On Thursday 26 April 2001 09:05, Andreas Metzler wrote: > > performance: By having images optimized for each processor on i386 > > users should see better performance. I don't believe performance > > numbers were quantified in the discussion but quantifying performance > > is probably important to evaluating this tradeoff. > > [...] > > Hello! > _Afair_ it is necessary to run a k6 (or athlon) "optimized" kernel to > use 3DNow! in applications like xmms or lame. This probably applys to > ISSE, MTTR and MMX, too.
I agree that having kernels compiled with support for different features is a good idea. We must provide an i386 kernel. There is software available which only runs with MMX. We should offer a kernel with MMX support which requires Pentium-MMX to support running this. For 3DNow! we should have a kernel which supports it. There is no need for a MTTR specific kernel. MTTR is not really needed as there is no software written which is unable to run without it. Our goal here should be compatibility with software. MTTR can increase speed significantly in certain situations, but there's lots of other ways of doing that for less effort which we aren't supporting. MTTR can allow you to work around broken hardware. But we can't provide enough kernels to support all combinations of broken hardware (I am sure that I could find a list of a dozen boolean options which are all needed to be in one state or another for various broken hardware - we can't provide 2^12 kernels). Also I think that we should have an SMP kernel in the list. Another issue is support for all the different SCSI controllers and RAID controllers. Perhaps we should spend more time working on initrd support? -- http://www.coker.com.au/bonnie++/ Bonnie++ hard drive benchmark http://www.coker.com.au/postal/ Postal SMTP/POP benchmark http://www.coker.com.au/projects.html Projects I am working on http://www.coker.com.au/~russell/ My home page
