The node processor on the iPSC/Z is a four MIP Intel 80386 processor which is four times faster than the 80286 of the iPSC/l. The iPSCI2 has Intel 80387 numeric co-wocessors. which run at about 250 Kflops. This is about five timefaster than the 80287, which ii the co-orocessor on the iPSC/l. The iPSC/Z also su~oorts a Weitek 1167 ootion. which runs at’ about 600 Kflops in 64-bit arithmetic an’d more than a megaflbp in. 32-bit arithmetic. The iPSC/Z can also be configured with the same high speed vector extension board as the iPSC/l, which has peak performance capabilities of 6.6 Mflops in 64-bit arithmetic and 20 Mflops in 32-bit. While these asymptotic computing rates are the same for both vector machines, the overhead associated with vector startup is much less on the iPSC/Z due to the faster 80386.