[Prev][Next][Index][Thread]
Re: 68040 clock doubling
>>>>> "John" == John de Bruin <j.bruin@auckland.ac.nz> writes:
In article <310F5CC3.5BCB@auckland.ac.nz> John de Bruin <j.bruin@auckland.ac.nz> writes:
John> I was reading in here about Motorola erroneously saying some
John> of its 040 chips run at 80MHz when their true speed is 40MHz
John> and ditto 33/66 MHz systems etc.
News propogation is funny. I haven't seen anyone claim that Motorola
has erroneously said anything, but just because I haven't seen it
doesn't mean it wasn't said. However, I said something similar, so
I'm going to clarify what I said. Here's my original statement:
"It's a little tricky, since some people call what you, I and
Motorola call a 40 MHz CPU a 80 MHz CPU and to them, a 33 MHz
becomes a 66 MHz CPU. At ARDI, we never use the inflated
numbers (even though it makes us look better), but when we say
25 MHz 68040, some people will think we're talking about a
theoretical 12.5 MHz CPU. Argh."
So my claim is that Motorola, myself and the person I was replying to
(Eric Bennett) would call the CPU that is in a Quadra 610 a 25 MHz
CPU. Eric's terminology was clear from his letter, I'm familiar with
my own terminology and I derived Motorola's terminology from page 12-1
of MC68040UM/AD "MC68040 32-bit microprocessor manual" (c) 1989.
Specifically, page 12-1 says:
"The following table provides ordering information pertaining to the
package type frequency, temperature and Motorola order number for the
MC68040.
Package Type Frequency Temperature Order Number
(MHz)
Pin Grid Array 25.0 TBD MC68040R25
R Suffix"
John> Why is this different from clock doubling as in the 486DX/2
John> CPU?
It is different because Motorola chose to call their original mc68040
CPU a 25 MHz CPU. I personally wouldn't care if they had chosen to
call it a 50 MHz CPU, in fact, it would make our own product look more
impressive, since we could say that a 75 MHz DX4 can run some Mac
programs as quickly as a 50 MHz 68040. Heck, let them call it a 500
MHz part and we'd then be able to say that a 75 MHz DX4 can run some
Mac programs as quickly as a 500 MHz 68040. However, it was
Motorola's choice, and I can understand their desire to call it what
they did.
John> I understand that most 68040s have two clock signals a PCLK
John> and a BCLK, the BCLK runs at 1/2 the speed of the PCLK.
John> And the BCLK is used for all bus timing, whereas the PCLK is
John> used for internal processor logic timing.
John> Now as far as I can tell the only difference between this
John> and the DX/2, is that the DX/2 only has one clock and uses a
John> clock 2x multiplier to derive the internal processor
John> speed. Why do the two clocks on the '040 _not_ have the same
John> effect as clock doubling?
I understand your point. If you read my original claim in context you
will see that my gripe is that when you have two different standards
for what N MHz means *for the same chip*, then things get confusing.
We choose to use the original meaning for two reasons:
a 25 MHz 68040 is roughly comparable to a 25 MHz 80486
we'd rather err on the side of people thinking Executor is
slower than it is and be pleasantly surprised than to use
the inflated number and have people be disappointed
John> Can the two run out of sync for instance?
John> Bye...
--Cliff
ctm@ardi.com
Follow-Ups:
References: