Swiego wrote on 2020-05-02, 09:29:

As a matter of intellectual curiosity, is there a max performance anyone has seen from disk I/o through a16 bit ISA slot?

Tweaking some settings and replacing the 15k SCSI disk (some HP branded enterprise drive) with a Maxtor Atlas 10k.4 72gb got me up to ~2.4MB/sec which is the best I’ve seen so far.

This should be of interest : http://www.vcfed.org/forum/showthread.php?564 … d-transfer-rate

TLDR : About 5.33 MegaBytes per second shared between all ISA peripherals seems to be the maximum (for an 8 MHz ISA bus)

Interesting thread. A lot of good evidence for why 5.33MB/s is a theoretical limit but the practical limit is closer to ~2.6MB/s for a ~60MHz system if I read the thread properly.

I realized I can up the adapter speed to 10MB/s (actually makes a 10% difference) and this gets me to a reliable 2.4MB/s with the Atlas 10k.4. At least this gives me a benchmark to target when exploring the Solid state options later. This is in DOS with aspi4dos.sys; it drops to 1.6MB/s without the driver, and is roughly the same in every W98 benchmark I’ve tried.

The setup with a 1542CF and the Atlas drive

Not bad for ISA I suppose...

I still wonder what Compaq was thinking when they designed a Pentium based machine with only an ISA bus at a time when PCI was coming onto the market . They could have at least gone with VLB . EISA would have been option too .

As I understand it, Compaq was a late adopter of Intel’s PCI standard and seems to have only begrudgingly adopted it. Even when they finally did, it was half-heartedly with EISA still receiving equal billing. It is kind of silly (but also unusual) to have an ISA-only Pentium.... there can’t be many of them.

It is an interesting curiosity .