Posts: 979
Threads: 27
Joined: Aug 2001
Well, it's open source, so really "anything you want", maybe with the exception of packaging porn or viruses. I mean, you could still do that but it's discouraged.
But on a more serious note, I'm positive there are optimizations to be made. I mean, it's a fairly new compiler, about 3 times as fast as VBDOS in the compiler program benchmark, which is currently the only existing FB benchmark (
).
Posts: 115
Threads: 11
Joined: Jun 2004
Almost certainly a lame comment, but...
Is it going to be worth checking out on all the various Win32 OSs at some point, both NT and 9x?
Posts: 336
Threads: 4
Joined: Nov 2004
I was thinking more along the lines of what Optimzation stuff has already been add and what does the Compiler leands it self to for other optimzation tecniques like Interprocedural Analysis or Function inlining.
Posts: 788
Threads: 53
Joined: Nov 2002
None, it's not a optimising compiler. But it generates good code. So it beats amny other basic compilers, such as powerbasic for one. And if he ever finished the DAG module then it will remove redudant instructions which will make it more effecient. It will also do strength reduction i guess.
oship me and i will give you lots of guurrls and beeea
Posts: 336
Threads: 4
Joined: Nov 2004
no i mean how easly does the compilers source code leand to adding things like Interprocedural Analysis , Function inlining, or Complex Branch Optimization. what i mean to add somthing like this would it mean reworking half the compiler or could these things be done modularly to the intermided language
Posts: 3,288
Threads: 167
Joined: Nov 2001
v3cz0r!!!! Pat yourself on the back!!!!
I just was able to get PTC to work with MingW and FB goes in toe to toe with it in terms of speed. And that's with Memset for clearing the buffer.
In fact I can't notice much speed difference on full screen vsynch off, on this 233mhz comp. Kudos my friend!!!!
BTW, you got mail. Reports, etc.
Marcade & Aetherfox:
http://rel.betterwebber.com/junk.php?id=25
That's another one in CPP.
You have to pass the points to the sine functions inside the loop to produce this fx. I don't know about the rose though...
Posts: 788
Threads: 53
Joined: Nov 2002
It has a full compiler pipeline. lexer -> parser -> AST -> intermediate code -> basic blocks -> machine code.
Between the basic block stage and machince code, a lot of higher level optimisation could be done pretty easly. Adding a DAG will do a huge difference, then once could also add flow graphs for loop optimisations and such. That's where the best optimisation can be done.
oship me and i will give you lots of guurrls and beeea
Posts: 336
Threads: 4
Joined: Nov 2004
over a week and no news on the compiler. oh well
Posts: 3,279
Threads: 170
Joined: Nov 2003
If he spent all day posting news about it, he'd never get the thing coded.
Let the man work!
I'd knock on wood, but my desk is particle board.
Posts: 922
Threads: 15
Joined: Jun 2003
Quote:over a week and no news on the compiler. oh well
Eh? Did you check the other topics on this forum? I doubt it..
Compiler development is going fine, but i need some time, i can't code that fast.. maybe til the end of this month there will be the release, no promises this time, sorry.