|
|
|
Back to newsletter 125 contents
The first edition of my book "Java Performance Tuning" was published over 10 years ago, and included over 300 performance tuning techniques. The majority of those techniques are still valid, though I have found that the low level ones are almost never required by me any more, as the Java runtime has become so much more efficient in the intervening years.
Of all the techniques that I published, the single technique that has generated the largest number of responses (that reached me) has been converting doubles to strings. Partly that's because it's one of the few techniques that I extracted from my book and published separately as an article; partly it's because many financial applications need to repeatedly display a large number of floating point numbers that are frequently changing, and it turns out that repeatedly converting those numbers to strings hundreds of times a second has always been a bottleneck.
When I published that conversion technique, I had expected that within a few years it wouldn't be needed, both because I expected the Java runtime would be so much faster, and because I expected the core Java classes to provide a fast conversion alternative to the slower one it already had. And yet even now I still get emails from people using, and requiring, my faster conversion. So I have a question for you all: do you still find a need for low level speed optimizations like this one? Specifically is the Java double formatting still noticeably slow enough for you that you need to use my (or another) faster conversion procedure, and if so, why?
Now on with this month's newsletter and all our usual Java performance tools, news, and article links. Over at fasterj we have a new cartoon about using priority queues; Javva The Hutt is back to tell us all about his recent training course and, as usual, we have extracted tips from all of this month's referenced articles.
Java performance tuning related news.
Java performance tuning related tools.
Back to newsletter 125 contents