[Zlib-devel] To ASM or not to ASM?

Mark Adler madler at alumni.caltech.edu
Sat Oct 4 13:40:57 EDT 2003


On Friday, October 3, 2003, at 05:55 PM, Cosmin Truta wrote:
> Based on my tests, and on the tests posted by Chris Anderson, I think
> that it's worth using the decompression ASM code (ASMINF), which makes 
> a
> noticeable difference; but it's not worth using the compression code
> (ASMV). In my opinion, the small speed improvements do not justify the
> risk of using ASM code.

That still leaves open the question: what is the risk of using ASMINF?  
How exhaustively has the assembler code been tested?  Has someone other 
than the author walked through the assembler source code to validate 
it?  I'm nervous since making it part of the zlib1.dll available from 
the zlib web site means that I'll have to support it.

mark





More information about the Zlib-devel mailing list