Re: Review of YDL 2 on Slashdot


Subject: Re: Review of YDL 2 on Slashdot
From: Timothy A. Seufert (tas@mindspring.com)
Date: Wed Jul 18 2001 - 13:14:48 MDT


At 10:16 AM -0400 7/18/01, Sean O. Denney wrote:

>It looks as if YDL 2 has gotten a review on /. I think it's a pretty
>good review, except the comparison between the G4/500, the PIII/800
>and the Athlon/1.2. The comparison they made was on kernel
>compilation and mp3 encoding.
>
>The flaw with this guys review was that he thought because RISC runs
>more efficiently and faster than CISC that compiling would be faster.
>Compiling for RISC is always slower because the instructions have to
>be broken down into smaller sizes (32-bit) rather then the variable
>size of x86 instructions.

I hope you don't go around spreading this kind of BS to counter the
other BS. It usually seems to be true that GCC is less efficient
when compiling for PPC, but this isn't the reason, if for no reason
other than the fact that x86 instructions are (on average) even
smaller than 32 bits.

The only valid way to compare compilation speeds is to make sure both
platforms are doing the exact same work, which means cross
compilation. Something like cross compiling the kernel (with the
same source tree and .config) targeting an entirely different
architecture (say, MIPS).

Most importantly, despite the propaganda from Apple, a G4/500 isn't
generally faster than 1 GHz x86 CPUs. Software that really meshes
well with AltiVec can compete. But for a program like GCC, it's
practically guaranteed that a PIII/800 and an Athlon 1.2 will beat a
G4-500.

   Tim Seufert



This archive was generated by hypermail 2a24 : Wed Jul 18 2001 - 12:19:32 MDT