Scor­pio News

at 8MHz. A Z80A CTC on a home brew board operates perfectly when the system clock is set to 4MHz. It seems very likely that Z80B peripherals will have sufficient margin to operate successfully at 8MHz. All the NE898 facilities tried to date have worked well. These are the CTC and PIO on board the SPCT, and the real time clock. The PIO is fully software and plug compatible with that on the GM811 and GM813 processors. Swapping from the Nascom only required a change of port addresses in the BIOS to allow the printer to operate. I have run the system with Gemini BIOSs V1, V2 and V3, and no problems have been experienced.

It is worth pointing out that in order to obtain standard RS232 baud rates a separate crystal clook is employed, and speed selection is by dip switch. All serial clocks are driven at the same frequency, producing the same baud rate except that each channel can be software set to /16 or /64. This is rather limiting if the serial interfaces are used for differing purposes. A split 1200/75 baud rate is just not possible.

A for way DIL switch is provided with the intention of providing boot options, although it would be possible to use these for other purposes. However, these switches are connected via the four most significant PIO lines on port A. If these lines are used the the switches should be left open, and the state of any external device connected to these lines is liable to alter the cold boot action. One option allowed by the Newburn Monitor is to boot the system from a serial port – this would permit the operation of a diskless configuration.

Performance

The Z180 gains in raw processing power over the original Z80 in several ways:

• Doubling the clock speed doubles the throughput
• Most instructions complete in fewer clock cycles giving up to 25% increase (Even the the NOP completes in three cycles instead of four – this must be of some use!) – however the Z80 used these extra clock cycles to perform memory refresh. The Z180 requires to insert additional refresh cycles, although on a less frequent basis.
• The new instructions, in particular the MLT instruction, can be used to good effect in new software
• Use of the other facilities in particular MMU and DMA can offer very worthwhile benefits

Some impression of the performance gains with pure Z80 code can be gauged from benchmark results, although these are often criticised for not bearing much relationship to real life processing requirements. I have run some of the tests listed in Paddy Coker’s articles in Scorpio News Issue 1, Volumes 1 & 2. The results are as follows: