I'm pretty sure only RZ1000 and CMD640 were bugged in a way you should avoid them. Winbond was generally very reliable.

Not sure about the W83769.

Since you mention the tape drive. It might be that the CMD640 data corruption bug won't affect you. Also many late BIOS versions for affected motherboards workaround the bug (by disabling read prefetch on flawed revisions of the chip). Also OS's newer than DOS have workarounds. This all cost performance but chances are you don't care too much about that.

mpe wrote on 2020-04-15, 09:42:

Since you mention the tape drive. It might be that the CMD640 data corruption bug won't affect you.

It will, especially when running any kind of backup software: https://www.mindprod.com/jgloss/eideflaw.html

On any mainboard with the CMD640 or RZ1000, you should really disable these and use an add-on card for IDE instead. I found the work-arounds in Win95 and updated BIOS'es to be non-working for me. I have a system here with the RZ1000 IDE controller on which I can cause data corruption by simply booting into Windows and going back to DOS. After that, the FAT is damaged, even though it has a BIOS revision that should fix all that. It doesn't. And this happens in Windows 95c as well as in Windows 3.1.

Thanks for the replies. S83769 sounds promising based on the replies. I'll have to test this out and report back.

The reason for the question is: I have a Travan 20-based SCSI tape drive on my Cyrix 5x86-133/4x system and after about 150 MB of written data (sometimes 200 MB), the tape gets chewed up. It has happened twice now. The pully's band seems OK to me and I'm not sure what the issue is. The tape cartridges are too costly now bother troubleshooting this issue. My solution was to put a Travan-20 based IDE tape drive in the system instead as the SCSI variants are harder to find.