PureBasic 6.01 LTS is released !

[jack]

thank you Fred for the update and for a newer gcc version

[luis]

Updated GCC on Windows to 12.2.0

Oh, that's nice, from 8.1 to the last one !

These are the results with a program I've used during this beta cycle to look for all the variations in performance we had (fast without volatile, slow with volatile, super-slow with just some volatile and a floating point problem, and finally now with some volatile, the floating point problem fixed and some very good performances),

Code: Select all

; x 86

; 6.01 B4
; ASM = 17440 
; ASM (opt) = 17220    
; C = 58286    
; C (opt) = 17343    

; 6.01 B5
; ASM = 17492 
; ASM (opt) = 17399    
; C = 58293
; C (opt) = 17522  

; x64

; 6.01 B4
; ASM = 15558 
; ASM (opt) = 15497    
; C = 20928    
; C (opt) = 10344    

; 6.01 B5
; ASM = 15671
; ASM (opt) = 15650    
; C = 18317   
; C (opt) = 10465    

It's interesting how the unoptimized C is pretty slow, while the optimized one is similar to the PB ASM @ 32 bits and sensibly faster @ 64 bits.

Also in general the 64 bit code is faster.

This is just one specific program with nested loops, floating point and a lot of memory access through pointers, so your results may vary but it's still interesting.

[BarryG]

C Backend: [Done] Compiler error for ProcessIdToSessionId

Just tried this with 6.01 beta 5, and it's not "Done" because the compiler still hangs due to using ASM with the C backend. I suggest maybe the compiler can detect such invalid code and raise an error, instead of showing "Compiling" for 10+ minutes?

[Psychophanta]

Let us try out the beta 5

[infratec]

WebGadget() reports now:

https://user-agents.net/string/mozilla- ... ecko-dnt-1

Which is IE 11.

[Fred]

C Backend: [Done] Compiler error for ProcessIdToSessionId

Just tried this with 6.01 beta 5, and it's not "Done" because the compiler still hangs due to using ASM with the C backend. I suggest maybe the compiler can detect such invalid code and raise an error, instead of showing "Compiling" for 10+ minutes?

Should be better with next version

[BarryG]

Wow, thanks Fred! I was under the impression it wasn't possible to fix (based on other people's comments). Look forward to testing it!

[StarBootics]

Apparently the version 6.01 LTS is out. Unfortunately I have a large program that compile fine with the ASM backend but I get an Assembler error with the C backend after that the IDE freeze completely. I don't have source code with asm code inside.

I'm under Debian 11 and I'm using PureBasic For Ubuntu 18.04.

I'm investigating and I hope I will find the error.

Edit : I have found my error. That being said, for that kind of error I should have got an Invalid Memory Access error and not an Assembly error. I will provide a small code example to demonstrate that.

Best regards
StarBootics

[skywalk]

Wow, you are right? Did not see the announcement
My large apps also hang on C compile.

[StarBootics]

Wow, you are right? Did not see the announcement
My large apps also hang on C compile.

I got it from the IDE when I have launch it. But no announcement on the forum about PB V6.01 LTS

Best regards
StarBootics

[Fred]

Yes, it's released ! You can grab it on your personal account. Thanks a lot to everyone during this bug-fix marathon

[idle]

Yes, it's released !

Thanks for the update really great to see the c backend progressing.

[Otrebor]

Thanks!!

[Fred]

A new quick fix version as just been pushed, so please update if you already got the final.

[Janni]

Thanks for all your hard work Fred!