Don’t bother!
There are cheaper ways to improve software performance 😉
CS research is mostly another form of “mathematical disease”.
Oh! Happy New Year nevertheless…
I just love these arrogant people who think that CS education and academia are useless. Of course, only they — real warriors — save the world and write the code. Unfortunately, quite often their code contains such horrible jokes as bubble sort. Which, in turn, other folks have to debug and to fix.
Kevembuanggasays:
Unfortunately, quite often their code contains such horrible jokes as bubble sort. Which, in turn, other folks have to debug and to fix.
Debugging bubble sort?
What are the chances that a bug show up in bubble sort relative to any other sort?
It’s ugly and horrendously inefficient but if you have a dozen items to sort it could be that it IS the right solution.
You didn’t really grok the points in the linked comment, did you?
Itmansays:
Chances are huge! I had to fix this once myself. The premature optimization problem arises usually, when there are two comparable algorithms and you chose a more complicated. Then, of course, it is a waste of time. But to chose the right algorithm, you should first understand the problem. That requires some CS knowledge. Moreover, Moore’s law would not hold forever. Sooner or later you have to optimize to get this 10 or 50 percent increase. In some applications such as large-scale applications (e.g. web search), you have to do this already.
More exactly, by “lightweight” you mean “compute-cost-effective” compression. Can compression save CPU cycles? Maybe. Sometimes. Would be good to explore and map that dimension somewhat.
Don’t bother!
There are cheaper ways to improve software performance 😉
CS research is mostly another form of “mathematical disease”.
Oh! Happy New Year nevertheless…
@Kevembuangga
Good points.
You may enjoy this blog post by Maarten van Emden on “What is Computer Science?”
http://vanemden.wordpress.com/2009/12/27/what-is-computer-science/
I just love these arrogant people who think that CS education and academia are useless. Of course, only they — real warriors — save the world and write the code. Unfortunately, quite often their code contains such horrible jokes as bubble sort. Which, in turn, other folks have to debug and to fix.
Unfortunately, quite often their code contains such horrible jokes as bubble sort. Which, in turn, other folks have to debug and to fix.
Debugging bubble sort?
What are the chances that a bug show up in bubble sort relative to any other sort?
It’s ugly and horrendously inefficient but if you have a dozen items to sort it could be that it IS the right solution.
You didn’t really grok the points in the linked comment, did you?
Chances are huge! I had to fix this once myself. The premature optimization problem arises usually, when there are two comparable algorithms and you chose a more complicated. Then, of course, it is a waste of time. But to chose the right algorithm, you should first understand the problem. That requires some CS knowledge. Moreover, Moore’s law would not hold forever. Sooner or later you have to optimize to get this 10 or 50 percent increase. In some applications such as large-scale applications (e.g. web search), you have to do this already.
More exactly, by “lightweight” you mean “compute-cost-effective” compression. Can compression save CPU cycles? Maybe. Sometimes. Would be good to explore and map that dimension somewhat.
@Bannister
Yes, you are quite right. By lightweight, I mean that it reduces the computational cost.
I had to fix this once myself.
OK, fix that C one:
for (i = size – 1; i;) {
for (j = i; ++j < size && a[j-1] > a[j]; swap(a[j-1], a[j]));
while (–i && a{i] <= a[i+1]);
}
🙂