Anyone up to help test a firmware modification to gain ~30% speed improvement? This only applies to environments that use the firmware for data transfer (not Dave Lyons GS/OS driver). It would probably be safest on CFFA 2.0 cards and later, but you're welcome to try it out on an earlier card - you may want to ask Rich before doing so. I noticed that the block read and write transfer routines were very cautious with their error checking. The original PC-AT ST-506 and IDE interfaces were designed to transfer the entire sector worth of data before error checking. The CFFA checks for error after every word transfer. Nothing wrong with that, but will slow things down significantly if there is no fear of a hardware bug that could lose data, thus the CFFA 2.0 recommendation. The change involves a slight re-order of the instructions and doesn't change any of the addresses used inside the firmware so only the AUX ROM needs to be updated.
I have been testing this on my Apple II+ running Pascal 1.3 which is very disk intensive. See my post under CFFA Applications. This change is even better than an accelerator, as Pascal spends much time reading and writing to disk. I will test with my IIe enhanced as soon as I figure out all the different versions of firmware to build.
Dave...