Is it just me or are compiled .exe au3 scripts running faster & less likely to crash?

[danaman123]

I'm running/coding/debugging an au3 project with 10k+ lines of code. I started to see the IDE run into straight up blank errors, even with no memory leak... and other strange errors I can't explain... such as requiring me to use the number() function to declare variables as integers. After 3k lines of code I found that compiling sections of the code into .exes and having them communicate to each other through a database or .txt solved all of my problems! The now APP runs way faster, bug free & crash free.

Anyways this is just my experience and I wonder if anyone else has had a similar experience?

Also I hope this post can help other's who are managing/developing a large Au3 project. Au3 is a power language & so much easier to read/write than python & rstudio or any other language for that matter.  I'd rather manipulate 1 million rows of data in an array format in Au3 that use rstudio...

I hope Au3 gets continued interest/development from the public! 

 

[Jos]

Quote

Is it just me or are compiled .exe au3 scripts running faster & less likely to crash?

It is you as a compiled script is the bytecode of the script and runtime module packed into the EXE, so basically doing the exact same thing.

... having said that: When you use au3stripper, the cases aren't the same anymore as the source of the compiled script is "much" smaller and we run a bit faster.

Jos

Edited August 15, 2020 by Jos

[Exit]

7 minutes ago, Jos said:

au2stripper

?

[Jos]

[danaman123]

3 hours ago, Jos said:

It is you as a compiled script is the bytecode of the script and runtime module packed into the EXE, so basically doing the exact same thing.

... having said that: When you use au3stripper, the cases aren't the same anymore as the source of the compiled script is "much" smaller and we run a bit faster.

Jos

I tried VBA & RStudio for running complex algorithms for training Ai, to my surprise all this Rstudio hype is BS, sometimes VBA is faster at certain tasks. Both Rstudio & VBA completed a calculation in around 15 seconds. 15 seconds per row for 100k rows would take me 416 hours or 10,000 days to complete... Both VBA & Rstudio couldn't perform any true multi-threading... VS Autoit I was able to run 10 compiled .exe processors for processing data & utilize all of my cores at 100%!   Which means Autoit it is 10x faster for processing data than Rstudio!

[argumentum]

3 hours ago, danaman123 said:

having them communicate to each other through a database or .txt solved all of my problems

..there are IPC that are quite good. Search around, give'em a try  

[jchd]

5 hours ago, danaman123 said:

Which means Autoit it is 10x faster for processing data than Rstudio!

A doubtfull assertion!

[RTFC]

16 hours ago, danaman123 said:

I'd rather manipulate 1 million rows of data

If you're doing numerical operations on this scale, then E4A (even when running interpreted only) is several orders of magnitude faster than pure AutoIt (and supports multi-processing with shared data containers on x64).

Edited August 16, 2020 by RTFC

[danaman123]

7 hours ago, RTFC said:

If you're doing numerical operations on this scale, then E4A (even when running interpreted only) is several orders of magnitude faster than pure AutoIt (and supports multi-processing with shared data containers on x64).

Thanks I'll check them out! I've never heard of them before, right now I'm running shared LAN computing with 6 computers processing a 6 way split data of the data with 8 .exes of Autoit running per computer. All 6 computers are at 100% CPU use. The array size read doesn't seem to affect the processing speed much but the algorithms itself uses the most CPU power and requires the most processing time.

[jchd]

Isn't it a good candidate for CUDA processing?