Top Supercomputers- India

January2021

Following is the ranking of the systems in terms of Rmax (Linpack Benchmark performance).

C - CPUs, G-GPUs, ICO - Intel Co-processors. Cores (4th column) that are not qualified with C/G/ICO refer to CPUs.
OEM - Original Equipment Manufacturer, SI - System Integrator.

Rank	Site	System	Core/Processor/Socket/Nodes	Rmax (TFlops)	Rpeak (TFlops)
21	Indian Institute of Technology,Kanpur	HPC Cluster based on Cluster Platform HP SL230s Gen8 nodes in SL6500 chassis, 768 nodes, 15360 cores, with total memory of 98 TB, RHEL 6.5 operating system, 500 TB total storage with DDN based Lustre file system, nodes connected by FDR Infiniband interconnect.Cluster Platform SL230s Gen8, Intel Xeon E5-2670v2 10C 2.5 GHz, Infiniband FDR. More Info OEM:HP, Bidder:HP	15360/2/Dual Intel Xeon E5-2670v2 ten-core processor at 2.5 GHz with Redhat Enterprise Linux 6.5 OS, 128 GB RAM.	295.25	307.2
22	Inter-University Centre for Astronomy and Astrophysics (IUCAA),Pune	Twenty Four numbers of Apollo 2000 Chassis, each having 4 * ProLiant XL170r Gen10 compute node with dual CPUs (Intel(R) Xeon(R) Gold 6248 CPU @ 2.50GHz, 20 Cores ) running Scientific Linux Operating System Version 7.5Intel(R) Xeon(R) Gold 6248 CPU @ 2.50GHz, 20 Cores ) running Scientific Linux Operating System Version 7.5, and connected by 50 G ETH. More Info OEM:HPE, Bidder:Concept	3840//	192.5	307.2
23	Vikram Sarabhai Space Centre (VSSC), Indian Space Research Organization (ISRO),Thiruvananthapuram	The SAGA Cluster has HP SL390 G7 servers and NVIDIA GPUs. It is a heterogeneous system consisting of 320 nodes with Intel Xeon E5530 and Intel Xeon E5645 CPUs, and C2070 and M2090 GPUs. The system has 153600 CUDA Cores on NVIDIA Tesla M2090 GPGPU and 1416 Intel Xeon Westmere-EP cores alongwith WIPRO servers (202 nos) of Intel Xeon Westmere and Nehalem cores, each with 2 numbers of NVIDIA GPGPU C2070 per server.dual Intel Xeon E5530 quad core and dual Intel Xeon E5645 hexa core CPUs, and dual 448-core NVIDIA C2070 and dual 512-core M2090 GPUs) w 40 Gbps Infiniband network More Info OEM:Wipro, Bidder:HP	3100C + 301824G/3/The system has three kinds of nodes: 1. 185 nodes (WIPRO Z24XX(ii) model Servers) consist of dual quad-core Intel Xeon E5530 CPU @ 2.4 GHz, and dual C2070 GPUs. 2. 17 nodes (WIPRO Z24XX(ii) model Servers) consist of dual hexa-core Intel Xeon E5645 CPU @ 2.4 GHz, and dual C2070 GPUs. 3. 118 nodes (HP SL390 G7 servers) consist of dual hexa-core Intel Xeon E5645 CPU @ 2.4 GHz, and dual M2090 GPUs. All nodes has 24 GB memory each.	188.7	394.76
24	Oil and Natural Gas Corporation (ONGC),Dehradun	Fujitsu Cluster, 260 nodes, 6240 cores, RHEL-6.5 OS, with Infiniband interconnect.(Intel Xeon E5-2670v3 @ 2.3 GHz dual twelve-core processor nodes) w/Infiniband More Info OEM:Fujitsu, Bidder:Wipro	6240/2/Dual Intel Xeon E5-2670v3 twelve-core processor at 2.3 GHz with RHEL-6.5 OS, 262 GB RAM.	182.09	229.63
25	MRPU,Chennai	ANTYA HPC cluster consists of Acer hardware servers having 236 CPU-only nodes, 22 GPU nodes, 2 High Memory nodes and 1 Visualization node with Red Hat Enterprise Linux OS, and connected by 100Gbps Enhanced Data Rate (EDR) InfiniBand (IB) network. Total storage of 2 PB using GPFS file system. \|2 x Ivy Bridge 12C E5-2697 V2@ 2.7GHz More Info OEM:Boston, Bidder:Locuz	9600//400	175	207
26	National Institute of Science Education and Research(NISER),Bhubaneswar	A HPC of 64 new compute nodes, 32 old compute nodes and 01 GPU node with 04 nos. of NVIDIA Tesla K40c has been named KALINGA for the School of Physical SciencesIt has 205 TB of storage with Luster Parallel File System and interconnected with 216-Port EDR 100Gb/s Infiniband Smart Director Switch populated with 108 ports.,of 64 new compute nodes, 32 old compute nodes and 01 GPU node with 04 nos. of NVIDIA Tesla K40c has been named KALINGA for the School of Physical SciencesIt has 205 TB of storage with Luster Parallel File System and interconnected with 216-Port EDR 100Gb/s Infiniband Smart Director Switch populated with 108 ports., More Info OEM:Netweb, Bidder:	3616//64 New Compute nodes, 32 old compute nodes and 1 GPU Node	161.42	249.37
27	Indian Institute of Technology(IIT),Jammu	The Supercomputer Agastya is based on a hybrid configuration of Intel Xeon Cascade Lake Gold 6248 20C, 2.5GHz Processors and NVIDIA Tesla V100. The system was designed with total peak computing capacity of 256 TFLOPS on CPU and 56 TFLOPS on GPU performance. The system uses the Lustre parallel file system of 795TB usable storage with 30 GB/Sec write throughput. The Internetworking is done via High Speed Low Latency Infiniband Architecture Mellanox HDR100 100Gbps using fat-tree topology configured as a 100% non-blocking network.is based on a hybrid configuration of Intel Xeon Cascade Lake Gold 6248 20C, 2.5GHz Processors and NVIDIA Tesla V100. The system was designed with total peak computing capacity of 256 TFLOPS on CPU and 56 TFLOPS on GPU performance. The system uses the Lustre parallel file system of 795TB usable storage with 30 GB/Sec write throughput. The Internetworking is done via High Speed Low Latency Infiniband Architecture Mellanox HDR100 100Gbps using fat-tree topology configured as a 100% non-blocking network More Info OEM:Netweb, Bidder:	3200//	161	256
28	Indian Institute of Science Education and Research (IISER),Thiruvananthapuram	HPE Apollo K-6000 chassis with 88 ProLiant XL230K Gen 10 compute nodes each with dual Intel Xeon Gold 6132 processors (Total 2464 cores at 2.6 Ghz), 3 ProLiant DL380 Gen 10 GPU nodes (Nvidia P100, 4 cards total), Intel omnipath infiniband 100 Gb, 500 TB Lustre FS DDN storageTotal 2464 cores at 2.6 Ghz), 3 ProLiant DL380 Gen 10 GPU nodes (Nvidia P100, 4 cards total), Intel omnipath infiniband 100 Gb, 500 TB Lustre FS DDN storage More Info OEM:HPE, Bidder:SS Information Systems Pvt Ltd .	2464//88compute nodes and total 2464cores .	141.3	205
29	Tata Consultancy Services,Pune	The EKA supercomputer is a Linux based cluster which uses most of the proven opensource cluster tools. EKA uses C7000 enclosures with 16 nos of BL460c blades populated and connected through 4X DDR Infiniband interconnect, this uses most efficient routing algorithms developed by CRL. HP Cluster Platform 3000 BL460c, 1800 nodes, 14400 cores, 29.5 TB total memory, 80 TB total storage based on HP SFS (Lustre based) storage system with 5.2 GB per second throughput, RedHat Linux operating system.Dual Intel Xeon 3 GHz quad core E5365 (Clovertown) w/Infiniband 4X DDR More Info OEM:HP, Bidder:HP	14400/2/dual Intel Xeon quad core E5365(Clovertown) (3 GHz, 1333 MHz FSB) processor sockets, 16 GB RAM, 73 GB	132.8	172.6
30	Indian Institute of Technology,Guwahati	Param-ishan- 126 compute nodes with dual Intel Xeon E5-2680 v3, 12 Core, 2.5 GHz; 16 compute nodes with dual Intel Xeon E5-2680 v3, 12 Core, 2.5 GHz and dual Nvidia K40 Card; 16 compute nodes with Intel Xeon E5-2680 v3, 12 Core, 2.5 GHz and dual Xeon Phi 7120p card; 500TB PFS from DDN; Brigh cluster Manager, Slurm; HPL submitted was run on all 162 nodes using cpu cores only. OS: CentOS 6.6, Storage using Lustre PFS DDN SFA-77003888 More Info OEM:Lenovo, Bidder:HCL	//162	122.71	155.52

,