Twospruces
Veteran Member
I have been spending some time modifying the TRS-80 model 100 to run at 2x clock IE 4.9152MHz.
The production 80C85 specification clearly does not support such operation.
And yet, it works well, under some conditions
* you need to use fast RAM
* you need to use fast ROM
.. and here fast means <200nsec. I use 120nsec SRAM and 150 nsec EPROM.
It occurs to me that, if one were writing specs for a processor, that they must have assumed something about the responsiveness of memory.
If Intel assumed "typical or slow" memory, then that would clearly reduce max speed, whereas in the timing budget they could assume that memory was fast, perhaps that supports a higher clock rate.
Is there some rule of thumb around this? What do processor vendors assume about memory speed when they are rating their devices?
I've been experimenting with different speed ratings on processors too. It is clear that processors with faster ratings can sustain higher speeds (As expected) but they all seem to run a fair bit faster than their ratings suggest.
thx
The production 80C85 specification clearly does not support such operation.
And yet, it works well, under some conditions
* you need to use fast RAM
* you need to use fast ROM
.. and here fast means <200nsec. I use 120nsec SRAM and 150 nsec EPROM.
It occurs to me that, if one were writing specs for a processor, that they must have assumed something about the responsiveness of memory.
If Intel assumed "typical or slow" memory, then that would clearly reduce max speed, whereas in the timing budget they could assume that memory was fast, perhaps that supports a higher clock rate.
Is there some rule of thumb around this? What do processor vendors assume about memory speed when they are rating their devices?
I've been experimenting with different speed ratings on processors too. It is clear that processors with faster ratings can sustain higher speeds (As expected) but they all seem to run a fair bit faster than their ratings suggest.
thx