and ofcourse agner.org
this guy rocks big time for pentium optimizations
Arkon
[QSoft Systems]
C/C++ Optimization
Blit,
My point exactly. It wasn''t that I didn''t think you understood it, but rather that I thought you glossed over it. It is one of those things people tend to take for granted once they have been doing it for awhile, but a common mistake starting out. I have worked in performance tuning for a long time professionally. Generally many people understand the general idea of algorithmic analysis, but they miss one little detail. All operations are not equal. With CPU it isn''t nearly as dramtic as it is with I/O. Executing a binary search against a huge file can be murder as you thrash the heads all over the disk drive.
My point exactly. It wasn''t that I didn''t think you understood it, but rather that I thought you glossed over it. It is one of those things people tend to take for granted once they have been doing it for awhile, but a common mistake starting out. I have worked in performance tuning for a long time professionally. Generally many people understand the general idea of algorithmic analysis, but they miss one little detail. All operations are not equal. With CPU it isn''t nearly as dramtic as it is with I/O. Executing a binary search against a huge file can be murder as you thrash the heads all over the disk drive.
Agreed, but thrashing the drive is going to be even worse on a linear search for large data sets. DB records are often stored in B+ trees for a reason...
All I''m saying is people (and especially new programmers) need to take note of asymptotic running time as well, instead of just focusing on other optimization techniques. Plenty of people get frustrated when their code which they''ve optimized heavily still performs badly, because they''re doing something that''s inherently slow algorithm-wise. Not to mention that code that''s been hacked to take all sorts of shortcuts that aren''t entirely obvious gets excessively cryptic, making it impossible to maintain.
All I''m saying is people (and especially new programmers) need to take note of asymptotic running time as well, instead of just focusing on other optimization techniques. Plenty of people get frustrated when their code which they''ve optimized heavily still performs badly, because they''re doing something that''s inherently slow algorithm-wise. Not to mention that code that''s been hacked to take all sorts of shortcuts that aren''t entirely obvious gets excessively cryptic, making it impossible to maintain.
This topic is closed to new replies.
Advertisement
Popular Topics
Advertisement