PureBasic 6.03 LTS is released !

[mk-soft]

@Fred,

Sorry to sound bad, but the code generated (executable) has the same size as beta 5 and it has the same speed (if not slower).

Am I missing something?

Thanks!

EDIT: I am referring to the C backend, of course.

The c-code output has changed somewhat.

[Quin]

Thanks for all your work squashing those pesky bugs, Fred!

[marcoagpinto]

Buaaaaaaaaaaaaaaaa

Beta 6 is slower than beta 5, even with all the optimisations I have been doing in the source-code of my software.

Windows 11, C-backend, PureBasic x64.

[Psychophanta]

@marcoagpinto
That's very scary
I hope it is solved.

[mk-soft]

Test this beta ide
Link: viewtopic.php?p=607026#p607026

[Fred]

Buaaaaaaaaaaaaaaaa

Beta 6 is slower than beta 5, even with all the optimisations I have been doing in the source-code of my software.

Windows 11, C-backend, PureBasic x64.

We didn't changed anything with code generation between beta 5 and 6, if you can see where it is slower please post a bug report.

[marcoagpinto]

Buaaaaaaaaaaaaaaaa

Beta 6 is slower than beta 5, even with all the optimisations I have been doing in the source-code of my software.

Windows 11, C-backend, PureBasic x64.

We didn't changed anything with code generation between beta 5 and 6, if you can see where it is slower please post a bug report.

@Fred

I had made an optimisation of my PhD software in beta 5 that made 20 runs of it to take less 40 seconds.

Now, I compiled with beta 6, and it takes longer than with the change above, more 5 seconds in 20 runs.

I know 5 seconds doesn't sound much, but I have to generate data of many thousands of runs and 5 seconds more in each 20 runs produces a huge difference in all these runs.

[Quin]

Are you able to give a code sample?

Buaaaaaaaaaaaaaaaa

Beta 6 is slower than beta 5, even with all the optimisations I have been doing in the source-code of my software.

Windows 11, C-backend, PureBasic x64.

We didn't changed anything with code generation between beta 5 and 6, if you can see where it is slower please post a bug report.

@Fred

I had made an optimisation of my PhD software in beta 5 that made 20 runs of it to take less 40 seconds.

Now, I compiled with beta 6, and it takes longer than with the change above, more 5 seconds in 20 runs.

I know 5 seconds doesn't sound much, but I have to generate data of many thousands of runs and 5 seconds more in each 20 runs produces a huge difference in all these runs.

[marcoagpinto]

Are you able to give a code sample?

How? The source-code is around 340K (340 000) lines with endless calculations and equations, it is impossible for me to know where in it is the speed loss.

[Psychophanta]

@marco
if you want to demonstrate your experience, then you should put to work imagination to show the issue.

[AZJIO]

How?

Code: Select all

StartTime = ElapsedMilliseconds()
Delay(1000) ; code under test
Debug ElapsedMilliseconds() - StartTime
Delay(1000) ; code under test
Debug ElapsedMilliseconds() - StartTime

Divide the code into blocks or divide it into 10 parts 340/10=34. After every 34000 lines, insert a time output line. Run on different versions to compare timestamps, which areas are running slower. Next, divide the slow section by 10, that is, insert a timestamp every 3.4K. And so we get to the desired function.

[Fred]

Code: Select all

- Added #PB_String_NoCaseAscii support for CompareMemoryString() to have much faster string comparison when only ascii characters are involved

[jacdelad]

Very nice, thanks!

[marcoagpinto]

Thank you, Fred!

[skinkairewalker]

awesome, Fred !
thanks