Former Russian President Dmitry Medvedev and the President of Moscow State University Victor Sadovnichy at the Lomonosov supercomputer presentation
“Lomonosov” supercomputer was installed at the Moscow State University in 2009. This supercomputer was created by Russian company “T-Platforms.” The system launch ceremony was attended by D.A. Medvedev, President of Russian Federation, who proposed to name the supercomputer after the great Russian XVIII century scientist.
2009 – First stage: designing, installation and commissioning of the base section of “Lomonosov”. The main computational supercomputer’s section consisted of 4160 dual-processor diskless compute nodes based on 4-core Intel Xeon 5570 processors, the second section included 260 dual-processor compute nodes with 4-core Intel Xeon X5570 processors and local hard drives. The total number of x86 processor cores was 35,360. In addition to x86 compute nodes, the supercomputer included 26 nodes based on PowerXCell8i accelerators. The total amount of memory was 56.5 TB, storage – 0.35 PB, and backup system capacity – 234 TB uncompressed. Supercomputer’s power consumption was 1.5 MW. At the time, its peak performance was estimated at 420 TFlops, and Linpack performance – 350 TFlops which resulted in a very good efficiency index of 83%. This performance level allowed “Lomonosov” to lead the list of the most powerful computers in CIS and Eastern Europe. It was ranked 12th in the global Top500 list in November, 2009.
2010 – The second stage in the supercomputer creation process. The system was supplemented with 640 diskless compute nodes based on TB2-XN computing platform and 40 compute nodes equipped with local HDD storage. Each of new compute nodes was equipped with 6-core Intel Xeon X5670 processor as CPU. The total amount of RAM increased to 79.92 TB, storage - to 1.75 PB. Supercomputer’s peak performance increased to 510 TFlops, and its Linpack performance was 397.1 TFlops. The efficiency was 77.8%. Its decline from preceding year’s level was caused by system’s heterogeneity, as compute nodes with different CPUs were used in the test.
2011 – The third stage: system expansion. Following the trend in the supercomputer industry, “Lomonosov” was additionally supplemented with 777 compute nodes equipped with GPU-based computing accelerators. As a hardware platform for nodes, TB2-TL solution was used where each node has two Intel Xeon E5630 CPUs and two NVIDIA X2070 computing accelerators. A total peak performance of the computer system was 1.37 PFlops, and Linpack performance – 674 TFlops. “Lomonosov” was ranked 13th in the June 2011 edition of Top500 list.
In June 2011, “Lomonosov” was included in the Graph500 list. According to tests results, the system ranked third (positions were allocated depending on the workload), but showed the best performance among all systems in the list. During tests, a result of 43,471,500,000 TEPS (Traversed Edges Per Second) was obtained using 8192 cores/4096 nodes based on Intel Xeon 5570 processors. Lately the system was ranked 2nd in the November 2011 list edition with 103,251,000,000 TEPS using 32,768 cores / 4,096 nodes based on Intel Xeon 5570 processors.
2012 – The fourth stage and yet another system expansion round. The supercomputer has been additionally equipped with 288 compute nodes with Intel® Xeon X5570/X5670 processors and GPU-based computing accelerators. Its total amount of memory has increased to 92 TB, and now computer consumes 2.8 MW. As a result of modernization, a peak performance of the computing system has been increased to 1.7 PFlops, and Linpack performance reached 901.9 TFlops.
Current rank in Top500 rating - #58 (November 2014).
Current rank in Top50 rating - #2 (April 2015).
«LOMONOSOV» KEY FEATURES
|Peak performance||1.7 PFlops|
|Linpack performance||0.9 PFlops|
|Main processor types||Intel® Xeon X5570 / X5670, NVIDIA X2070|
|Primary / Secondary compute nodes||T-Blade2 / T-Blade1.1, PeakCell S|
|X86 compute nodes||5 104|
|GPU compute nodes||1 065|
|PowerXCell compute nodes||30|
|X86 processors||12 346|
|X86 cores||52 168|
|GPU CUDA cores||954 240|
|Total RAM||92 TB|
|System / Service / Management Network||QDR Infiniband 4x / 10G Ethernet / Gigabit Ethernet|
|External storage||3-level storage:
|Operating system||Clustrx T-Platforms Edition|
|Total area (supercomputer)||252 m2|
|Total equipment weight||More than 75 tons.|
|Power consumption||2.6 MW|
1.7 PFlops «Lomonosov» supercomputer