MULU & MULS accurate timings
Page 1 of 1

Author:  da_norf [ Tue Sep 09, 2008 10:16 am ]
Post subject:  MULU & MULS accurate timings

It appears to me that the instructions MULU & MULS are always weighted with the number of cycles corresponding to the worst possible case (70 cycles + n cycles for calculating the effective address) in v3.7.11 Beta.

According to M68000 User's Manual :

MULS, MULU — The multiply algorithm requires 38+2n clocks where n is defined as:
MULU: n = the number of ones in the <ea>
MULS: n=concatenate the <ea> with a zero as the LSB; n is the resultant number of 10
or 01 patterns in the 17-bit source; i.e., worst case happens when the source
is $5555.

For exemple in v3.7.11 Beta :
MULU #160,d1
"costs" 74 cycles whereas in real life it should "cost only" 38+2*2+4=46 cycles if I understand correctly the User's Manual.

Will we see these accurate timings in a next version ?

Author:  profkelly [ Wed Sep 10, 2008 1:03 pm ]
Post subject: 

From EASy68K Help.

"The cycle counter does not take into account pipeline efficiencies. The cycle number displayed tends to be a worst case value and is only intended for relative comparison purposes."

Author:  profkelly [ Thu Sep 11, 2008 3:20 am ]
Post subject: 

Sim68Kv3.9.1 will use the aforementioned 38+2n formula to calculate execution times.

Page 1 of 1 All times are UTC
Powered by phpBB® Forum Software © phpBB Group