Communication Is Important
•
So far the algorithm is “send-then-compute”
•
Try overlapping and computing
•
Use broadcast communication
B
C
P
a
P
b
P
e
P
f
P
i
P
j
P
m
P
n
P
c
P
d
P
g
P
h
P
k
P
l
P
o
P
p
P
a
P
b
P
e
P
f
P
i
P
j
P
m
P
n
P
c
P
d
P
g
P
h
P
k
P
l
P
o
P
p
+
x
1
1
x
2
2
x
3
3
x
k
k
+
+...
=
A
P
a
P
b
P
e
P
f
P
i
P
j
P
m
P
n
P
c
P
d
P
g
P
h
P
k
P
l
P
o
P
p