Group 64

Memorandum M-2661

Digital Computer Laboratory Massachusetts Institute of Technology

Cambridge

39.

Massachusetts

SUdJECT: WI Operation Tim8s

To:

S&EC

Group, Group 61,

Group 64

]'rom : ^H. Denman

Date: ]'ebruary 1, 1954

Page 1 of

3

Abstract: The operation times for the

WWI

instructions (except bo and bi) have been both calculated (based on the computer opera-

tion) and measured (using the real time c10qk). The results are given.

rrhe operation times have been measured for the WI Computer, wi th the following conditions:

10 It is assumed that the instruction is stored in MCM. and that if the address of the instruction refers to a storage address, this address is also in the MOM.

2. The times listed below indicate the number of pulses, from the fundamental computer pulse generator, required for each operation. This generator is supposed to operate at exactly 1 megacycle per second. If it does, these numbers are also the number of rnicros~conds required for

\..

each operation. However, this pulse generator is LC controlled, and therefore may not operate at exactly 1 megacycle. H~ncet the actual times for the operations may not be these numbors of microseconds. The measurements made during this test indicate that the error in assuming

the pulses to be exactly 1 microsecond apart was less than ~ of 1% at the time of the testo

3.

For In-Out instructions (si, rd~;; rc), it is assumed that the interlock is cleared and ^any delay counts associated with these orders are ignored. Times for bi a.nd bo are not given. as they are chiefly dependent on the In-Out equipment referred to by the previous s1 instruetion o

M-2661 ^{Page 2} These times have been calculated on the basis of the operation of the computer, and were verified by measurements using the real-time clock:

o

1 2

3 4

5

6 7 d

9 10 11

13 14 15 16 17

1-3

19

20 21 22

23

24

25

26

OPEHATIOl~

si

illegal inst.

bi 1"d bo 1"0

illega.l inst.

sb ts td ta ck ab ex cp sp ca cs ad su em sa ao d.m

mr } mh

dv

TI¥lE t in)ksec. (or clock pulses)

31

(see

3

above)

not meas. (see

3

^above)

16 "

^It

"

not rneas.

" "

^It

24

" ^{" "}

16

24-

32 32

24

32 32 16 16

24

24 24 24

.24

24 32 24

36-43

f.Pime for this oper~tion varies in accordance with the numhe;c of l's in the magni tude of the number in the AC when this instr. is given. The mini- mum time ()6.,M-sec) is for all 0 ^I S in fAG

I;

^the

maximum ( 43..)kSec ) for all 11 s ~n

lAO'.

^{The time}

is a linear ;functiop. of t1;le number of lis be- tween 1 and 15.rounded up, and therefore the average time for this iustr. ;):'01" random numbers

is 39 • .5~sec.

73

11.1 ... 2661 lruM3ER

27 28

29

)0 31

H.J:CWA:'Jjf

Page 3

OPERATION

_TI~ffU. in ;tsec. (or clock pulses)

:~: ^n: 1

^16-41

srr

J

sf 33-31

clh

Time for this op3ration varies in accordan\~

wi th the numher n in the adclresses of these instructions. n is taken modulo

32.

''che minimum time(16~sec.) is for n=O;the maxi- mum time(4~sec.) for n~3l. However, the

time is .not a linear function of n between these points, due to the occurrence of restoration pulses.

Tire for this instruction varies in accor- dance with the number of a's between the binary ~oint and the first 1 in the magni- tude of the bina.ry fraction represented by the number in the

Ae

and Eli.. If the mag- ni tude of the number in the AC and. BR is

~

i.

the numl:>er is scale fa·ctored 0 times t

gi ving the minumum time of' 3~ sec. for this operation. If the magnitude of the number in the AC is a and the 3R contains all O·s, the operation takes the maximum

time of al~ sec. (giving $cale factor of

33).

The time is not a linear function of n between these values, due to restoration pulses.

clc S9.me as shift orders above (16-LH~ sec.)

md 24

Signed

:_#._-_11. ___ ' _pO_~ ___

----..tV1'V _ _

H. H. Denman

Approved: __ ' __

~

________________ __

c.

^{W. Adams}