|6MHz,||1 wait state||–||5.7|
|6MHz,||0 wait states||–||7.0|
|8MHz,||1 wait state||–||7.7|
|8MHz,||0 wait states||–||9.2|
|10MHz,||1 wait state||–||10.2|
|10MHz,||0 wait states||–||11.5|
A program supplied with the system, SPEED, was used to switch between these.
You can also switch clock speeds directly from the keyboard, but not wait
states. The default power-on state is set via SETUP, which stores your choice
into battery backed memory.
In general I can see of no reason why you should want to switch clock speeds,
and normally leave the system at 10MHz. However, a few of the programs that I
have are games, and these are far too fast at 6MHz, let alone at 10! It is in
fact interesting to note that some of the genes don’t seen to alter, even when
you change the clock speed. Presumably they are using one of the programmable
timers that are a standard part of the AT spec.? As far as I am aware the only
time you HAVE to change clock speed is with some clever formatting and backup
programs that drive various chips directly, and have very specific and critical
timing loops in the code. As I don’t seem to have any of these programs this
doesn’t worry me.
I tried a variety of software on the system, and it all ran without problem,
with the exception of BASIC! By trial and error I discovered that BASIC was
fine at all speeds except 10MHz, 0 wait states. I started to wonder if BASIC
was one of these programs that falls over at high speeds – I couldn’t understand
how it could be for a moment, but that was how it looked. What would happen was
that a BASIC program that would run fine at 10MHz, 1 wait state (from now on
10/1, etc), would give various syntax errors at 10/0. I went with my system to
see Dave Hunt. These funny programs ran at 10/0 on his system – aha, a hardware
fault on mine! I then said “Look how they fall over on mine”. They didn’t!
Hmmm! Then several minutes later, at 10/0, they started to fail. A temperature
dependent fault, temporarily cured by a cold spell in the car? OK, let’s leave
the system on for a bit longer. Lo and behold, the same problem at 10/1.
Back to Elonex
I took the system back to Elonex, and they disappeared into a back room. When
they reappeared they said “changed some chips – try that”. I did, leaving it on
for some time with no problem. Then, running Norton, I noticed that it was back
at 10/1 again – they’d reset the link! I queried this, and was told that all
the initial systems that they had had had run at 10/0 OK. But then they had
started to get problems, so they were all now sold set to 1 wait state.
I pointed out that the advert said 0 wait states, and was told “yes it will run
with 0 wait states, but not at 10 MHz. We don’t say it will do both at the same
time”. For some reason I let this pass, but have subsequently decided that I
may raise this again. After all, the SETUP program allows 10/0 to be set as a
default, and the manual makes various references to it. In fact, I have
subsequently met someone else who has recently bought an Elonex system and he
has noticed a Norton of only 10.3, yet his SETUP, AS DELIVERED, shows it is set
to 10/0. Obviously the system is supposed to be capable of it, but I expect
you’ll see Elonex’s adverts changed in the near future.
At the time of writing, I don’t think you’ll find another 10MHz systems at the
same price, in fact I’m not aware of even an 8MHz system that’s within £100, and
so I suppose it can be argued that one shouldn’t quibble about the lack of this
last bit of possible performance. If you are looking for a system with that
extra edge, do check carefully. Some suppliers make no comment at all about
wait states, and more often than not this means that the system has then
permanently set. Of course, if you’re wealthy, there’s always the 12MHz systems
which have recently become available, and if you’re absolutely stinking rich you
can get an 80386 based system!