In its evolution, the computer industry has reached the point when con
siderable computing power can be packaged on a single microprocessor c
hip. At the same time, costs of designing a computer system around suc
h a CPU are growing. For these reasons we decided to explore a possibi
lity of using commercially available symmetric multiprocessors (SMP) a
s building blocks for the LQCD computer. Careful analysis of the archi
tecture allowed us to build a QCD primitive library running close to t
he peak performance on the UltraSPARC processor. As a result, multithr
eaded QCD code (both the heatbath and the Wilson fermion inverter) run
s at about 50% efficiency on a single SMP. The communication between d
ifferent CPUs is handled by a coherent memory system. Currently we are
planning to connect several SMPs with a high bandwidth network into a
single system.