The 11th list of the Top50 most powerful supercomputing systems in Russia has been released. The list itself is in Russian (I presume; it’s at least in a language I can’t read that looks Russian), but Russian HPC manufacturer T-Platforms has issued their customary release which gives us a view into what’s happened in the six months since the last release (they aren’t carrying it on their website, yet, but HPCwire is). The list is maintained by the Research Computing Center of Lomonosov Moscow State University and the RAS Joint Supercomputer Center.
The big message for me is that there hasn’t been much change
The top twenty of the list has remained unchanged. Like in the previous edition, the list is topped by MVS-100K supercomputer installed at the RAS Joint Supercomputer Center with Linpack-measured performance of 71,28 Tflops. The second position is occupied by SKIF MSU Chebyshev supercomputer based on blade systems developed by T-Platforms company. The supercomputer is installed at the Research Computing Center of Lomonosov Moscow State University and has Linpack-measured performance of 47,3 Tflops (computing efficiency – 78,9 percent). Since deployment in 2008 the supercomputer capacities have been fully loaded with computational jobs.
The 11th edition of the list has shown the minimal performance growth rate since the rating foundation. The total actual performance of the systems has grown from 382.6 Tflops to 387.1 Tflops within half year. The rating has updated by 10 percent – 5 systems out of 50 are new or upgraded.
As one of many possible contrasts with the list maintained by the Top500, the threshold for entry to the Top500 list was about 17 TFLOPS in June. 978 Gflops gets you onto the Russian list.
The shares of processor vendors have remained unchanged in the 11th edition of the rating: Intel – 37 systems, AMD – 7, IBM – 5 and Hewlett Packard – 1 system.