In order to get 30 MB/s out of ISA you would need to run it at ~30 MHz (default is ~ 8 MHz). 7 MB/s is about right unless you've overclocked it a lot.

MHDD scanned the drive at 30MB/s and later tonight at 26MB/s. The drive it self formated extremely fast and windows 95 installed also extremely fast.
So my only assumption is that it is running at 25MB ~ 30MB speeds using Tekram DC-300B.

I did some research and it seems that overclocking the ISA bus is possible, where some boards had limiters that were adjustable and others didn't
Must be that this board does it automatically?

I found an program TweakBIOS so i will try to do tests with that, as well as with disabled cache on Tekram DC-300B.

Stojke wrote:

MHDD scanned the drive at 30MB/s and later tonight at 26MB/s.

Mhdd is not any way a benchmark program.
It count not transfer itself, but time between reading two nearest cluster, and since you controller is cached it can fake this seek time and report a smaller value.

Stojke wrote:

The drive it self formated extremely fast

Yes, this is how cache controller works, all writes is cached, all verify is retry data momently, so format is performed very fast.

Stojke wrote:

and windows 95 installed also extremely fast.

Yes, this is how cache controller works. Widows copy a bunch of small files, usually it means disk should read fat, write file, modify fat and again. This leads to lot work for disk heads and slowdown process greatly, so actual transfer rate is dropped to tens or hundred kb/s. And cache controller can hold all fat-changes and write only once in a few seconds, that mean it write mostly linear and transfer speed is limited only to bus speed.

Stojke wrote:

So my only assumption is that it is running at 25MB ~ 30MB speeds using Tekram DC-300B.

No, this is can't be true.

Stojke wrote:

I did some research and it seems that overclocking the ISA bus is possible

Not that fast. Usually you can try to run ISA on 10 or 12 mhz, but not every isa card can work on that speed. So 16mhz I assume as a theoretical limit, I think only specially designed card can work on that speed, bun none of massive developed cards.

Than that explains it all. Than again, isn't this a desired effect, even If its not pure speed?

Stojke wrote:

isn't this a desired effect, even If its not pure speed?

Of course. I've had an isa cache controller tekram dc-600 back in days, that I've used from early 486 up to later Pentium mobos.
It has only about 5 mb/s linear transfer bottleneck, but due a cache most operations in windows 9x and os/2 is still faster, than on onboard ide, with ~20 mb/s linear speed.

Cool, thanks for the explanation 😀
I really like VLB SCSI controllers. If I ever get an Cyrix 5x86 cpu I might build a system with one.