AND quietest AND coolest-running desktop in the world?
My parents have one of these and the two fans are inaudible from more than 12 inches away. They happen to have a Western Digital hard drive so you can still hear a slight idle whine and you definitely hear seeks, but I would assume most Dells today would have drives with fluid bearings. In any case, if the hard drive is the noisiest component, you can slap it in a $25 NoiseControl NoVibes frame, problem solved (though unfortunately I don't think you can fit it in a PowerMac).
I read somewhere that when Dell started selling a lot of PCs in Singapore that they began getting serious about quietness; apparently Singaporeans care a lot about office ergonomics. So Dell created an acoustics lab, and the results were so good that they started quieting all(?) their computers.
(I don't know if this applies to their Precision workstations or their top of the line Dimension XPS, as I haven't heard those. But it seems to be the case for all mainstream Dimensions.)
I personally disagree that the Mac is amazingly quiet for having any fans. But after spending considerable effort silencing my own computer, and lurking around the SilentPCReview forums, my scale of what's quiet and what's not has been recalibrated.
A PC case of similar size has maybe four external drive bays and five hard drive bays, and five PCI slots plus an AGP slot.
A PC case with similar expandability, like the Antec Aria, is much smaller. (Though the Aria doesn't have two CPUs and can't hold 8 sticks of memory.)
While the insides of most PC cases may not look neat and tidy, it's child's play to buy or build a quiet, powerful PC that never overheats no matter what the load. So while PC cases may not (in general) have air channels, I'd say they work as well as they need to (though Intel apparently disagrees, since the BTX form factor is coming and is supposed to be designed for airflow).
I don't think you could reasonably expect a mainstream manufacturer to give up the expandability for an academic increase in cooling efficiency, although Shuttle has clearly demonstrated that at least some people are willing to give up expandibility for a slick form factor.
The G5's are beautiful though; I wouldn't mind owning one.
What's the difference? No phase change in a conventional circuit? And if so, how does the liquid circulate with no pump?
pic
I wish they had just said "heat pipe" instead of "innovative liquid cooling", which to us obviously sounds like a traditional liquid cooling setup a la Koolance. For me, that would be a big negative; I would really rather avoid the complexity and risks of liquid cooling.
Most Dell desktops are a fair sight quieter than the G5, from what I've seen (and heard). It's impressive that Apple got a computer that needs nine fans to be fairly quiet, but with most Dells, well, you only need two fans (one in the PSU and one over the CPU, with a plastic hood).
The PowerMacs are impressive for all sorts of other reasons, but saying "amazingly quiet for having nine fans" is progress seems backward to me. Same with the liquid cooling; it's only there because the G5 runs incredibly hot. I'm sure people will be touting liquid cooling as an advantage, when really I think it should be viewed as a necessary evil. Wouldn't it be better to use a processor that doesn't need to be liquid cooled??
It looks like all netTunes does is remote the iTunes UI, pixel for pixel. If so, and if it works with iTunes 4.6 at all, the AirTunes feature should work fine.
That's not what "defensive patenting" means. It means your competitor may come at you with their patent on invention X, so you should be prepared to strike back with your patent on inventions Y and Z. Like a nuclear arms race, the desired outcome is not so much annihilation of your enemy as it is a cautious stalemate (or cross-licensing deals). Not that I think this is a good state of affairs, but "defensive patenting" is not an empty argument.
I want to give the JBoss folks the benefit of the doubt, and I'm sure many others in the Java/J2EE community want to, too... but they just keep making it so damn hard.
All you'd really need to do is put together a UI and an indexing service... both easy to do with .NET and not terribly difficult to do with Java.
Oh, I think some of those products also index the messages stored in your e-mail client. Not sure how you'd go about that... but at least you could leave an API so somebody else who knows could do it.
Anyway, yeah, I understand that. My question is whether, for the specific purpose of spam filtering, it results in improved performance, and if so whether it's been documented anywhere.
The clustering stuff is certainly interesting for other purposes, and I'm glad there are people out there not only writing the software, but integrating it into the OS. The graphic and industrial designers aren't the only smart people at Apple.
Again, what's intuitive doesn't play out in practice... this seems to be a common theme in the world of statistical spam filtering. For example, you'd think the word "free" would be pretty spammy... in my corpus, it only gets a score of .406 (where 0 is least spammy and 1 is most spammy, and an e-mail must have an aggregate score of .9 to be classified as spam). On the other hand, "sir" gets .945 and "madam" gets .987.
According to the FAQ of SpamBayes (I think), they're always getting suggestions of ways to tweak their algos that would "obviously" improve the result, but in almost every case it either makes no difference or hurts accuracy, when actually tested on real data.
There's a big difference between an opinion of that nature coming from an employee of the company in question, vs. the opinion of a more or less objective third party. I for one was definitely duped by some of these fake posts into thinking JBoss was much more popular and much better received than it actually is.
I work for a software startup that has been getting a fair bit of buzz in the media and in the blogsphere. Anytime I join a discussion, the first thing I say is that I work for the company. To omit this information is simply disingenuous, and it's something I believe many of JBoss's most outspoken detractors would never do.
It's true that a lot of prominent people in the Java community and esp. on TSS.com have very deep grudges against JBoss. No doubt there's a lot of bias on both sides of the fence. But the JBoss crew always seems to fight a little dirtier, or in this case, a lot dirtier.