CompuServe Messages

#SCSI Problem.

    31-Jan-94 10:01:33
Sb: #73443-#SCSI Problem.
Fm: BILL LEACH 71330,2621
To: Trent Hawkins 73614,533
Trent; Beware! There is nothing wrong with Amy's SCSI-I reselection. It is the drives that were in error. Thus, if you connect a SCSI-I drive to your system, the potential is there for you to have reselect problems. BTW, you example is one where you problem would not notice the difference between reselection on and reselection off. AP does not hit the drive for large transfers in each access. I am not certain but AP may only be requesting or writing a few block at a time. There are lots of parameters associated with the Amy's SCSI system that will affect performance. Most of them are set for what is generally optimal for reselection off, multiple drives present, moderately heavy system load (multitasking) and moderately heavy DMA in use. Under such setting as are "standard", setting reselection probably results in differences that are only measurable but not perceivable. Also, the faster the drive, the less difference will actually exist and of course SCSI-II with SCSI-II drives will have a significantly smaller difference. Though I am not recommending actually doing this unless you want to embark upon a few days (weeks?) of careful testing: You could increase the MaxTransfer and buffer settings dramatically. Then play around with copying very large files (set copy's blocking size to some attainable high value also). Using activity of AP will not be valid for testing because AP appears to be specifically "tuned" to insure that it does not interfer with its own use of the serial port DMA. That is, it appears that Steve took the appoach that he should set his disk useage such that he would be able to keep up with the fastest and most demanding use of the serial port but no more. Or in other words, he keeps his demands for disk access to a small enough blocking factor to be "a good neighbor" in a multitasking system. -bill