>I recently bought 2 Sony HB-F700D MSX2 machines. They seemed to be
>totally the same, but after trying them, one of the machines seems to
>have a diskdrive which is really very much slower than the drive in the
>other machine. After a while I found out that the fast drive-machine has
>Disk Basic version 1.2 and the slow one 1.0. But I think there's been no
>'surgery' on these machines to replace the diskROM, and the machine with
>1.2 has a lower serial-number than the 1.0-one! Isn't that strange?
Maybe the 'surgery' was done without leaving traces? For example, some disk
ROMs are put in an EPROM and can be replaced without visible physical
changes in the machine.
>>> 30 PRINT TIME/50 : REM 60 for NTSC mode
>>>
>>> Can other people please check this on their MSX too? I'm wondering...
>>> (IF this is a valid way to measure the disk-speed ofcourse!)
>>
>> No it isn't!
>> The Disk routines turns of the interrupts while reading/writing to disk.
>> So it could be possible that you have one interupt registration when the
>> next EI is done while in reality there could be already two or tree
>> 'interupt times' passed (But since the interrupt line was still high
>> from the first one they went by unnoticed)
>
> But the slow drive gave 2.3 while the fast drive gave 0.23... Sounds
> reasonable, doesn't it?
David is right, you only measure the number of interrupts that got through.
For example, take two equally fast disk ROMs, ROM A re-enables interrupts
whenever possible and ROM B keeps interrupts disabled during the whole disk
I/O. According to your test, ROM B would be much faster, but in fact it isn't.
Just because you got the right results doesn't mean you did the right test...
A better test would be:
10 get time t$:print t$
20 for i=0 to 31:bload "file"+hex$(i)+".bin":next i
30 get time t$:print t$
I deliberately used 32 seperate files because that way caching won't give
an unfair advantage. If the directory and FAT are cached, the test will run
faster, but that is only fair because caching does actually make loading
faster. But if you load the same file 32 times, a caching disk ROM will
gain an advantage that is no longer a reflection of it's real speed.
Most disk ROM don't cache files however, so you can probably use the same
file every time.
Another good test: use a sector copier (one that uses the disk ROM, so
don't choose FastCopy!) and take a stopwatch to time how long a disk copy
takes.
Bye,
Maarten
****
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/)
****
