joef@extro.ucc.su.OZ.AU () writes:

quoted text
>  Well, I'm impressed!  Comms is _REALLY_ fixed. 

Yes.. I don't seem to be dropping any characters anymore.  (I haven't
tried at drastically higher bps rates yet, but it looks good. Thanks,
Linus)

But, there *is* a problem. The whole system is *much* slower. Screen
updates come in chunks of a half-dozen lines at at time, and emacs
takes *ages* to load -- some times it's even aborted, saying `Alarm
Clock'. I'm not game to go near gcc like this.

While it's hard to say, and I haven't tried any real tests, it doesn't
appear to be a disk problem -- programs are slow to print screenfuls
of text which should not require any disk IO. There's plenty of time
to watch emacs' clever screen-painting algorithm at work when moving
through files, for instance. It almost looks as though the `turbo'
switch has been turned off, but that's not really possible since I
don't have one wired in.  Swapping VC's is quick, like before, for
what little that information is worth.

This is using the stock bootimage from banjo, on about the slowest
machine possible: 38(6+7)sx-16 with 3Mb of memory (384k eaten by the
board) It's been slow, but quite usable, up to now.  Stripping out the
(in my case unneeded) SCSI drivers and math emulation might help, but
the problem seems to be more than limited memory. Shift-ScrollLock
shows some free pages.

I haven't yet tried this kernel on my larger machine, but it looks
like something took a serious performance hit - maybe something that
isn't easily noticeable on faster machines, like an interrupt timing
problem?

Is it possible that something in the new system start-up code has
switched the chip down to slow speed, after all?  I don't have a turbo
LED wired in either..

Does anyone else notice a significant drop in speed?

Thanks to everyone for their work on this wonderful system!
--
Dan.