By fifthposition
via pl.atyp.us
Published: Jul 13 2010 / 09:34
So, you have a program that uses efficient algorithms with a well-parallelized implementation, and it’s neither I/O-bound nor memory-bound. Will it be faster in C? Yes, it very well might. It might also be faster in Fortran, which is why many continue to use it for scientific computation but that hardly makes it a good choice for more general use. Everyone thinks they’re writing the most performance-critical code in the world, but in reality maybe one in twenty programmers are writing code where anything short of the most egregious bloat and carelessness will affect the performance of the system overall.
-
Continuous Delivery and Long-Term Requirements
-
SQL Server Compact Code Snippet of the Week #15 : flush data to disk immediately
-
ABC: Always Be Coding
-
Quotes of the Week: Complexity
-
Setting up Static and Dynamic (LACP) Link Aggregation (802.3ad) on D-LINK DES-3526/DES-3550 using the CLI.
-
The Kirk and Spock Approaches to Big Data
Comments
dmandpenfold replied ago:
I would say it would be faster in C. When you think about it.
rfssdm replied ago:
I used to work for a company that did everything in C "for performance reasons". The code base was so bloated that they could probably have increased performance significantly by simplifying their architecture, so they could tell the wood from the trees.
I remember one bug fix I did where I found a redundant computation that must have been in there for years. After removal, the underlying product feature was three times as fast :-)
Voters For This Link (45)
Voters Against This Link (0)