Apple has serious R&D demands - if they stayed with IBM, they would need a cooler G5 so they could finally put them in PowerBooks. From IBM's perspective, that's probably not a good investment. It would cost IBM a lot of money to develop a cooler G5, and they might not sell enough to Apple to justify that cost. Particularly when they have a sure thing in GameCubes and the next generation consoles.
I can see IBM telling Apple, "It's not you, it's me."
"Hi. Here is an essay based entirely on personal anecdotes and conjecture that validates my own personal view, and also places me in an elite category of computer users. Have a nice day."
Well, sorta. It's just that in college, there are so few people, ratio wise, who are on sports teams that most people care about. But trust me, those few who are are regarded as gods. I went to Virginia Tech, and a certain quarterback who now plays in the NFL approached someone I knew and just said "Want to hit this?" She didn't because she didn't know who he was yet, but the girl next to her jumped at the chance.
I know I'll be marked as a troll as it goes against the popular perception on/.
Sorry to pick on you, but I can't stand it when people say "I know I'll be modded as a troll/flamebait" or "I have karma to burn, so..." The poster is basically trying to martyr themself. If you have an opinion which you think runs contradictory to what most other people think, just post it as is. I've seen far too many posts with these disclaimers marked up to 4 and 5 for it to actually be true.
I don't think that extrapolating from programming contest results to a nation's programmers' general ability to code is valid. Matloff points out excellent reasons why this doesn't work, but he pays attention mainly to statistics of the rankings and varying amount of training time.
Simply, I don't think that being good at these contests necessarily is the same at being good at producing software in industry or even research. I don't like solving problems under strict time constraints, so I've never volunteered to take part in math or programming competitions. It's simply not fun for me. I like problem solving when I'm free to take the time to explore the design space and maybe go off on tangents that might eventually prove worthwhile (but often don't). Some people enjoy solving problems under strict time constraints; I'm just not one of them. I enjoy other activities that others do not. It's just personal preference.
In the end, we always have time constraints - projects have deadlines, research papers have submission dates - but measuring the amount of time in hours vs. days, weeks or months make a very big difference in how much freedom you have to explore the problem.
I also reccomend "The Selfish Gene" by Dawkins. In it, he goes into detail about the mechanism by which evolution happens, our genes. It's a fascinating book.
Auto-vectorization is not the reason why Intel's compiler is better. It certainly helps, but in my experience, not much. Intel's compiler just does better optimizations across the board. Which is no surprise: Intel is making a compiler for thier chipsets. They have inside knowledge of what's best to do when for a particular chip. Further, Intel's compiler is marketed as the fastest compiler for x86, which as far as I know, is true. Hence, they spend a lot of time on the optimizations.
GCC, on the other hand, has a different goal: get a working compiler on as many platforms as possible.
It is an architectural change - one big enough to warrant increasing from 3 to 4. But the author's question is "Are the optimizations better now?" to which he answers "No." But he also make it quite clear that in the future, they may be.
One of the major changes in the 4.0.0 release is the internal reorganization that allows for more aggressive optimizations. Hence, he tested how the optimized performance of the latest 3.x versus 4.0.0. How do you tell the compiler to optimize? Well, you have to pass it "lots of flags."
He's a "maverick" in the sense that aside from the original Star Wars, he made very successful and expensive movies outside of the studio system. Empire and on are essentially independent films.
If it was sold in a box in stores, people who don't know what it is might pick it up and try to install it. Very few people have 64-bit processors - certainly the average consumer who buys software from a retailer does not. This way they can control it: you only get the 64-bit version of Windows if you actually have a 64-bit processor.
The poster implied their reason was lack of support. I think it's lack of interest.
Read some Linux kernel code, particularly the system calls. Most of them are structured in such a way that they do lots and lots of error checking, and do a goto to near the end of the function that performs cleanup (mainly, deallocation of structures) if any of the error checks fail. It's really the most elegant and obvious way to accomplish the task.
You're right; I shouldn't have said "floating point operations". It's any operation done over a large array. As for how effective they are, I've only seen marginal gains with the SSE instructions on floating point intensive benchmarks with large data sets. That makes me rather skeptical about how effective they'd be in applications which aren't as compute bound.
Vectorized instructions really only make a difference when you're doing floating point operations on very large arrays. I don't think an Office productivity application does much of that.
True, but I agree with the author that it's not flamebait. Adversarial, yes, but I don't consider that suffecient for a flamebait. His post makes a legitamite point, whereas I consider something flamebait if it exists only to be aversarial.
Not so sure about Google and Fark which are purely online, but it seems logical that traditional newspapers will pass on the cost to their print subscribers.
It does seem logical, until you realize that most of a newspaper's revenue comes from advertising, not subscriptions. This extra cost will likely have much more effect on the cost of their advertisements than the cost of their subscriptions.
Precisely. How dare people try to indicate when information will no longer be correct and the cache should be refreshed.
That is done through - wait for it - the TTL. Setting a low TTL means your data is more likely to be accurate, but increases communication costs. Setting a high TTL means your data is less likely to be accurate, but decreases communication costs. Obviously there is a happy medium somewhere, depending on the application.
Nothing in my post said we should cache information indefinitely; that's not a cache, that's storage. But eliminating it entirely would bring top level domain servers to their knees.
Slashdot Mirror
Yours is basically the conclusion I came to.
Apple has serious R&D demands - if they stayed with IBM, they would need a cooler G5 so they could finally put them in PowerBooks. From IBM's perspective, that's probably not a good investment. It would cost IBM a lot of money to develop a cooler G5, and they might not sell enough to Apple to justify that cost. Particularly when they have a sure thing in GameCubes and the next generation consoles.
I can see IBM telling Apple, "It's not you, it's me."
"Hi. Here is an essay based entirely on personal anecdotes and conjecture that validates my own personal view, and also places me in an elite category of computer users. Have a nice day."
The word "official" is not in the story. So, no, it's not "official."
Tough talk. But it wasn't your quality of life on the line.
Well, sorta. It's just that in college, there are so few people, ratio wise, who are on sports teams that most people care about. But trust me, those few who are are regarded as gods. I went to Virginia Tech, and a certain quarterback who now plays in the NFL approached someone I knew and just said "Want to hit this?" She didn't because she didn't know who he was yet, but the girl next to her jumped at the chance.
I know I'll be marked as a troll as it goes against the popular perception on /.
Sorry to pick on you, but I can't stand it when people say "I know I'll be modded as a troll/flamebait" or "I have karma to burn, so..." The poster is basically trying to martyr themself. If you have an opinion which you think runs contradictory to what most other people think, just post it as is. I've seen far too many posts with these disclaimers marked up to 4 and 5 for it to actually be true.
I don't think that extrapolating from programming contest results to a nation's programmers' general ability to code is valid. Matloff points out excellent reasons why this doesn't work, but he pays attention mainly to statistics of the rankings and varying amount of training time.
Simply, I don't think that being good at these contests necessarily is the same at being good at producing software in industry or even research. I don't like solving problems under strict time constraints, so I've never volunteered to take part in math or programming competitions. It's simply not fun for me. I like problem solving when I'm free to take the time to explore the design space and maybe go off on tangents that might eventually prove worthwhile (but often don't). Some people enjoy solving problems under strict time constraints; I'm just not one of them. I enjoy other activities that others do not. It's just personal preference.
In the end, we always have time constraints - projects have deadlines, research papers have submission dates - but measuring the amount of time in hours vs. days, weeks or months make a very big difference in how much freedom you have to explore the problem.
I also reccomend "The Selfish Gene" by Dawkins. In it, he goes into detail about the mechanism by which evolution happens, our genes. It's a fascinating book.
Offtopic? The entire post is two sentences; one to adress the parent post, and the other directly references the article. How can that be offtopic?
It's a typo. Get over it. It's italicized because were I speaking that sentence outloud, I would have stressed that word.
Auto-vectorization is not the reason why Intel's compiler is better. It certainly helps, but in my experience, not much. Intel's compiler just does better optimizations across the board. Which is no surprise: Intel is making a compiler for thier chipsets. They have inside knowledge of what's best to do when for a particular chip. Further, Intel's compiler is marketed as the fastest compiler for x86, which as far as I know, is true. Hence, they spend a lot of time on the optimizations.
GCC, on the other hand, has a different goal: get a working compiler on as many platforms as possible.
It is an architectural change - one big enough to warrant increasing from 3 to 4. But the author's question is "Are the optimizations better now?" to which he answers "No." But he also make it quite clear that in the future, they may be.
One of the major changes in the 4.0.0 release is the internal reorganization that allows for more aggressive optimizations. Hence, he tested how the optimized performance of the latest 3.x versus 4.0.0. How do you tell the compiler to optimize? Well, you have to pass it "lots of flags."
I think it's worth abbreviating names that are used often. Think ls versus list or even listdirectory.
He's a "maverick" in the sense that aside from the original Star Wars, he made very successful and expensive movies outside of the studio system. Empire and on are essentially independent films.
If it was sold in a box in stores, people who don't know what it is might pick it up and try to install it. Very few people have 64-bit processors - certainly the average consumer who buys software from a retailer does not. This way they can control it: you only get the 64-bit version of Windows if you actually have a 64-bit processor.
The poster implied their reason was lack of support. I think it's lack of interest.
Read some Linux kernel code, particularly the system calls. Most of them are structured in such a way that they do lots and lots of error checking, and do a goto to near the end of the function that performs cleanup (mainly, deallocation of structures) if any of the error checks fail. It's really the most elegant and obvious way to accomplish the task.
You're right; I shouldn't have said "floating point operations". It's any operation done over a large array. As for how effective they are, I've only seen marginal gains with the SSE instructions on floating point intensive benchmarks with large data sets. That makes me rather skeptical about how effective they'd be in applications which aren't as compute bound.
Not really. By "very large" I mean the arrays are at least several MB. Most spreadsheets don't have that much data.
Vectorized instructions really only make a difference when you're doing floating point operations on very large arrays. I don't think an Office productivity application does much of that.
True, but I agree with the author that it's not flamebait. Adversarial, yes, but I don't consider that suffecient for a flamebait. His post makes a legitamite point, whereas I consider something flamebait if it exists only to be aversarial.
Not so sure about Google and Fark which are purely online, but it seems logical that traditional newspapers will pass on the cost to their print subscribers.
It does seem logical, until you realize that most of a newspaper's revenue comes from advertising, not subscriptions. This extra cost will likely have much more effect on the cost of their advertisements than the cost of their subscriptions.
Precisely. How dare people try to indicate when information will no longer be correct and the cache should be refreshed.
That is done through - wait for it - the TTL. Setting a low TTL means your data is more likely to be accurate, but increases communication costs. Setting a high TTL means your data is less likely to be accurate, but decreases communication costs. Obviously there is a happy medium somewhere, depending on the application.
Nothing in my post said we should cache information indefinitely; that's not a cache, that's storage. But eliminating it entirely would bring top level domain servers to their knees.
(And money is cheap!)
Almost sounds like it should be a Yogi Berra-ism.
I get a flamebait for a response to a troll? Nice.