A team of boffins has managed to create a protocol which can carry out 7.8 million MPI tasks on 1,966,080 cores of the Sequoia Blue Gene/Q supercomputer system.
According to the research spec this means that the protocol can manage to process 33 trillion events in 65 seconds which is hugely faster than the 504 billion events/second carried out in the past.
Researchers Peter Barnes, Christopher Carothers, David R. Jefferson and Justin M. Lapre have dubbed their protocol Time Warp and it is a parallel discrete-event simulation synchronization that automatically uncovers the available parallelism through its error detection and rollback recovery mechanism.
They noticed a 97 times performance improvement when scaling from 32,768 to 1,966,080 cores thanks to cache performance improvements when running at peak scale.
The system uses a new, long range performance metric, called Warp Speed, which grows linearly with an exponential increase in the PHOLD event-rate.
In true Star Trek style, the machine has warp speeds. Currently it is at Warp Speed 2.7 and it will be nearly 150 years before we expect to reach Warp Speed 10.0.
The paper on Warp Speed can be found by a jump to the left and then a step to the right here.