On Sat, 19 May 2012, Thorsten Glaser wrote: > Finn Thain dixit: > > >Has anyone benchmarked those binaries (running on the hybrid platform) > >against native 68020 and Coldfire binaries (running on their respective > >native platforms)? > > Since I?ve seen an strace and been explained what SYS_333 is I believe > benchmarks of modern-eglibc m68k are beyond salvagable anyway. > > Maybe a new ABI, with a register reserved for TLS, would actually be > faster, despite being more generic, since it saves kernel calls?
You could benchmark that also. But it isn't really relevant to the problem. > > When doing a generic *nix distribution, and it?s not called Gentoo, you > *need* to produce generic executables. (Which is why ARM is > fundamentally doomed, as you have not only one kernel image per board, > but also arm, armeb, armel(debian), armel(ubuntu), armhf by now. And > with devices being outdated before they hit the market, and the market > being a cheap throw-away one, I see it getting only worse.) That kind of fragmentation may or may not be a problem for CF in the future ... who knows? Finn > > bye, > //mirabilos > -- To UNSUBSCRIBE, email to debian-68k-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/alpine.LNX.2.00.1205191344471.4964@nippy.intranet
