Well, here's the test then. Do those with autism go on to have more children or grandchilden than average?
That's only a good test if your focus is on individuals rather than genes. Even if you don't have offspring, you may promote the further expression of genes you inherit if you contribute to the survival or reproduction of other people with those genes.
Consider a small, somewhat genetically-related tribe. If this tribe produces some individuals who are less fertile but increase survival for their relatives, the genes for those individuals may continue to be expressed in future generations.
Or the geek-friendly version: even though Jim over there is a really crappy farmer, he keeps on solving hard problems about construction and weapons. Let's say he's gotten a bunch of recessives all in one place. Jim's siblings also have half-doses of some of those recessives; if he helps them do better, the recessives are more likely to propagate, even though his siblings are "normal".
I hope THIS mask rev of Opteron (Hammer) chip will be faster than January 2002 PowerPC G4 chips.
[...] The Mac Dual 1 Ghz g4 is faster than all existing dual AMD motherboards in RC5 benchmark by almost 100%.
[..] Funny "Mhz myth" there showing itself I guess... Apple now is selling even FASTER machines [...]
I can see the new "Switch" ad now (white background, jerky cuts):
"I'm a network administrator and so are my friends" "We steal computer power from our employers, at school, wherever we can find it, to run this Are See Five thing"
"Peace, love, and strong crypto"
"So I noticed the Apple computers were pretty fast at kicking out keyblocks" "I had to have one"
"Say it with me: Brute-force known-plaintext attacks" "That's what makes a computer cool"
"If I'm going to spend a few thousand dollars on a computer, it's gotta be the best at at least one thing"
"Hi, I'm Anonymous Coward. I'm a crack user."
[Apple logo]
Cmon. The estimated SPECint numbers are wonderful news. They're a lot closer to reflecting what most of us do with these machines than key-agile stream ciphers. Beating up x86 weenies with the RC5 key rates will just make them buy a couple of $400 Athlons to stick in the closet and gloat about price/key/sec performance. (That's counting electricity too.)
it would have been awarded the Pulitzer Prize had the board not considered it obscene
It is obscene; that judgement is correct.
But it is also divine.
In fact, you can pick an arbitrary pair of opposite, highly charged adjectives, and both will likely be a fair description of Gravity's Rainbow. So as well as being tedious and boring, it is also challenging and endlessly fascinating. Not to mention deadly serious and deadly humorous. I can't think of a novel that has more influenced my worldview than this one.
A-and how can you say no to a book that has lame calculus humor in grafitti, or a bunch of drunk Army engineers chasing the protagonist, singing limericks about Doing It with the German V2 rocket hardware?
There was a young fellow named Hector,
Who was fond of a launcher-erector.
But the squishes and pops
Of acute pressure drops
Wrecked Hector's hydraulic connector
(Hints for the first-time reader of GR: you don't have to understand it the first time through. Hell, you can just skim it. It's still funny and interesting. Also, gin helps a lot.)
FTP chooses arbitrary inbound ports for non-passive data connections as well, and just about every firewall sold today supports stateful packet inspection to open up the TCP (src,port,dest,port) quad for the duration of the transfer.
"Professional" firewalls support the same kind of sniffing for H.323, according to a quick survey I did last year, and it's easy to set up. The biggest difference in deployment is that we already have a high degree of trust/dependence on TCP through firewalls, and much less warm fuzzy feelings about UDP.
Just remember: it doesn't matter whether it's secure; you have to convince other people that it's secure.
Even if Apple ever were to switch to making x86-based Macs (and you, the reader, are significantly more likely to bang Anna Kournikova than to see an x86-based Mac for sale), they would put something proprietary in those machines, maybe even in every component of those machines, and change the Mac OS to refuse to boot if it doesn't detect that proprietary something.
In the world I come from, proprietary hardware gratuitously inserted into a system to control where software can be run has a name.
It's called a dongle.
Apple's in a funny position. What they sell is hardware, but the core value provided to customers is software.
(Don't believe me about the software? Look at how well the Mac clones sold, and how many people here are drooling over the possibility of inexpensive, non-crippled hardware to run OS X...)
Let me second #1. K&R2 is one of the best "my language" books ever written. You need a copy if you don't have it.
I have a special place in my heart for "Smalltalk-80: The Language And Its Implementation.". When I first encountered it in 1989 or so, it changed how I thought about programming. Now that object-oriented programming is part of the ambient, I don't think it would be so startling, but it's interesting to see where it came from.
These days, you can only get "Smalltalk-80: The Language" which cuts out the last bit, which describes how the language was actually implemented on the hardware available as of 1983. I suppose it's not that relevant to see the bytecode used on the Alto when there are systems like Squeak around that are both portable and quite performant on commodity x86 hardware, but I still find the details of how to implement on 1983 hardware interesting.
Re:All languages are NOT equally good
on
Perl 6 Synopsis 5
·
· Score: 2
(In case you're wondering where that PHP link went above, bounce through this link to defeat Doug's silly slashdot Referrer detection.)
Perhaps if you read Doug's testing methodology you'll notice he doesn't configure PHP for performance.
The suggested performance config changes Doug cites are:
--disable-debug. This is the default, so specifying it again doesn't matter.
--enable-inline-optimization "If you have much memory and are using gcc, you might try this."
This does work, although you have to read to the end of the./configure --help options to find it. It speeds up the Shootout tests by around 15%.
Unfortunately, 15-20% isn't very much. PHP's ranking relative to other language implementations doesn't change, even with this improvement.
FWIW, Debian doesn't configure with --enable-inline-optimization.
Tight control of hardware and software are the very things that keep Apple alive and relevant
Great. What's in it for me?
There's this persistent meme that goes something like "Apple has to do this thing we don't like, but what's good for Apple is good for us." No, what's good for Apple is good for Apple. But they're in the business of delivering appealing, competitive products, right? So they don't get any free passes for bad behavior because "they have to eat"---their competition has to eat too.
All languages are NOT equally good
on
Perl 6 Synopsis 5
·
· Score: 2
Perl rocks. PHP rocks. C rocks. Pascal rocks. Bash rocks. One just isn't any better than any other, if the one you are using gets you the desired results (speed, speed of coding, ease of use, correct answers to problems, etc.).
What a wussy response. So what you're saying is that those languages are all good, except when they're not.
I love Intercal because it destroys all the "they're all very nice" language relativity arguments. Here's a language that's specifically designed to be as annoying as possible. I dare you to advocate Intercal in the same way you did above.
Both the PHP language and its implementation have significant problems. Regular users of PHP already have their own list of language design annoyances ("it has to be a global??") and you can see some of the implementation problems here. You will note PHP's implementation getting beaten by Tcl, gawk, xemacs, and njs.:-(
PHP would have been better off if the implementors had used an existing language like Lua (80k of x86 code for standalone interpreter+core libraries!) and focused on the embedding features unique to the application area.
Apple's current high-quality/low-price hardware strategy
Ah, this post must come from the alternate history where Apple's decisive leadership in PReP allows them to innovate and fill new niches, leaving MSI and ABit to worry about the low-quality/low-price volume market for MacOS hardware. Sony's Macs, following the high-quality/high-price strategy, still attract many users on brand name appeal, of course. AppleSoft's tiered OS licensing allows the Mac industry to ship low cost machines while making high margins on licenses to run on high-end hardware. (Taligent's OS has been delayed until 2H03, surprise.)
And about this time, Dell is figuring out that its low-quality/high-price hardware strategy isn't quite working out, and Microsoft is thinking of buying them to get them an efficient supply chain.
Seriously, I think what you mean is "high-quality/high-value". If you mean anything.
Gahhhhh, not the 925. Just about every keyboard of that era had HUGE springs under every letter---it HURT to press them down.
I've got one in storage, and whenever I pull it out I'm dismayed at what I put up with back then.
Re:Why you don't always go to the Supreme Court
on
2600 Drops DeCSS Appeal
·
· Score: 5, Insightful
As Phil Agre pointed out, Bush seems to be headed for a dispute with the Constitution itself. Bush is quoted in the Washington Post (skip to the last paragraph):
Bush said the ruling "points up the fact that we need common-sense judges who understand that our rights were derived from God. And those are the kind of judges I intend to put on the bench."
Which seems in contradiction to Article VI.3 of the US Constitution:
Clause 3:
The Senators and Representatives before mentioned, and the Members of the several State Legislatures, and all executive and judicial Officers, both of the United States and of the several States, shall be bound by Oath or Affirmation, to support this Constitution; but no religious Test shall ever be required as a Qualification to any Office or public Trust under the United States.
Interesting.
As you'd expect from that, the Presidential Oath or Affirmation in Article II.1.8 is "merely"
"I do solemnly swear (or affirm) that I will faithfully execute the Office of President of the United States, and will to the best of my Ability, preserve, protect and defend the Constitution of the United States."
If that price analysis above is correct, you can buy three of these devices at $160 for the price of a single Zaurus.
I still think there's a niche for low-end Linux PDAs. Of course, Softfield is selling their improved 16M, battery-charging Agenda VR3 hardware for ~$130 as well...
The technological means here are just an outcome of policy. Make sure you've got the policy right before you spend a bunch of time playing with routers.
Imagine someone's standing outside a locked car. They've got a slimjim, and are fishing around inside the door.
If it's their car, they can do whatever they like to get past the lock. Hell, they could just brick it and drive off.
If it's somebody else's car, they're breaking the law. That is, if they don't have permission from the owner of the vehicle to do that; I can't use a slimjim so I delegate this to AAA or a locksmith. In fact, if it's somebody else's car, they aren't allowed to open an unlocked cardoor and fish around inside, even though there's no lock in the way.
Doing a bunch of port blocking is like that lock. It can provide some mechanical resistance to what you don't want, but the ultimate protection is the law or policy. When some other IM system springs up that you haven't managed to block yet, you want your users to know that they shouldn't be using that either, even though the car door is unlocked.
Good communication of policies can help a lot. My experience is that I can get much better results when I explain not only the rule, but the motivations behind it, and why it matters to the people who need to follow it. What you really want are users who are on your side, and can help look out for problems. If you can't get that, well, maybe they don't like the rule at all, but they understand why it's there and how it relates to their role in the organization.
Sometimes it helps to write the policy document first. Here's the start of one for a hypothetical usage policy for IM:
Yoyodyne Partners performs work for its clients that is often quite sensitive. We have a duty to protect their information. This is both a matter of ethics and contract; our contract agreements state that we will use appropriate methods to do protect against disclosure or misuse. Failing to protect their information could lead to legal sanction, and a loss of future work for the company.
Instant Messaging, like email, can be a valuable business tool. We have an email usage policy that describes what's appropriate for the use of email. IM, like email, has significant information security risks in some cases. Unencrypted mail may be intercepted on the Internet, as can instant messages. In fact, instant messages may be even more vulnerable to tampering and intercept than email.
The commercial instant messaging tools in common use at Yoyodyne don't have the same kinds of information security protection as our email system does; there's no way to encrypt or sign messages or a conversation, or even verify that you're talking to who you think you're talking to. Therefore:
Yoyodyne Partners prohibits the use of Instant Messaging tools to transmit or discuss any material considered Proprietary or YP Confidential.
This includes using your personal systems, like a home computer.
And at this point your policy-makers have a choice between leaving it at that or adding "...and because the risk of accidental disclosure is high, and to demonstrate to our clients that adequate safeguards are in place, we will block common IM systems at our corporate firewall.". But maybe you don't need to block, if your employees are already good enough to carry out this duty in other forms.
Oops, gotta run. Whaddya expect from a slashdot post anyway?
Yeah. But IA64 made a lot of sense for Intel, given their market position when the effort started.
Think back to Rambus. (Back?) Intel got a lot of options on Rambus stock, provided that Intel could ship n percent of systems using Rambus memory. If Intel had no significant chipset competition, this would be easy. But it turned out there was enough competition to give people a choice of chipsets, and hence memory technologies.
Still, the P4 seems consciously designed to play to Rambus strengths. It chews memory bandwidth like candy through prefetching, which helps cover the higher Rambus latencies. I think Intel took a performance hit relative to AMD when the market preferred DDR SDRAM.
Anyway, it's a great story for Intel if they could control the future of PC technology. Rambus gets rich, Intel gets rich, you pay more. Three cheers for AMD for breaking this.
IA64 now looks similiar. If it wasn't for the aura of inevitability associated with the Itanic, nobody would be particularly thrilled with it. The initial SPECint numbers where it barely kept up with a SPARC were the first practical warning---if you don't count the schedule slips.
If IA64 was inevitable, everybody would have to pay up to transition to it. If it was the banner Win64 platform, a lot of places would be buying them regardless of relative price/performance. But because it looks like AMD will eat IA64 from the low end, and with POWER4 staring down from the high end, there's no longer an obvious niche where IA64 dominance is inevitable.
Good idea. Then I can tell him or her about the Counter-Strike dreams I've been having.
At least the C-S dreams are a lot more interesting and have more narrative depth than the Tetris dreams I had when I was addicted to that game.
The C-S dreams are a little familiar. Back in the 1980s I had my clock radio set to the local National Public Radio "News And Information" station. I usually had the alarm set to "radio", waking me up in the middle of "Morning Edition". Which led to a lot of half-dreams about guerrillas in Central America....
Street Fighter may have had the better gameplay, but Mortal Kombat was so much better at the imagery. To this day, I sometimes see events through the MK filter.
For example, look at the Napster chapter 11 story. What I see is the sky gone dark, a dazed, wobbling Napster logo, and on the screen in big red letters: FINISH HIM.
But back this story.
Microsoft is playing as Scorpion again. Microsoft always plays Scorpion---can't get enough of those harpoon combos. ("GET OVER HERE!" crunch "aieeee"). Stuck over in the corner is Liu Kang, crouched. Linux hasta be Liu Kang---he's the good guy, right?
Scorpion is beating the crap out of Liu Kang, or he would be if Liu wasn't blocking. Blocking most of it at least. At some point Scorpion gets cocky and overextends and you see Liu land...
...the uppercut...
Dan Forden's head slides out....
TOASTY!!!
After Scorpion flies, Liu Kang morphs back into Shang Tsung. Huh, I guess the hit belonged to IBM, that ancient, shape-shifting sorcerer. I bet some flaming skulls are up next. This makes it less clear who is the good guy or the bad guy.
What a great troll. GPRs are a reasonable measure of CPU size; they tend to serve as a proxy for address space as well. But GPRs are the main justification for labeling the 8080 an 8-bit chip, even though you can use HL to address 16 bits of memory.
But instruction size is just silly, and why I think you're trolling. The Athlon in this box uses variable-length instructions; most are 8 bits long. The Alpha, normally considered a 64-bit machine, uses 32-bit instructions. The Itanic puts three instructions into a 128-bit bundle, making its instruction length about 42 bits. The NEC Vr4181 in the Agenda VR3 PDA in my pocket has a 16-bit data bus and modes for both 32- and 64-bit GPRs.
Oh and the Vr4181 has both the standard 32-bit MIPS II instruction set and the 16-bit MIPS16 set, with instructions to switch between them.
From a programmer's point of view the data bus, physical address pins, and the size of instructions are just implementation details. What's important is the instruction set architecture, and the computing model defined by it. In both MIPS II and MIPS16 modes, the Vr4181 has 32-bit GPRs and a flat 32-bit address space. (With a little kernel hacking, it'd be 64-bit GPRs and addressing, but that would be silly.) When I take my code to a Vr4131, which has a 32-bit data bus, I don't have to change anything.
That's why I consider the 68000 to be a 32-bit architecture. Except for performance, my code will run identically on the 68008, 68000, and 68020, with their external 8, 16, and 32 bit data buses.
For the new Transmeta chip, this evaluation strategy says that it's still a 32-bit chip. Programmers outside of Transmeta don't directly program the device, so it makes no difference what the internal ISA is. The externally visible ISA is still the variable-length 8-bit IA-32 architecture, with its 32-bit GPRs. I'd bet they aren't implementing the cheap hacks to get 36-bit physical addressing...
You know, the funny thing about "market-oriented" is that it seems to mean "encourages bigger and better corporations though vertical and horizontal integration". Which actually means fewer economic decisions are being made by markets, and more are made by managers at firms.
Sometimes it seems like "the market" is just the excuse for transfering decision power around (expecially away from your consumers), and reveals a deep mistrust of the use of markets as the basis for broad economic decision-making.
Slashdot Mirror
That's only a good test if your focus is on individuals rather than genes. Even if you don't have offspring, you may promote the further expression of genes you inherit if you contribute to the survival or reproduction of other people with those genes.
Consider a small, somewhat genetically-related tribe. If this tribe produces some individuals who are less fertile but increase survival for their relatives, the genes for those individuals may continue to be expressed in future generations.
Or the geek-friendly version: even though Jim over there is a really crappy farmer, he keeps on solving hard problems about construction and weapons. Let's say he's gotten a bunch of recessives all in one place. Jim's siblings also have half-doses of some of those recessives; if he helps them do better, the recessives are more likely to propagate, even though his siblings are "normal".
[...] The Mac Dual 1 Ghz g4 is faster than all existing dual AMD motherboards in RC5 benchmark by almost 100%.
[..] Funny "Mhz myth" there showing itself I guess... Apple now is selling even FASTER machines [...]
I can see the new "Switch" ad now (white background, jerky cuts):
"I'm a network administrator and so are my friends" "We steal computer power from our employers, at school, wherever we can find it, to run this Are See Five thing"
"Peace, love, and strong crypto"
"So I noticed the Apple computers were pretty fast at kicking out keyblocks" "I had to have one"
"Say it with me: Brute-force known-plaintext attacks" "That's what makes a computer cool"
"If I'm going to spend a few thousand dollars on a computer, it's gotta be the best at at least one thing"
"Hi, I'm Anonymous Coward. I'm a crack user."
[Apple logo]
Cmon. The estimated SPECint numbers are wonderful news. They're a lot closer to reflecting what most of us do with these machines than key-agile stream ciphers. Beating up x86 weenies with the RC5 key rates will just make them buy a couple of $400 Athlons to stick in the closet and gloat about price/key/sec performance. (That's counting electricity too.)
It is obscene; that judgement is correct.
But it is also divine.
In fact, you can pick an arbitrary pair of opposite, highly charged adjectives, and both will likely be a fair description of Gravity's Rainbow. So as well as being tedious and boring, it is also challenging and endlessly fascinating. Not to mention deadly serious and deadly humorous. I can't think of a novel that has more influenced my worldview than this one.
A-and how can you say no to a book that has lame calculus humor in grafitti, or a bunch of drunk Army engineers chasing the protagonist, singing limericks about Doing It with the German V2 rocket hardware?
(Hints for the first-time reader of GR: you don't have to understand it the first time through. Hell, you can just skim it. It's still funny and interesting. Also, gin helps a lot.)FTP chooses arbitrary inbound ports for non-passive data connections as well, and just about every firewall sold today supports stateful packet inspection to open up the TCP (src,port,dest,port) quad for the duration of the transfer.
"Professional" firewalls support the same kind of sniffing for H.323, according to a quick survey I did last year, and it's easy to set up. The biggest difference in deployment is that we already have a high degree of trust/dependence on TCP through firewalls, and much less warm fuzzy feelings about UDP.
Just remember: it doesn't matter whether it's secure; you have to convince other people that it's secure.
Is it just me, or did anybody else see this exact story come out of the random Slashdot story generator?
In the world I come from, proprietary hardware gratuitously inserted into a system to control where software can be run has a name.
It's called a dongle.
Apple's in a funny position. What they sell is hardware, but the core value provided to customers is software.
(Don't believe me about the software? Look at how well the Mac clones sold, and how many people here are drooling over the possibility of inexpensive, non-crippled hardware to run OS X...)
CDPD may be outdated and slow, but is GPRS flat rate?
Judging from the prices paid for the 3G spectrum licenses, it looks like the 3G providers were looking to extract ~$100 month per subscriber....
Income smoothing was a problem for Microsoft too.
Here's a Seattle Times article talking about the settlement.
Let me second #1. K&R2 is one of the best "my language" books ever written. You need a copy if you don't have it.
I have a special place in my heart for "Smalltalk-80: The Language And Its Implementation.". When I first encountered it in 1989 or so, it changed how I thought about programming. Now that object-oriented programming is part of the ambient, I don't think it would be so startling, but it's interesting to see where it came from.
These days, you can only get "Smalltalk-80: The Language" which cuts out the last bit, which describes how the language was actually implemented on the hardware available as of 1983. I suppose it's not that relevant to see the bytecode used on the Alto when there are systems like Squeak around that are both portable and quite performant on commodity x86 hardware, but I still find the details of how to implement on 1983 hardware interesting.
Perhaps if you read Doug's testing methodology you'll notice he doesn't configure PHP for performance.
The suggested performance config changes Doug cites are:
"If you have much memory and are using gcc, you might try this."
This does work, although you have to read to the end of the ./configure --help options to find it. It speeds up the Shootout tests by around 15%.
Unfortunately, 15-20% isn't very much. PHP's ranking relative to other language implementations doesn't change, even with this improvement.
FWIW, Debian doesn't configure with --enable-inline-optimization.
Great. What's in it for me?
There's this persistent meme that goes something like "Apple has to do this thing we don't like, but what's good for Apple is good for us." No, what's good for Apple is good for Apple. But they're in the business of delivering appealing, competitive products, right? So they don't get any free passes for bad behavior because "they have to eat"---their competition has to eat too.
What a wussy response. So what you're saying is that those languages are all good, except when they're not.
I love Intercal because it destroys all the "they're all very nice" language relativity arguments. Here's a language that's specifically designed to be as annoying as possible. I dare you to advocate Intercal in the same way you did above.
Both the PHP language and its implementation have significant problems. Regular users of PHP already have their own list of language design annoyances ("it has to be a global??") and you can see some of the implementation problems here. You will note PHP's implementation getting beaten by Tcl, gawk, xemacs, and njs. :-(
PHP would have been better off if the implementors had used an existing language like Lua (80k of x86 code for standalone interpreter+core libraries!) and focused on the embedding features unique to the application area.
Ah, this post must come from the alternate history where Apple's decisive leadership in PReP allows them to innovate and fill new niches, leaving MSI and ABit to worry about the low-quality/low-price volume market for MacOS hardware. Sony's Macs, following the high-quality/high-price strategy, still attract many users on brand name appeal, of course. AppleSoft's tiered OS licensing allows the Mac industry to ship low cost machines while making high margins on licenses to run on high-end hardware. (Taligent's OS has been delayed until 2H03, surprise.)
And about this time, Dell is figuring out that its low-quality/high-price hardware strategy isn't quite working out, and Microsoft is thinking of buying them to get them an efficient supply chain.
Seriously, I think what you mean is "high-quality/high-value". If you mean anything.
I've got one in storage, and whenever I pull it out I'm dismayed at what I put up with back then.
As you'd expect from that, the Presidential Oath or Affirmation in Article II.1.8 is "merely"
with no "so help me God" or anything like that...Disproof by example: one Patricia Miranda Torvalds is now 5. I wonder how she feels about VM implementations....
See the press release and the PHOTOGRAPHIC PROOF.
(What, you were expecting a personal testimonial? Go browse at -1.)
I still think there's a niche for low-end Linux PDAs. Of course, Softfield is selling their improved 16M, battery-charging Agenda VR3 hardware for ~$130 as well...
Imagine someone's standing outside a locked car. They've got a slimjim, and are fishing around inside the door.
If it's their car, they can do whatever they like to get past the lock. Hell, they could just brick it and drive off.
If it's somebody else's car, they're breaking the law. That is, if they don't have permission from the owner of the vehicle to do that; I can't use a slimjim so I delegate this to AAA or a locksmith. In fact, if it's somebody else's car, they aren't allowed to open an unlocked cardoor and fish around inside, even though there's no lock in the way.
Doing a bunch of port blocking is like that lock. It can provide some mechanical resistance to what you don't want, but the ultimate protection is the law or policy. When some other IM system springs up that you haven't managed to block yet, you want your users to know that they shouldn't be using that either, even though the car door is unlocked.
Good communication of policies can help a lot. My experience is that I can get much better results when I explain not only the rule, but the motivations behind it, and why it matters to the people who need to follow it. What you really want are users who are on your side, and can help look out for problems. If you can't get that, well, maybe they don't like the rule at all, but they understand why it's there and how it relates to their role in the organization.
Sometimes it helps to write the policy document first. Here's the start of one for a hypothetical usage policy for IM:
And at this point your policy-makers have a choice between leaving it at that or adding "...and because the risk of accidental disclosure is high, and to demonstrate to our clients that adequate safeguards are in place, we will block common IM systems at our corporate firewall.". But maybe you don't need to block, if your employees are already good enough to carry out this duty in other forms.
Oops, gotta run. Whaddya expect from a slashdot post anyway?
The parent in its entirety is plagarized from sections 4.5 and 4.11 of the comp.unix.questions FAQ.
If you track down a text version, you'll find those sections were written in 1994....so at least the poster is correct that this is not so new.
AMD design engineers run into an Intel strategy exec at a conference. Intel guy says:
Think back to Rambus. (Back?) Intel got a lot of options on Rambus stock, provided that Intel could ship n percent of systems using Rambus memory. If Intel had no significant chipset competition, this would be easy. But it turned out there was enough competition to give people a choice of chipsets, and hence memory technologies.
Still, the P4 seems consciously designed to play to Rambus strengths. It chews memory bandwidth like candy through prefetching, which helps cover the higher Rambus latencies. I think Intel took a performance hit relative to AMD when the market preferred DDR SDRAM.
Anyway, it's a great story for Intel if they could control the future of PC technology. Rambus gets rich, Intel gets rich, you pay more. Three cheers for AMD for breaking this.
IA64 now looks similiar. If it wasn't for the aura of inevitability associated with the Itanic, nobody would be particularly thrilled with it. The initial SPECint numbers where it barely kept up with a SPARC were the first practical warning---if you don't count the schedule slips.
If IA64 was inevitable, everybody would have to pay up to transition to it. If it was the banner Win64 platform, a lot of places would be buying them regardless of relative price/performance. But because it looks like AMD will eat IA64 from the low end, and with POWER4 staring down from the high end, there's no longer an obvious niche where IA64 dominance is inevitable.
Four cheers for AMD.
At least the C-S dreams are a lot more interesting and have more narrative depth than the Tetris dreams I had when I was addicted to that game.
The C-S dreams are a little familiar. Back in the 1980s I had my clock radio set to the local National Public Radio "News And Information" station. I usually had the alarm set to "radio", waking me up in the middle of "Morning Edition". Which led to a lot of half-dreams about guerrillas in Central America....
For example, look at the Napster chapter 11 story. What I see is the sky gone dark, a dazed, wobbling Napster logo, and on the screen in big red letters: FINISH HIM.
But back this story.
Microsoft is playing as Scorpion again. Microsoft always plays Scorpion---can't get enough of those harpoon combos. ("GET OVER HERE!" crunch "aieeee"). Stuck over in the corner is Liu Kang, crouched. Linux hasta be Liu Kang---he's the good guy, right?
Scorpion is beating the crap out of Liu Kang, or he would be if Liu wasn't blocking. Blocking most of it at least. At some point Scorpion gets cocky and overextends and you see Liu land...
After Scorpion flies, Liu Kang morphs back into Shang Tsung. Huh, I guess the hit belonged to IBM, that ancient, shape-shifting sorcerer. I bet some flaming skulls are up next. This makes it less clear who is the good guy or the bad guy.What a great troll. GPRs are a reasonable measure of CPU size; they tend to serve as a proxy for address space as well. But GPRs are the main justification for labeling the 8080 an 8-bit chip, even though you can use HL to address 16 bits of memory.
But instruction size is just silly, and why I think you're trolling. The Athlon in this box uses variable-length instructions; most are 8 bits long. The Alpha, normally considered a 64-bit machine, uses 32-bit instructions. The Itanic puts three instructions into a 128-bit bundle, making its instruction length about 42 bits. The NEC Vr4181 in the Agenda VR3 PDA in my pocket has a 16-bit data bus and modes for both 32- and 64-bit GPRs.
Oh and the Vr4181 has both the standard 32-bit MIPS II instruction set and the 16-bit MIPS16 set, with instructions to switch between them.
From a programmer's point of view the data bus, physical address pins, and the size of instructions are just implementation details. What's important is the instruction set architecture, and the computing model defined by it. In both MIPS II and MIPS16 modes, the Vr4181 has 32-bit GPRs and a flat 32-bit address space. (With a little kernel hacking, it'd be 64-bit GPRs and addressing, but that would be silly.) When I take my code to a Vr4131, which has a 32-bit data bus, I don't have to change anything.
That's why I consider the 68000 to be a 32-bit architecture. Except for performance, my code will run identically on the 68008, 68000, and 68020, with their external 8, 16, and 32 bit data buses.
For the new Transmeta chip, this evaluation strategy says that it's still a 32-bit chip. Programmers outside of Transmeta don't directly program the device, so it makes no difference what the internal ISA is. The externally visible ISA is still the variable-length 8-bit IA-32 architecture, with its 32-bit GPRs. I'd bet they aren't implementing the cheap hacks to get 36-bit physical addressing...
Sometimes it seems like "the market" is just the excuse for transfering decision power around (expecially away from your consumers), and reveals a deep mistrust of the use of markets as the basis for broad economic decision-making.