Can Fasm compile faster?

Index > Main > Can Fasm compile faster?

Author

Thread

rocketsoft

Joined: 26 Jan 2010
Posts: 189

rocketsoft 07 Feb 2011, 15:33

System:I7-870@3.94Ghz

Self compile of rasm:44 millisecond (7 passes, 124000 bytes, mostly code)
Self compile of fasmw:94 millisecond (6 passes, 138000 bytes, mostly code)
note: i increased the resolution of fasmw´s timer to 1 millisecond

As u can see RASM compiles OVER twice as fast as FASM
It only matters offcourse when compiling VERY BIG sources

Rasm has no preprocessor... that could explain the difference
Fasm uses binary trees... and Rasm uses a faster database system

I have no need yet for preprocessing in any of my applications
u only loose EQU, which does literal substitution

07 Feb 2011, 15:33

revolution
When all else fails, read the source

Joined: 24 Aug 2004
Posts: 20418
Location: In your JS exploiting you and your system

revolution 07 Feb 2011, 15:40

Are you trying to start a speed war?

Without a preprocessor then that means you have no macros? If so then ewww, so many tedious things to code with no macros to do the drudge work for you.

07 Feb 2011, 15:40

Tomasz Grysztar

Joined: 16 Jun 2003
Posts: 8356
Location: Kraków, Poland

Tomasz Grysztar 07 Feb 2011, 15:55

Can fasm compile faster? Most certainly - however the improvements would be mostly not worth the time spent on developing them. And another bad news - fasm 2 (if/when I finally write it) is most probably going to be slower than 1.x.

As for the comparison - though the method of comparing the time in which assemblers compile itself was sometimes used by Betov, it is really the "apples vs oranges". If you want to do get any valuable results, you should compile some identical (up to the syntax differences) source with both tools. Randall Hyde once did such a benchmarking, and you can find the test generator here, but I have another nice source for testing. It is something that was a real project once posted on Win32Asm community board - I attach the source of C program which in turn generates the assembly source.
BTW, the test shows that fasm 1.69.31 is noticeably slower than 1.68. So the trend towards the slow 2.0 is already visible. Wink

Description:	Test source generator.	Download
Filename:	GenAsm2.c
Filesize:	19.96 KB
Downloaded:	543 Time(s)

07 Feb 2011, 15:55

YONG

Joined: 16 Mar 2005
Posts: 7997
Location: 22° 15' N | 114° 10' E

YONG 08 Feb 2011, 09:13

Tomasz Grysztar wrote:

And another bad news - fasm 2 (if/when I finally write it) ...

Oh no!

I thought that fasm 2 would be available by 2020. But now ... I believe that fasm 2 will never be available! Sad

08 Feb 2011, 09:13

idle

Joined: 06 Jan 2011
Posts: 440
Location: Ukraine

idle 08 Feb 2011, 10:00

2 -> 1.69.xx, y-not :p
div_64(used by fasm) can be replaced with amd's variant, where 2 divs used as maximum
i can explain it in a bit different way if you do not understand the algo

08 Feb 2011, 10:00

rocketsoft

Joined: 26 Jan 2010
Posts: 189

rocketsoft 08 Feb 2011, 15:15

Offcourse RASM has macros (Its very Tasm compatible)
preprossesing is only needed INSIDE macros
Normal code does not need preprossesing

08 Feb 2011, 15:15

ouadji

Joined: 24 Dec 2008
Posts: 1081
Location: Belgium

ouadji 08 Feb 2011, 16:25

Quote:

Normal code does not need preprossesing

in this case why not give the choice ? (with a directive)
By default the preprocessing is "on",
but if you don't use macros,
the directive "preprocessing_off" to disable preprocessing.

Code:

format PE large GUI 4.0
entry start
preprocessing_off
start:

_________________
I am not young enough to know everything (Oscar Wilde)-

08 Feb 2011, 16:25

SFeLi

Joined: 03 Nov 2004
Posts: 138

SFeLi 08 Feb 2011, 17:09

ouadji, for what? Even fasm itself compiles in less than a second.

08 Feb 2011, 17:09

ouadji

Joined: 24 Dec 2008
Posts: 1081
Location: Belgium

ouadji 08 Feb 2011, 17:45

yes, i'm agree.
fasm suits me perfectly as it is.

_________________
I am not young enough to know everything (Oscar Wilde)-

08 Feb 2011, 17:45

MinhHung

Joined: 10 Sep 2010
Posts: 51
Location: Viet Nam

MinhHung 08 Feb 2011, 21:22

can UPX make fasm faster. http:\\upx.sf.net

08 Feb 2011, 21:22

Ivan2k2

Joined: 08 Sep 2004
Posts: 80
Location: Russia, Angarsk

Ivan2k2 09 Feb 2011, 10:51

MinhHung wrote:

can UPX make fasm faster. http:\\upx.sf.net

UPX will not make executables faster

09 Feb 2011, 10:51

revolution
When all else fails, read the source

Joined: 24 Aug 2004
Posts: 20418
Location: In your JS exploiting you and your system

revolution 09 Feb 2011, 10:53

But maybe fasm can make UPX faster.

09 Feb 2011, 10:53

ManOfSteel

Joined: 02 Feb 2005
Posts: 1154

ManOfSteel 09 Feb 2011, 12:45

MinhHung wrote:

can UPX make fasm faster. http:\\upx.sf.net

It will not make the internals of fasm go faster. On the other hand, it will slow down (just a little a bit) the execution of the fasm executable since it now has to uncompress itself before execution.

revolution wrote:

But maybe fasm can make UPX faster.

09 Feb 2011, 12:45

revolution
When all else fails, read the source

Joined: 24 Aug 2004
Posts: 20418
Location: In your JS exploiting you and your system

revolution 09 Feb 2011, 12:53

ManOfSteel wrote:

It will not make the internals of fasm go faster. On the other hand, it will slow down (just a little a bit) the execution of the fasm executable since it now has to uncompress itself before execution.

Well that is part of the debate I guess. Can UPX uncompress faster than the OS can load the extra bytes from the HDD? For first time runs the answer might be yes. For subsequent runs coming from cache the answer might be no. For smallish programs like fasm it probably makes absolutely no perceivable difference to the user at the keyboard.

09 Feb 2011, 12:53

MinhHung

Joined: 10 Sep 2010
Posts: 51
Location: Viet Nam

MinhHung 09 Feb 2011, 13:13

hehe. Sorry. I think best compiler not only fast , best compiler easy to understand->code easy. Now, computer processor very fast. I think 'code easy' bester than 'fast compiler'

09 Feb 2011, 13:13

f0dder

Joined: 19 Feb 2004
Posts: 3175
Location: Denmark

f0dder 13 Feb 2011, 20:06

revolution wrote:

ManOfSteel wrote:
It will not make the internals of fasm go faster. On the other hand, it will slow down (just a little a bit) the execution of the fasm executable since it now has to uncompress itself before execution.
Well that is part of the debate I guess. Can UPX uncompress faster than the OS can load the extra bytes from the HDD? For first time runs the answer might be yes. For subsequent runs coming from cache the answer might be no.

Especially since the from-disk-bytes will then be served from cache, but you'll still be doing the decompression overhead - and on top of that, multiple instances of the same program won't benefit from copy-on-write page sharing.

_________________

- carpe noctem

13 Feb 2011, 20:06

edfed

Joined: 20 Feb 2006
Posts: 4352
Location: Now

edfed 13 Feb 2011, 21:21

a computer, even used for an intensive threading, have a lot of time without doing anything.
fasm does really something, it lets you speak directlly to the computer and take control of the machine. then no matter if it is slow (and it is not), it does something that no other assembler does, it lets you code assembler easy, and uses a very good syntax, and a powerfull macro processor, then, it can be a little slower than other assemblers (and it is not), because it does something that it is the only one to do.

maybe fasm2 will not be so good, just wait to see what IDE it will have. bloat like a lot of IDE or clean as fasmw1.xx?

13 Feb 2011, 21:21

f0dder

Joined: 19 Feb 2004
Posts: 3175
Location: Denmark

f0dder 13 Feb 2011, 23:33

edfed wrote:

maybe fasm2 will not be so good, just wait to see what IDE it will have. bloat like a lot of IDE or clean as fasmw1.xx?

Any reason the current IDE has to change much just because the assembler backend becomes more powerful?

13 Feb 2011, 23:33

Teehee

Joined: 05 Aug 2009
Posts: 570
Location: Brazil

Teehee 17 Feb 2011, 14:58

17 Feb 2011, 14:58

mattst88

Joined: 12 May 2006
Posts: 260
Location: South Carolina

mattst88 23 Feb 2011, 06:28

f0dder wrote:

edfed wrote:
maybe fasm2 will not be so good, just wait to see what IDE it will have. bloat like a lot of IDE or clean as fasmw1.xx?
Any reason the current IDE has to change much just because the assembler backend becomes more powerful?

It's the same logic that causes him to use Windows 98. Smile

_________________
My x86 Instruction Reference -- includes SSE, SSE2, SSE3, SSSE3, SSE4 instructions.
Assembly Programmer's Journal

23 Feb 2011, 06:28

< Last Thread | Next Thread >

Forum Rules:

You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum
You cannot attach files in this forum
You can download files in this forum