Scor­pio News


July–September 1988 – Volume 2. Issue 3.

Page 32 of 39

Initially I also cross-connected line 12 to line 14. This did not work, since both drives were selected simultaneously, so I used a DVM to find which of the lines 12, 14 or 16 became active when B: was selected, and I found that line 14 (line 12 ex main PCB) was the one. This line was therefore used, but I was left rather confused, since this should have been drive select 1. Without a fair amount of dis-assembly of the machine, I could not see what drive select link was in place on the floppy in the machine, but I suspect it was drive select 1! If this was the case, then line 14 (drive select 2) must have been being asserted when drive 1 was required, and drive select 1 for drive 21!

The addition of this drive made things much easier. It would also have been possible to have connected say a FD55F, or a FD35F (with suitable connectors), for 720 Kb format.

The Norton ‘SI’ – System Information program gave the following results when tried on three PC systems, all running PCDOS 3.2 :–

1)IBMPC XT(4.77 MHz 8088)1.0
2)IBM PS2/50(10 MHz 80286)10.0 (Wait states ?)
3)Gemini AT(12 MHz 80286)13.5 (No wait states)

the above figures are averages over several runs with ‘SI’. The Gemini on occasions gave figures as high as 15.9, but I suspect that the results may get more variable as speeds increase. I do not know how accurate this program is, and it is obviously only testing one aspect of the system, but it is indicative of the relative machine performance, excluding disk I/O. It is supposed to indicate how much faster a system is than a bog-standard ‘PC’, and the first result above verifies this.


It is difficult to criticise the Gemini AT in respect of the standard of construction and performance. It appears to be capable of carrying out its tasks quickly and without problems.

It was let down initially by absence of Documentation and Support Software, non-standard memory-mapping, and lack of support for user expansion of the Floppy Drives, but these aspects are easily remedied. Fundamental design errors in the hardware are a different matter, and are not easily overcome, but the Gemini ‘AT’ did not appear to suffer in this respect.

Provided Gemini supply adequate documentation and support software, the machine should be capable of providing the standard of performance and reliability needed by more demanding enterprises, and it can be recommended.

Page 32 of 39