Peter Berczik obtained record benchmark of sustained 25 Teraflop/s for a single five million body direct N-body simulation on 100 Tesla graphical processing units (GPU) on the GPU clusters of the Institute of Process Engineering IPE of Chinese Academy of Sciences in Beijing in March 2009. The IPE is operating of order 1000 GPU cards with a cumulated peak speed of 1 Petaflop/s. It's hardware is the prototype for similar clusters to be opened at other CAS institutions (see top news). See preliminary plot. March 2009 (this work is in collaboration with This email address is being protected from spambots. You need JavaScript enabled to view it.and This email address is being protected from spambots. You need JavaScript enabled to view it., and also partly shows results obtained from the kolob frontier kolob frontier GPU clustersee below.