|
Post by tenochtitlanuk on Nov 27, 2020 14:32:34 GMT -5
A poster on JB forum was interested in changing bmp files, pixel by pixel. Something I've done a lot.
In the words of the Irishman asked how to get to Dublin and replying 'Well I wouldn't start from here'. Gonna be slow....
The method I've showed recently, of working on ppm images at file level, produces the two transforms of a 500x500 image seen below in about 2.5 seconds each in LB5 on Linux on an old laptop. First image read as triplets of RGB. Re-written with RGB interchanged or rotated.
Slower on LB4, and problems in JB for some reason. Will post the points or problems that came up on the appropriate forum section. For example, he unfinished time$() always giving times in ms was worrying when it implied I must have gone to sleep for an hour or so!!
I'd hoped to have the same code run on all three platforms for comparison, but have yet to achieve it. Page on my site soon with code...
|
|
|
Post by tenochtitlanuk on Nov 28, 2020 9:13:27 GMT -5
For interest, JB was slightly faster than LB4, but LB5 wins hands down....
Sorry I didn't correct for the faultyLB5 time$( "seconds") reporting in milliseconds!
|
|
|
Post by Carl Gundel on Nov 28, 2020 18:42:33 GMT -5
For interest, JB was slightly faster than LB4, but LB5 wins hands down.... Sorry I didn't correct for the faultyLB5 time$( "seconds") reporting in milliseconds!
Thanks. Can you provide source code?
|
|
|
Post by tenochtitlanuk on Nov 29, 2020 11:23:43 GMT -5
|
|
|
Post by Carl Gundel on Dec 1, 2020 15:42:33 GMT -5
For interest, JB was slightly faster than LB4, but LB5 wins hands down....
Sorry I didn't correct for the faultyLB5 time$( "seconds") reporting in milliseconds! It's interesting that JB is slightly faster than LB4. It seems pretty consistent, but they are running the same code so it's surprising.
|
|