Re: [hatari-devel] fdc test for real hardware

[ Thread Index | Date Index | More Archives ]

On 22/09/2013 21:49, Roger Burrows wrote:
On 22 Sep 2013 at 19:32, Nicolas Pomarède wrote:

I'm improving the timings for the floppy controler and I'd like to have
some results for the attached program on real hardware (STF or STE, not
TT or Falcon).

Could someone with a working STF/STE test this program and post the
results ? I need results from different machine to measure the relation
between RPM and index duration.

On my 1040STF:
Run 1:
   0000994d 000002fd
   0000994a 000002fd
   00009949 000002fd
   0000994a 000002fd
   00009949 000002fd
Run 2:
   0000994e 000002fc
   00009949 000002fd
   00009949 000002fc
   0000994a 000002fc
   0000994a 000002fc

On my 1040STe:
Run 1:
   00009acd 00000218
   00009afe 000001e7
   00009b31 000001b4
   00009b61 00000259
   00009abc 00000229
Run 2:
   00009b3e 0000027c
   00009a9d 00000248
   00009ace 00000216
   00009afe 000001e8
   00009b30 000001b4

P.S. The 1040STe is not used very often.

Thanks for your results ; are these pal or ntsc model ?

ragnar75 also sent me some ste results and like yours, they're much less stable than the stf ones :

00009a03    00000234
00009a45    000001f2
00009a89    000001ad
00009acd    0000023e
00009a3b    000001fb

I don't really understand why ; my test program set SR to 2700, so I don't think my waiting loop is delayed by some STE specific hw. This is as if the RPM was not constant as on STF. But anyway, 1 unit in the 2 measured values is 40 cycles, so in the end it makes only a very small rpm variation.


Mail converted by MHonArc 2.6.19+