Mark said:
Eric, thanks for the info.
http://www.siimage.com/products/storage.asp#semic
I saw that this particular chip is for SATA but my need is to
attach extra PATA drives.
Watch out for Silicon Image if you're using an nforce2 motherboard. I've
seen reports of problems and may be encountering one myself--when I finish
testing I intend to post the results.
I haven't seen any widespread reports of data corruption with Promise
products. Googling I find that most of the hits that actually associated
disk corruption with Promise controllers were discussing the changes that
need to be made to the Linux kernel to fix the bug in the kernel that was
causing the corruption. I believe I found 2 accounts of data corruption
with Promise controllers on any OS but Linux--one of those concluded that
he had a broken controller, while the other didn't say he had the problem
himself and when pressed couldn't provide a link to anybody else who had.
But you can google this yourself.
Meanwhile, _everything_ has problems with interrupt sharing--the Promise
boards are not unique in that regard--you'll see the fanboys tell you how
interrupts are designed to be shared, but the fact is that sharable
interrupts were a kluge developed around the time that EISA came into
existence whose only purpose was to provide a clumsy workaround for the
paucity of interrupts on the original PC and PC/AT, and interrupt sharing
has never been completely debugged, which is why, finally, some
motherboards are coming out that have more than 15 interrupts--if sharing
worked so all-fired well you have to ask yourself why Intel and Microsoft
devoted time and effort to developing the Advanced Programmable Interrupt
Controller (APIC) spec and why nvidia and the rest have been so quick to
implement it.
If someone tells you that he's sharing interrupts between two devices that
generate a high volume of traffic on an ongoing basis and he hasn't had
problems, append to his statement a "yet".