8 Jan
2006
8 Jan
'06
noon
Scott Stevens wrote:
On Sat, 07 Jan 2006 17:29:22 +0100
Holger Veit <holger.veit at ais.fraunhofer.de> wrote:
Concerning that playstation game which I don't know, I'd rather
interpret that as advertising bullshit. One could simply count the
number of people involved in programming the core and estimate how many
lines of assembly one would have needed to write. If the result were
even in the 10**5..6 range or even higher, I'd consider this unrealistic
unless one speaks of excessive code duplication. Some code like the OS/2
kernel is about 2 thirds of assembly code, the rest is 16 bit or 32 bit
C, but this is about 800K executable size - not really millions of LOC.
An 800k executable can be the result of millions of lines of
assembly code.
Certainly at least a million or several.
I do not talk about comment lines or slack header files but lines that
actually produce a single or multibyte instruction.
I also do not talk about compilers, code generators and linkers which
can *produce* an 800k file, but are by itself surely larger than its result.
And given an executable is typically not pure code, but contains
structural data (debug symbols, references, fixups, etc.) the 800k
executable turns out to be even less than 800000 code or data bytes
effectively.
Holger