
24
CHAPTER
1 Introduction
10,000
1,000
t
100
10
0.1
F
• MPP
peak
■
CRAY
peak
h
F
F
F
-
-
=
Xmp/416(4)
I
CM-200
·
CM-2
· y
ASCI
Redj
Paragon
XP/S MP
yS
(6768l·/
Paragon
XP/S
MP yr
(1024)·^
CM-5
·
s*TiD
yS
T932(32)
■
./Paragon
WIS^"*^"^
/ ·,^---
/>^^B
C90(16)
""Delta
^^-^~\
^1pSC/860
Ymp/832(8)
"
nCUBE/2(1024)
I
I
I
I I
1985
1987
1989
1991
1993 1995
1996
FIGURE 1.11 Performance
of
supercomputers and MPPs on the UNPACK peak performance
benchmark. Peak performance
in
GFLOPS
for
solving dense linear equations is shown
for
the leading
CRAY multiprocessor vector supercomputer and the fastest MPP systems. Note the change in scale from
Figure 1.10 (MFLOP ...