sven@cs.widener.edu (Sven Heinicke) (12/09/90)
What is quicker? int a = 0,i = 0; or int a,i; i = a = 0; and why? This type of thing has been bothering me for a long time. -- sven@cs.widener.edu Widener System Manager
gwyn@smoke.brl.mil (Doug Gwyn) (12/09/90)
In article <1990Dec08.222943.1581@cs.widener.edu> sven@cs.widener.edu (Sven Heinicke) writes:
-What is quicker?
- int a = 0,i = 0;
-or
- int a,i;
- i = a = 0;
-and why?
-This type of thing has been bothering me for a long time.
Why would you care? They're obviously not going to be bottlenecks..pfalstad@phoenix.Princeton.EDU (Paul John Falstad) (12/09/90)
In article <1990Dec08.222943.1581@cs.widener.edu> sven@cs.widener.edu (Sven Heinicke) writes: >What is quicker? > > int a = 0,i = 0; > >or > > int a,i; > i = a = 0; Depends. If these are local variables, then most likely they will produce the same code. The latter case is simply postponing the initialization. If these are global variables, the former case is quicker (in UNIX, anyway), since the compiler doesn't have to generate any code to initialize the two variables. In the latter case, it puts a and i in bss, but in the former case, it puts them in the initialized data segment. Your implementation could be totally different however. :-) -- From somewhere in front, you hear a repulsive squelching mumble, as if a chattering orangutan has suddenly had its mouth crammed full of earthworms. "You're invading my personal space!" shrieks the travel agent, rearing back from you like an offended snake. She is probably from California.
fjb@metaware.metaware.com (Fred Bourgeois) (12/11/90)
In article <14694@smoke.brl.mil> gwyn@smoke.brl.mil (Doug Gwyn) writes: >In article <1990Dec08.222943.1581@cs.widener.edu> sven@cs.widener.edu (Sven Heinicke) writes: >-What is quicker? >- int a = 0,i = 0; >-or >- int a,i; >- i = a = 0; >-and why? >-This type of thing has been bothering me for a long time. >Why would you care? They're obviously not going to be bottlenecks.. ... and if they *still* bother you, try generating assembly listing and see what the compiler does with them. ;-) -- Fred Bourgeois, MetaWare Inc., 2161 Delaware Avenue, Santa Cruz, CA 95060-2806 fjb@metaware.com ...!uunet!metaware!fjb Colorless Green Ideas Sleep Furiously, and so do I.
c145gmk@utarlg.utarl.edu (GORDON KEEGAN) (12/11/90)
In article <1990Dec08.222943.1581@cs.widener.edu>, sven@cs.widener.edu (Sven Heinicke) writes... >What is quicker? > > int a = 0,i = 0; > >or > > int a,i; > i = a = 0; > >and why? > >This type of thing has been bothering me for a long time. Our C compiler under VMS 5.3-1 produced the following assembler code for the respective C segments: int a=0, i=0; F8 AD 00 D0 000B movl #0,-8(fp) F4 AD 00 D0 000F movl #0,-12(fp) and int a, i; i = a = 0; F8 AD 00 D0 000B movl #0,-8(fp) F4 AD F8 AD D0 000F movl -8(fp),-12(fp) The second will be marginally slower, but only just... ----------------------------------------------------------------------------- | Gordon Keegan || Bitnet : c145gmk@utarlg | | Systems Programmer || THEnet : UTARLG::C145GMK | | Academic Computing Services || Internet: c145gmk@utarlg.utarl.edu | | University of Texas, Arlington || AT&TNet : 817-273-2208 | ----------------------------------------------------------------------------- | Beauty does what beauty does best: it's beautiful... | -----------------------------------------------------------------------------
userAKDU@mts.ucs.UAlberta.CA (Al Dunbar) (12/11/90)
In article <1990Dec08.222943.1581@cs.widener.edu>, sven@cs.widener.edu (Sven Heinicke) writes: >What is quicker? > > int a = 0,i = 0; > >or > > int a,i; > i = a = 0; > >and why? > >This type of thing has been bothering me for a long time. > >-- >sven@cs.widener.edu Widener System Manager The correct answer to the first question is "the one that takes less time to execute". At least that is about all that can be said until you say what compiler you are using. Once you've said that, the answer can be further refined by using a stopwatch. Since the first question cannot be answered, neither can the second, except to say "because it takes less time to execute". having said the above, a simple compiler *might* generate quicker code for the second case, because once the zero value is generated in a register to assign to a, it needn't be generated again. I believe, however, that most compilers worth the diskette they are sold on would optimize these fragments into identical code. -------------------+------------------------------------------- Al Dunbar | Edmonton, Alberta | "this mind left intentionally blank" CANADA | - Manuel Writer -------------------+-------------------------------------------
harrison@necssd.NEC.COM (Mark Harrison) (12/11/90)
In article <14694@smoke.brl.mil>, gwyn@smoke.brl.mil (Doug Gwyn) writes: > In article <1990Dec08.222943.1581@cs.widener.edu> sven@cs.widener.edu (Sven Heinicke) writes: > -What is quicker? > -This type of thing has been bothering me for a long time. > Why would you care? They're obviously not going to be bottlenecks.. Maybe he isn't worried about the speed, but about understanding how the compiler works. -- Mark Harrison harrison@necssd.NEC.COM (214)518-5050 {necntc, cs.utexas.edu}!necssd!harrison standard disclaimers apply...
bright@nazgul.UUCP (Walter Bright) (12/12/90)
In article <1990Dec08.222943.1581@cs.widener.edu> sven@cs.widener.edu (Sven Heinicke) writes:
-What is quicker?
- int a = 0,i = 0;
-or
- int a,i;
- i = a = 0;
-and why?
Why not:
1. Compile and disassemble the output, and see if there is any difference
in the code generated.
2. Write a simple timing loop to see which is faster.ccastdf@prism.gatech.EDU (Dave) (12/13/90)
bright@nazgul.UUCP (Walter Bright) writes: >In article <1990Dec08.222943.1581@cs.widener.edu> sven@cs.widener.edu (Sven Heinicke) writes: >-What is quicker? >- int a = 0,i = 0; >-or >- int a,i; >- i = a = 0; >-and why? Well, int a=0,i=0 generates the following lines of assembly: _main: jmp .L22 .L21: movl $0,-4(%ebp) movl $0,-8(%ebp) and int a,i; i=a=0; generates these lines: _main: jmp .L22 .L21: movl $0,-4(%ebp) movl $0,%eax movl %eax,-8(%ebp) Therefore, on a 386 Sequent with no optimization, the first is more effieient. Dave -- David Frascone Georgia Institute of Technology, Atlanta Georgia, 30332 Office of Computing Services--User Assistant
bright@nazgul.UUCP (Walter Bright) (12/15/90)
In article <18539@hydra.gatech.EDU> ccastdf@prism.gatech.EDU (Dave) writes:
/Therefore, on a 386 Sequent with no optimization, the first is more effieient.
Why are you concerned with which is more efficient if you turn off optimization?rlf@dptspd.sat.datapoint.com (Rory Foster) (12/19/90)
sven@cs.widener.edu (Sven Heinicke) writes: >What is quicker? > int a = 0,i = 0; >or > int a,i; > i = a = 0; >and why? >This type of thing has been bothering me for a long time. Well, I looked at it and (without optimizing) the first case was "quicker". I was using the UNIX SVR4 C compiler. If your compiler has an assembly option, (-S in Johnson's PCC) use it to see for yourself. When I optimize, my compiler shows no difference. So there you have it, it sort of just depends on what you're using at the time. Regards, Rory Foster rlf@dptspd.sat.datapoint.com
zsinwatt@qut.edu.au (James Watt) (12/19/90)
In article <rlf.661536169@dptspd>, rlf@dptspd.sat.datapoint.com (Rory Foster) writes: > sven@cs.widener.edu (Sven Heinicke) writes: > >>What is quicker? > >> int a = 0,i = 0; > >>or > >> int a,i; >> i = a = 0; > >>and why? > >>This type of thing has been bothering me for a long time. > > > Well, I looked at it and (without optimizing) the first case was "quicker". > I was using the UNIX SVR4 C compiler. If your compiler has an assembly option, > (-S in Johnson's PCC) use it to see for yourself. > > When I optimize, my compiler shows no difference. So there you have it, > it sort of just depends on what you're using at the time. On a HP9000 840 ( without optimisation), one is better than two, and with optimization they are basically similar. However if you use the int a,i=0; form of declaration and optimize it, it outperforms both one and two. In reality there are far too many variables to make this kind of exercise have any real relevance to the meaning of life in general. -- |----------------------------------------------------------------------------| | James Watt | AARnet: james@water.fit.qut.edu.au ARPA: ZSINWATT@QUT.EDU.AU | |----------------------------------------------------------------------------|