LC for Raspberry Pi

[Richard Gaskin]

jonathandlynch wrote:

 > Why is LC 7 slower than 6?

There may be additional factors specific to the RPi build, but in 
general those differences were due to Unicode support and unoptimized 
refactoring related to that and other large-scale scopes of work.

With v9 much of the earlier performance drop has been regained, with key 
work on done common things like lineoffset and some other chunk 
expressions, arrays, and more.

Some rendering operations may still benefit from optimization, esp. on 
Windows, but overall I've not seen much in that area on Linux.

Given the larger data sizes of some Unicode strings you can still expect 
some performance loss in some areas, but in my experience nothing like 
moving from v6 to v7.

If you come across anything prohibitive please submit a bug report.  The 
team has worked through a number of performance-related bug reports, and 
sometimes when they see a new recipe it'll alter them to opportunities 
they hadn't considered.

The lineoffset fix was among those.  In v6 and earlier chunk delimiters 
could only be a single character, but v7 introduced multi-char 
delimiters.  Kinda cool if you need it, but a more costly substring 
compare is needed to support that.  With v8 they were able to fork the 
operation so that when a delimiter is only a single char it uses the 
older routine, saving the cooler-but-costlier substring compare only for 
when needed.

-- 
  Richard Gaskin
  Fourth World Systems
  Software Design and Development for the Desktop, Mobile, and the Web
  ____________________________________________________________________
  Ambassador at FourthWorld.com                http://www.FourthWorld.com