- why start with the NYtimes? Why not try some cheaper media first (like local newspapers), to see how the responses are.. and if it works, to scale up to more expensive papers like the NY times.
That's a good idea, but I think the answer is pretty simple: it takes too much time and coordination. Which local papers? Where? How do we determine what's effective to move onto larger papers? All of that work might actually be more expensive than one ad in a very prominent paper.
The whole reason people are interested in the Prisoner's Dilemma (and Iterated) is that they are models for situations encountered in science. This isn't a comepetition like chess where we're trying to see who (as in the humans) are the best at it. We're trying to see what interesting results come of it. This is an interesting result.
So as to not keep you guessing, I suspect your private school was both better funded and had a larger percentage of middle to high income kids. You said your school took all economic backgrounds - but what was the breakdown, and how "low" on the economic ladder? At my public high school, about 66% of the students were on free or reduced lunch. (I was on my newspaper and I crunched the numbers one time.) Our school was also not, in general, a bastion of academic achievemnt.
For standardized tests like the SATs, there's a strong correlation between performance and household income.
Oh, and if someone at my school had physically attacked our principal in the cafeteria, the administrators would have restrained the kid very quickly, and the kid would have been expelled. It wouldn't have been instantaneous like in your school, but it certainly would have happened.
("Free or reduced lunch" means the kid's guardians have an income low enough that the state is willing to sell lunch to him/her for a reduced price or free.)
Please forgive me, I could only make it about half way through your first paragraph. Has it ever occurred to you that it's attitudes like yours that are a barrier to women being in CS?
When you get into systems research, the distinction is not as well defined. To be interesting, the research requires an implementation, and that implementation is often the most time consuming and interesting part of the research.
Why don't we change the name of computer science to something more appropriate. Algorithmics? Computational theory?
Neither of those work because they're too small. Both algorithm analysis and theory of computation are subsets within the overall field of Computer Science. "Symbolic processing" wouldn't capture everything, either. Just as people always point out that there's more to CS than programming, there's more to CS than algorithms, too.
I also think that Computer Science is an odd term for the field. I've never been able to draw the line between what's science and what's engineering when it comes to "Computer Science" and I don't think I ever will be able to.
Ever consider going back to get a Master's in CS? From what you say, I think it's something you'd find satisfying. I know it's difficult to going back to being a poor student, but it might be worth it.
If I remember correctly, I then need to go and re-download and configure the extensions I'm using. That's a pain in the ass. I think Firefox is a great browser, but this is something the developers need to fix - and I have confidence that they eventually will.
You seem to feel that if this is the case, then it's some sort of dire problem that hurts us all. It's not. This is a site that says "We found these stories interesting." It's maintained by several people, so there are going to be inconsistencies, and yeah, some stories will be missed. It's not that big of a deal.
And think about if it was not the case - then having multiple editors gains you nothing. The workload for ten editors would be the same as for one; they'd still have to look at the same number of stories. I know I wouldn't want to work like that.
No, we don't have a problem. Editors select stories that they think should go up. That's their right. Some editors will have different preferences. There is nothing wrong or unexpected about that. This is their site, not yours.
The desktop is a 2D environment. There is no need for a 3D mouse. If a 3D desktop environment actually catches on, then people will need a 3D mouse. Until then, it's a novelty.
I've just started using Firefox, and the best plugin I know of for it is Super DragAndGo. If you drag a link to empty space on the webpage, that link is opened in a new tab. It's so simple, but it's the best new web browsing feature I've seen in a long time.
I guess all of those operating systems papers I read last year that involved implementation weren't computer science. And I guess the paper I have in front of me, "The Performance Implications of Thread Mangagement Alternatives for Shared-Memory Multiprocessors" isn't computer science since they actually coded their ideas.
What he meant was they're not releasing all of it, just some of it.
"There have been a lot of conversations in the company in the past two months about (how) . . . it's time for us to give something back. So our technical director, Craig Silverstein, has started a project to look at all the Google code and start figuring out what parts of it we want to give back," Rosing said.
I agree with you. I was raised with VS as an undergrad. I'm at a different school for grad school, and we're all Linux, all the time. I thought I'd be at a loss when it came to coding, I've found I'm really not. My enviromnent is vi and make - which makes life easier, because these days, I'm writing code for SMPs, and my desktop is not one. I just ssh in and go about as I always do.
So yes, I agree, VS is awesome, but I actually find I don't miss it. Now, if I was on a very large project that had dozens of people and 100,000+ lines of code, then maybe I would miss VS, but for my single person research projects, I have no need.
In these cases, no inlining is taking place, or it makes no difference.
The metchall.cpp is one I was wondering about. Now that I think about it, it makes sense that no inlining is happening; you're calling a virtual member function through a pointer. You have to go through the virtual function table at run time, the compiler can't know ahead of time what to inline.
Inlining isn't going to help fib.cpp - you can't inline a recursive function. For the matrix.cpp case, the function call overhead is nothing compared to the aglorithm it's running. Something you can do with C++ that you can't do with Java is hand optimize the matrix multiplication to avoid cache misses. That can speed it up by orders of magnitude. To my knowledge, you can't do that in Java. The (small) speed up you saw when you told it what processor you're using might have to do with keeping stuff in the L1 cache, or even on the registers.
I should have RTFA. My fault. His reasoning was "On the other hand, -O3 performs space-speed tradeoffs," however, the man page for g++ that I've always read says:
-O3 Optimize yet more. -O3 turns on all optimizations specified by -O2 and also turns on the -finline-functions and -frename-registers options.
From what I know, the STL performs better if you inline - it basically gets rid of a lof the wrapper classes. And on the method call comparison - the one where Java was faster by an order of magnitude - g++'s optimizer might be smart enough to inline that call.
Slashdot Mirror
- why start with the NYtimes? Why not try some cheaper media first (like local newspapers), to see how the responses are.. and if it works, to scale up to more expensive papers like the NY times.
That's a good idea, but I think the answer is pretty simple: it takes too much time and coordination. Which local papers? Where? How do we determine what's effective to move onto larger papers? All of that work might actually be more expensive than one ad in a very prominent paper.
Invalid input should still be handled sanely. Think of a compiler.
The whole reason people are interested in the Prisoner's Dilemma (and Iterated) is that they are models for situations encountered in science. This isn't a comepetition like chess where we're trying to see who (as in the humans) are the best at it. We're trying to see what interesting results come of it. This is an interesting result.
So as to not keep you guessing, I suspect your private school was both better funded and had a larger percentage of middle to high income kids. You said your school took all economic backgrounds - but what was the breakdown, and how "low" on the economic ladder? At my public high school, about 66% of the students were on free or reduced lunch. (I was on my newspaper and I crunched the numbers one time.) Our school was also not, in general, a bastion of academic achievemnt.
For standardized tests like the SATs, there's a strong correlation between performance and household income.
Oh, and if someone at my school had physically attacked our principal in the cafeteria, the administrators would have restrained the kid very quickly, and the kid would have been expelled. It wouldn't have been instantaneous like in your school, but it certainly would have happened.
("Free or reduced lunch" means the kid's guardians have an income low enough that the state is willing to sell lunch to him/her for a reduced price or free.)
The five minutes I saw of Batman and Robin is the worst movie I've ever seen.
Please forgive me, I could only make it about half way through your first paragraph. Has it ever occurred to you that it's attitudes like yours that are a barrier to women being in CS?
Why does this affect girls more? Because society doesn't encourage girls to be social outcasts.
CS majors have to be social outcasts?
When you get into systems research, the distinction is not as well defined. To be interesting, the research requires an implementation, and that implementation is often the most time consuming and interesting part of the research.
Why don't we change the name of computer science to something more appropriate. Algorithmics? Computational theory?
Neither of those work because they're too small. Both algorithm analysis and theory of computation are subsets within the overall field of Computer Science. "Symbolic processing" wouldn't capture everything, either. Just as people always point out that there's more to CS than programming, there's more to CS than algorithms, too.
I also think that Computer Science is an odd term for the field. I've never been able to draw the line between what's science and what's engineering when it comes to "Computer Science" and I don't think I ever will be able to.
Ever consider going back to get a Master's in CS? From what you say, I think it's something you'd find satisfying. I know it's difficult to going back to being a poor student, but it might be worth it.
If I remember correctly, I then need to go and re-download and configure the extensions I'm using. That's a pain in the ass. I think Firefox is a great browser, but this is something the developers need to fix - and I have confidence that they eventually will.
I'd like to know too, but you and I don't have the right to "demand" anything. At least I don't - I don't pay to read this site.
I can generally tell how hard I'm exercising by how much it hurts.
You seem to feel that if this is the case, then it's some sort of dire problem that hurts us all. It's not. This is a site that says "We found these stories interesting." It's maintained by several people, so there are going to be inconsistencies, and yeah, some stories will be missed. It's not that big of a deal.
And think about if it was not the case - then having multiple editors gains you nothing. The workload for ten editors would be the same as for one; they'd still have to look at the same number of stories. I know I wouldn't want to work like that.
Eventually those individual pieces need to syncrhonize again, and sometimes that syncrhonization is more expensive than the computation itself.
No, we don't have a problem. Editors select stories that they think should go up. That's their right. Some editors will have different preferences. There is nothing wrong or unexpected about that. This is their site, not yours.
The desktop is a 2D environment. There is no need for a 3D mouse. If a 3D desktop environment actually catches on, then people will need a 3D mouse. Until then, it's a novelty.
I've just started using Firefox, and the best plugin I know of for it is Super DragAndGo. If you drag a link to empty space on the webpage, that link is opened in a new tab. It's so simple, but it's the best new web browsing feature I've seen in a long time.
I think Gettysburg had an intermission.
I guess all of those operating systems papers I read last year that involved implementation weren't computer science. And I guess the paper I have in front of me, "The Performance Implications of Thread Mangagement Alternatives for Shared-Memory Multiprocessors" isn't computer science since they actually coded their ideas.
I agree with you. I was raised with VS as an undergrad. I'm at a different school for grad school, and we're all Linux, all the time. I thought I'd be at a loss when it came to coding, I've found I'm really not. My enviromnent is vi and make - which makes life easier, because these days, I'm writing code for SMPs, and my desktop is not one. I just ssh in and go about as I always do.
So yes, I agree, VS is awesome, but I actually find I don't miss it. Now, if I was on a very large project that had dozens of people and 100,000+ lines of code, then maybe I would miss VS, but for my single person research projects, I have no need.
In these cases, no inlining is taking place, or it makes no difference.
The metchall.cpp is one I was wondering about. Now that I think about it, it makes sense that no inlining is happening; you're calling a virtual member function through a pointer. You have to go through the virtual function table at run time, the compiler can't know ahead of time what to inline.
Inlining isn't going to help fib.cpp - you can't inline a recursive function. For the matrix.cpp case, the function call overhead is nothing compared to the aglorithm it's running. Something you can do with C++ that you can't do with Java is hand optimize the matrix multiplication to avoid cache misses. That can speed it up by orders of magnitude. To my knowledge, you can't do that in Java. The (small) speed up you saw when you told it what processor you're using might have to do with keeping stuff in the L1 cache, or even on the registers.
Why stop at -O2 and not use -O3 for g++?