Board index » delphi » best Compiler for best Code

best Compiler for best Code

Hi together,

i think some time ago, there was a press release (or maybe it was just in a
readme) from borland concerning compiling assembler code and the efficiency
of borlands delphi-compiler. The essence was, that you dont need to code in
Assembler anymore as Borlands pascal-compiler produces <nearly> optimal
code.

I need this statement (as a quote - thats why i need the original source
and type of publication) for my thesis - does anyone remember?

thanks in advance!

 

Re:best Compiler for best Code


Quote
"Holger Dinkel" <Hogda...@web.de> wrote in message

news:Xns91C2C7A862770HogDaRog@131.188.3.24...

Quote
> Hi together,

> i think some time ago, there was a press release (or maybe it was just in
a
> readme) from borland concerning compiling assembler code and the
efficiency
> of borlands delphi-compiler. The essence was, that you dont need to code
in
> Assembler anymore as Borlands pascal-compiler produces <nearly> optimal
> code.

> I need this statement (as a quote - thats why i need the original source
> and type of publication) for my thesis - does anyone remember?

I don;t, however I believe that ever since the pentium II it;s been fairly
pointless programming in assembler since one can rarely get enough of a
pay-off in terms of performance versus effort. It;s also with the pentium or
PII that linked lists as opposed to dynamic arrays became a much of a
muchness for most situations, which is why you don;t see so many linked
lists these days.

Also do you know about Very Long Word processors (like the Itanium)? They
make assembler a total waste of time, and exceptionally difficult any way.

Greg

Re:best Compiler for best Code


Hi Greg

What makes the Itanium, coding in assembler more difficult then a 80*86 ?

Just curious.

victor

Quote
"Greg Lorriman" <greg_lorri...@bigfoot.com> wrote in message

news:1014837567.8211.0.nnrp-13.9e98bd70@news.demon.co.uk...
Quote
> "Holger Dinkel" <Hogda...@web.de> wrote in message
> news:Xns91C2C7A862770HogDaRog@131.188.3.24...
> > Hi together,

> > i think some time ago, there was a press release (or maybe it was just
in
> a
> > readme) from borland concerning compiling assembler code and the
> efficiency
> > of borlands delphi-compiler. The essence was, that you dont need to
code
> in
> > Assembler anymore as Borlands pascal-compiler produces <nearly> optimal
> > code.

> > I need this statement (as a quote - thats why i need the original
source
> > and type of publication) for my thesis - does anyone remember?

> I don;t, however I believe that ever since the pentium II it;s been fairly
> pointless programming in assembler since one can rarely get enough of a
> pay-off in terms of performance versus effort. It;s also with the pentium
or
> PII that linked lists as opposed to dynamic arrays became a much of a
> muchness for most situations, which is why you don;t see so many linked
> lists these days.

> Also do you know about Very Long Word processors (like the Itanium)? They
> make assembler a total waste of time, and exceptionally difficult any way.

> Greg

Re:best Compiler for best Code


"Greg Lorriman" <greg_lorri...@bigfoot.com> wrote in
news:1014837567.8211.0.nnrp-13.9e98bd70@news.demon.co.uk:

Quote
> I don;t, however I believe that ever since the pentium II it;s been
> fairly pointless programming in assembler since one can rarely get
> enough of a pay-off in terms of performance versus effort. It;s also
> with the pentium or PII that linked lists as opposed to dynamic arrays
> became a much of a muchness for most situations, which is why you don;t
> see so many linked lists these days.

well, for me personally its not 100% sure if manually inserted ASM-code
really speeds up things or not. All this bitshifting-stuff is not so easy to
implement in Delphi/Pascal, is it? An example: I have an asm-routine which
performs an 32Bit-XOR and results the number of bits (or "1") set in the
result (for some sort of correlation). I havent found a really quick routine
for that in Delphi, but thats just btw...

Quote
> Also do you know about Very Long Word processors (like the Itanium)?
> They make assembler a total waste of time, and exceptionally difficult
> any way.

ok but do YOU develop programs for the Itanium? ;-)

Nevertheless: Thank you!

Re:best Compiler for best Code


Quote
> What makes the Itanium, coding in assembler more difficult then a 80*86 ?

I read in an magazine (Computer'Techniek - Dutch, of origin German) that the
Itanium takes 3 asm-instructions at a time and executes them paralel. The
only catch in this is that these instructions may not use the same
registers. This makes it verry difficult to optimize. If you can't find a
suitable instruction you have to insert nops...
That is why Intel is helping the developers of the gcc-compiler (linux) with
their optimizing routines.

I hope this clears some things up.

Mark Elissen

"balbaro" <balb...@skynet.be> schreef in bericht
news:dJnf8.1634$tC.151@news.easynews.com...

Quote
> Hi Greg

> What makes the Itanium, coding in assembler more difficult then a 80*86 ?

> Just curious.

> victor

> "Greg Lorriman" <greg_lorri...@bigfoot.com> wrote in message
> news:1014837567.8211.0.nnrp-13.9e98bd70@news.demon.co.uk...
> > "Holger Dinkel" <Hogda...@web.de> wrote in message
> > news:Xns91C2C7A862770HogDaRog@131.188.3.24...
> > > Hi together,

> > > i think some time ago, there was a press release (or maybe it was just
> in
> > a
> > > readme) from borland concerning compiling assembler code and the
> > efficiency
> > > of borlands delphi-compiler. The essence was, that you dont need to
> code
> > in
> > > Assembler anymore as Borlands pascal-compiler produces <nearly>
optimal
> > > code.

> > > I need this statement (as a quote - thats why i need the original
> source
> > > and type of publication) for my thesis - does anyone remember?

> > I don;t, however I believe that ever since the pentium II it;s been
fairly
> > pointless programming in assembler since one can rarely get enough of a
> > pay-off in terms of performance versus effort. It;s also with the
pentium
> or
> > PII that linked lists as opposed to dynamic arrays became a much of a
> > muchness for most situations, which is why you don;t see so many linked
> > lists these days.

> > Also do you know about Very Long Word processors (like the Itanium)?
They
> > make assembler a total waste of time, and exceptionally difficult any
way.

> > Greg

Other Threads