ColdFire/68K Microcontrollers and ProcessorsのトピックRe: Coldfire V1 Math support

Coldfire V1 Math support

davepinnh — Tue, 03 Jun 2008 10:13:34 GMT

How do the JM64/128 perform a multiply?

I see from the manual there are optional MAC, DIV, and EMAC functions as part of the V1 solution. So does the JM64 or JM128 implement any of these hardware functions?

Im guessing not. So how many cycles for a 16x16 multiply on the JM128?

Re: Coldfire V1 Math support

kef — Wed, 04 Jun 2008 20:10:53 GMT

MUL.L and MUL.W instructions are present in MCF51QE. Should be present in JM too.

Re: Coldfire V1 Math support

davepinnh — Wed, 04 Jun 2008 20:37:26 GMT

OK, great. But without a EMAC or hardware assist, how many clock cycles to do a 16x16 multiply or a MULS.L in a V1.

Re: Coldfire V1 Math support

kef — Thu, 05 Jun 2008 00:19:35 GMT

OK, with TPM free running at least prescaler:

volatile long a=5;

volatile long b=5;

foo()

{

t = TPM1CNT;
   a = a*b; // 2 loads, muls.l and store
   a = a*b;
   a = a*b;
   a = a*b;
   a = a*b;
   a = a*b;
   a = a*b;
   a = a*b;
   a = a*b;
   a = a*b;
t = TPM1CNT-t;

}

after foo executed, I found 77 in the t. Results should be similar repeating the same with shorts