in the 386 era, there were several brands of ide controllers: lg, umc, winbond, hmc, ali and some others that i can't remember now. has anyone tested whether they all perform similarly, or which one is faster?
IDE of this vintage doesn't really go through the controller.
The controller at most modifies the I/O addresses and IRQ that the hard drive responds to, but otherwise just directly connects the hard drive to the ISA bus.
Data transfers are done via Programmed I/O directly with the hard drive itself, not through the controller. This changes in the late 486 and Pentium ages, where the IDE controllers are actually PCI devices and can perform PCI DMA transfers into memory.
So the controller really doesn't matter on a 386.
Is is accurate that a ~2.1MB/s max sustained transfer rate is about the best one can expect with IDE PIO over 16 bit ISA?
I did quite a bit of benchmarking a few months ago and this was about the best I could realize with the machines, controllers and drives I had.
As an aside I also explored SCSI over ISA and could achieve just a bit more... about 2.4MB/s.
Before 1994 those were not controllers, but adaptors. The controller was on the disk and the adaptor was a simple device to connect a disk to the ISA bus. The chip plays no role in performance.
There was actually a handful of ISA EIDE controllers, but they are very rare and they won't get any faster.
Theoretical limit of PIO mode 0 is 3.3 MB/s
In practice, 2+ MB/s is perfectly OK
Żywotwór planetarny, jego gnijące błoto, jest świtem egzystencji, fazą wstępną, i wyłoni się z krwawych ciastomózgowych miedź miłująca...
Never made such testing. But once i experimented with ISA bus overclock on UMC MB457. The only IDE card that was stable on 25MHz was PIC 4020.
Grzyb wrote on 2020-06-13, 18:52:Theoretical limit of PIO mode 0 is 3.3 MB/s
In practice, 2+ MB/s is perfectly OK
Especially when you consider that the CPU is doing the work in PIO mode and my be a limiting factor .
BreakPoint wrote on 2020-06-13, 18:57:Never made such testing. But once i experimented with ISA bus overclock on UMC MB457. The only IDE card that was stable on 25MHz was PIC 4020.
Is that a typo ? I cannot find anything about a "PIC 4020" IDE controller ?
PIC P4020
BreakPoint wrote on 2020-06-13, 19:05:PIC P4020
Thank you. So it is a "PIC P4020" . This one definitely exists. Too bad there is not much documentation about it .
EDIT : Finding info about "PIC" is definitely not helped by all the stuff on the Internet about "PIC microcontrollers".
