26 Jan
2010
26 Jan
'10
10:44 a.m.
On 1/26/10, Dave McGuire <mcguire at neurotica.com> wrote:
On Jan 26, 2010, at 12:54 PM, Dave Caroline wrote:
> ...everyone at the time would mention the clock rate of the Z80
> claiming it was faster, they never notice the clock is divided, basic
> instruction rate was about 1 meg on the Z80.
Yep. I remember those arguments - fanboys would get excited because
their number was larger than your number, but ignore how much work got
done per clock tick.
One of the "worst" for that (the ratio of work-per-ticks, not fanboys)
was the 1802. If you executed a 3-cycle instruction (not uncommon),
it was 24 clock ticks. 1.7MHz was a typical number (half of NTSC
colorburst), so you were doing well to get to 0.1 MIPS. By
comparison, a 1MHz 6502 is a few times faster.
This same problem happens all the time
today...people assuming
that "more gigahertz" means "faster processor", even when it
doesn't. It's funny.
It's a number - easy to memorize and throw around, especially when you
don't bother to understand what's really going on in one tick.
I was a dyed-in-the-wool Z80 guy in my early
years... and
mostly ignored the 6502 due to the lack of registers.
I started with the 6502 and the 1802 - very different register models.
I do have to admit that more than once, I was frustrated by the 6502
only having three 8-bit registers and a non-orthogonal instruction set
(no TYX or TXY or PHX or PHY for example). The 65C02 made up for a
lot, but it came along late enough that it didn't help much.
ISTR that a 4MHz Z80 could match or outperform average code on a 1MHz
6502, but clever 6502 code could really rip.
-ethan