On Wed, 11 Nov 1998, Nestor Soriano wrote:
> ...
>
> >it is an almost perfect way, but I think it can be done faster. (each time
> >loading FFFF into a register, checking for FFFF, ...)
>
> LD HL,sectornumber
> INC HL
> LD A,H
> OR L
> JP Z,big_sector_number
> DEC HL
> JP small_sector_number
or even better:
LD HL,sectornumber
LD A,H
AND L
INC A
JP Z,big_sector_number
small_sector_number
Anyway, I think that we can use every cycle @3.58MHz. I still prefer a new
entry. Disk access times will become smaller and smaller.
>
> Is this so slow compared with the disk access time?
>
> >Perhaps we can add a new entryaddress(es) into the diskroms: DISKIO for
> >32bit sectors, ...
>
> This is also a good idea. In fact, when BIOS was updated for MSX2, new VRAM
> read and write routines for accessing 64K VRAM were added. Theoretically,
> old VRAM access routines can only access first 16K VRAM. This is the same
> case, then.
indeed!!!
>
> >If DOS3 will read/write a large drive it should use this routine instead
> >of #4010. (normal DISKIO)
>
> Use of #4010 allows DOS to be hardware independent. But then sector numbers
> are limited to 16 bit. How to solve this? With drivers?
What problem do you mean?
In general #4010 is only called by DOS kernel. In all other cases when
your own program makes use of #4010, you know what size the disk you want
to access has. So if it is >32MB, use the 'new diskio' otherwise use
#4010.
Remark: you can always read the bootsector of a disk with #4010. Useful
to determine its size. This can also be done by DOS3 kernel.
(Like I've mentioned before: current DOS2 fetches the #4010 address from a
kind of diskparametertable. (I'm not 100% sure of this))
>
> >BTW. I was wondering: what has FAT16 to do with partitions >32MB. Can't
> >you use FAT12? The clustersize will be large though.
>
> Cluster size in sectors must be a power of 2. So, maximum sectors per
> cluster is 128; using 4096 clusters we get a maximum partition number of
> 256Mb.
>
> But hey, you really think that 64Kb clusters are practical? If we can
> modify DOS in order to use 32bit sector numbers, let's do it complete and
> implement also FAT16!!
I totally agree. I only wanted to be sure that FAT16 is theoretically not
necessary for >32MB partitions.
Like some people said before; 24bit sectornumber will be enough to fit all our
needs for the coming decades ;-) (8GB partitions)
>
> >On the other hand I think FAT16 will be faster, because you can manipulate
> >16bits instead of 1byte and a half. Anyway that's something for DOS3.
>
> Well, this is an extra advantage. But I think that such speed problems are
> not critical; remember that disk access times are much bigger that any
> cluster calculation time.
any gain of speed is welcome
jon
('first think and then do')
****
MSX Mailinglist. To unsubscribe, send an email to [EMAIL PROTECTED] and put
in the body (not subject) "unsubscribe msx [EMAIL PROTECTED]" (without the
quotes :-) Problems? contact [EMAIL PROTECTED] (www.stack.nl/~wiebe/mailinglist/)
****
