Rockbox mail archive
Subject: Re: ATA test, round 2
From: Jens Arnold (arnold-j_at_t-online.de)
> I'm very glad that my case is no longer a problem for the
> speed optimizations!
This sounds very good indeed!
> I tried the 1-Feb-2004 build again, just to recheck that the
> problem really was there, and as expected, i got a corrupt
> .rockbox folder...
> Only with
> [Recent ucl] -> [ROLO] -> [Feb build] -> [Restart w/ ON+F1] ->
> [Archos firmware] -> [Feb build] did i get the old problem
It seems that Rockbox now does some initialization that it
didn't (properly) back in February, and that the Archos fw also
does not do. So I ask you if you could check two more
1) Please check if you get still no problems if you use the fast
build with "Disk Poweroff" set to "Yes".
2) Put ajbrec-fast.ajz into the root and rename it to
ajbrec.ajz. After disconnecting usb, switch off the box.
Then boot into the fast build via Archos fw with F1+On.
Check if it still works fine.
> By the way, it may be just me, but if believe that compared to
> the old optimizations, these recent ones are way faster. Maybe
> it's just the feeling of handling a 4 month old build, but i
> remember that the old speed optimizations, when they were
> released, didn't made a big diference, but with this new
> build, i feel that browsing and skiping tracks is faster, even
> compared to the rather recent ucl i already had in the box.
I think you're right on this one. The first asm optimized
disk reading routine (by Jörg) was on average ~30% faster than
the C version. The current one (disabled in the official builds)
is on average ~80% faster than the C routines.
In addition, the disk writing is now optimized also. Even the
official builds use an optimized C writing loop for some time,
which is more than twice as fast as the February version. The
asm optimized version enhances the speed by another factor of 2,
bringing it on par with the reading.
Again, thanks for all the testing!
Page was last modified "Jan 10 2012" The Rockbox Crew