harshakudaravally Posted February 1, 2013 Report Posted February 1, 2013 [b]1. High Performance Computer (HPC):[/b] A HPC to deliver at least 5.0 Teraflops performance, as per the attached specs, to run most of the common applications in a company, integrated with the other parts of the data center. [b]REQUIREMENTS OF HPC INFRASTRUCTURE [/b] Item 1: Compute Servers - 16 Nos At least 2*2.6 GHz Processor (8 Cores / 8 FLOPS per Processor), 2GB/Core memory ECC Memory; 2*500GB HDD, QDR IB HCA/ 10 GigE HCA, Dual GigE Ports; Memory per node to be scalable to 256GB in future. Item 2: Master Node: 1 No At least 2*2.6 GHz Processor (8 Cores / 8 FLOPS Per Processor), 4GB/Core Memory ECC Memory; 2TB usable HDD Space, QDR IB HCA/ 10 GigE HCA, Dual GigE Ports; Item 3: Compute Switch: Ports: 24 fixed SFP+ ports 10 GigE of Low latency or suitable IB switch Performance: 900-Gbps switching capacity Preferably latency less than 300 nanoseconds. Item 4: Management Switch: 24 Port Layer 2 GigE Switch. The switch proposed should connect to unified Switching fabric at company. Item 5: Rack - Industry standard Rack/ OEM Rack suitable for Quoted Server and Storage has to be proposed Item 6: Software: • Cent OS Operating System, Intel Compilers – 2 user license, C & Fortran Compiler, Math Libraries. • Vendor specific Infiniband/GigE stack on Linux OS, if available, should be supplied. • Vendor specific MPI implementation on Linux OS, if available, including MPI over IB and IpoIB if required, should be supplied. • Scheduler – Scheduler propsoed should be fully supported by bidder. Scheduler should be from Open grid Engine, PBS Pro and LSF only. • NO Open source, community edition cluster management tool to be proposed. • GUI Based Job submission portal • Users should only apply the jobs to nodes that have the required characteristics, therefore matching the best job to the best resource to get maximum performance. • All Application like Ansys CFX, Fluent, Ansys Mech, MD/MSC Nastran, LSDyna, Abacus, Hmmer, Gromacs, Fasta, Blast, Amber, Gaussian, ClusterlW, Altair Optistruc, Converge, Femfat, Altair Radioss to be integrated with the portal proposed. Hello guys above stated is the requirement, can some one help me out with this.. Very urgent requirement.
ChampakDas Posted February 1, 2013 Report Posted February 1, 2013 mathi poyindhi...[img]http://www.desigifs.com/sites/default/files/CroppedBrahmishock_zps694518cc.gif?1350460826[/img] help
cherlapalli_jailer Posted February 1, 2013 Report Posted February 1, 2013 requirement is not clear or may be not my cup of coffee?? R u looking for informatica or data storage??
harshakudaravally Posted February 1, 2013 Author Report Posted February 1, 2013 [quote name='cherlapalli_jailer' timestamp='1359747648' post='1303206168'] requirement is not clear or may be not my cup of coffee?? R u looking for informatica or data storage?? [/quote] Looking for a HPC cluster
PMREDDY19 Posted February 1, 2013 Report Posted February 1, 2013 Enti ba... e High Performance story vesaavu.. Architect role ki job specification.. interview question maadiriga undi.. Edina Order place chestaa.. quotes prepare cehyyamannara enti.. akkad Servers purchase cheyyali.. total : 16 +1 servers konni routers.. konni sitches.. inkaa evevo required softwares Item 1: Compute Servers - 16 Nos At least 2*2.6 GHz Processor (8 Cores / 8 FLOPS per Processor), 2GB/Core memory ECC Memory; 2*500GB HDD, QDR IB HCA/ 10 GigE HCA, Dual GigE Ports; Memory per node to be scalable to 256GB in future. Item 2: Master Node: 1 No At least 2*2.6 GHz Processor (8 Cores / 8 FLOPS Per Processor), 4GB/Core Memory ECC Memory; 2TB usable HDD Space, QDR IB HCA/ 10 GigE HCA, Dual GigE Ports; Item 3: Compute Switch: Ports: 24 fixed SFP+ ports 10 GigE of Low latency or suitable IB switch Performance: 900-Gbps switching capacity Preferably latency less than 300 nanoseconds. Item 4: Management Switch: 24 Port Layer 2 GigE Switch. The switch proposed should connect to unified Switching fabric at company. Item 5: Rack - Industry standard Rack/ OEM Rack suitable for Quoted Server and Storage has to be proposed Item 6: Software: • Cent OS Operating System, Intel Compilers – 2 user license, C & Fortran Compiler, Math Libraries. • Vendor specific Infiniband/GigE stack on Linux OS, if available, should be supplied. • Vendor specific MPI implementation on Linux OS, if available, including MPI over IB and IpoIB if required, should be supplied. • Scheduler – Scheduler propsoed should be fully supported by bidder. Scheduler should be from Open grid Engine, PBS Pro and LSF only. • NO Open source, community edition cluster management tool to be proposed. • GUI Based Job submission portal • Users should only apply the jobs to nodes that have the required characteristics, therefore matching the best job to the best resource to get maximum performance. • All Application like Ansys CFX, Fluent, Ansys Mech, MD/MSC Nastran, LSDyna, Abacus, Hmmer, Gromacs, Fasta, Blast, Amber, Gaussian, ClusterlW, Altair Optistruc, Converge, Femfat, Altair Radioss to be integrated with the portal proposed.
harshakudaravally Posted February 1, 2013 Author Report Posted February 1, 2013 [quote name='PMR aka OM' timestamp='1359748338' post='1303206268'] Enti ba... e High Performance story vesaavu.. Architect role ki job specification.. interview question maadiriga undi.. Edina Order place chestaa.. quotes prepare cehyyamannara enti.. akkad Servers purchase cheyyali.. total : 16 +1 servers konni routers.. konni sitches.. inkaa evevo required softwares Item 1: Compute Servers - 16 Nos At least 2*2.6 GHz Processor (8 Cores / 8 FLOPS per Processor), 2GB/Core memory ECC Memory; 2*500GB HDD, QDR IB HCA/ 10 GigE HCA, Dual GigE Ports; Memory per node to be scalable to 256GB in future. Item 2: Master Node: 1 No At least 2*2.6 GHz Processor (8 Cores / 8 FLOPS Per Processor), 4GB/Core Memory ECC Memory; 2TB usable HDD Space, QDR IB HCA/ 10 GigE HCA, Dual GigE Ports; Item 3: Compute Switch: Ports: 24 fixed SFP+ ports 10 GigE of Low latency or suitable IB switch Performance: 900-Gbps switching capacity Preferably latency less than 300 nanoseconds. Item 4: Management Switch: 24 Port Layer 2 GigE Switch. The switch proposed should connect to unified Switching fabric at company. Item 5: Rack - Industry standard Rack/ OEM Rack suitable for Quoted Server and Storage has to be proposed Item 6: Software: • Cent OS Operating System, Intel Compilers – 2 user license, C & Fortran Compiler, Math Libraries. • Vendor specific Infiniband/GigE stack on Linux OS, if available, should be supplied. • Vendor specific MPI implementation on Linux OS, if available, including MPI over IB and IpoIB if required, should be supplied. • Scheduler – Scheduler propsoed should be fully supported by bidder. Scheduler should be from Open grid Engine, PBS Pro and LSF only. • NO Open source, community edition cluster management tool to be proposed. • GUI Based Job submission portal • Users should only apply the jobs to nodes that have the required characteristics, therefore matching the best job to the best resource to get maximum performance. • All Application like Ansys CFX, Fluent, Ansys Mech, MD/MSC Nastran, LSDyna, Abacus, Hmmer, Gromacs, Fasta, Blast, Amber, Gaussian, ClusterlW, Altair Optistruc, Converge, Femfat, Altair Radioss to be integrated with the portal proposed. [/quote] Hardware and software required is already populated and OEM are working on the quote part. Is there any one who can give me a design on how to implement this.
PMREDDY19 Posted February 1, 2013 Report Posted February 1, 2013 Cent OS Run cheyyagalgina.. Intel Hardware purchase cheyayli emo.. memu ithey.. HP DL 380 . DL 580 Servers use chestaamu ikkada U S lo.. They come with Intel X.** processors.. RHEL tho nadustaayi..
harshakudaravally Posted February 1, 2013 Author Report Posted February 1, 2013 [quote name='PMR aka OM' timestamp='1359748542' post='1303206312'] Cent OS Run cheyyagalgina.. Intel Hardware purchase cheyayli emo.. memu ithey.. HP DL 380 . DL 580 Servers use chestaamu ikkada U S lo.. They come with Intel X.** processors.. RHEL tho nadustaayi.. [/quote] Bhayya r u good at implementing this?
PMREDDY19 Posted February 1, 2013 Report Posted February 1, 2013 [quote name='SH Kudaravally' timestamp='1359748524' post='1303206308'] Hardware and software required is already populated and OEM are working on the quote part. Is there any one who can give me a design on how to implement this. [/quote] whatever u r asking is way above than the information/knowledge than I know.. But nee Requirements ni batti.. I can see some kind of standing up a .. 1 MAster connecting to 16 Servers. with a new network to be setup in a Seperate Rack inside the datacenter.. with some dedicated Switches. chitti & k2s emiana chustaaru emo.. try cheddamu.. k2s ki konchem idea undaali e area lo..
PMREDDY19 Posted February 1, 2013 Report Posted February 1, 2013 [quote name='SH Kudaravally' timestamp='1359748626' post='1303206325'] Bhayya r u good at implementing this? [/quote] ledhu baa nenu application implement chestaanu kani. motham server team kuda ma daniloney unnnaru.. will try to check with someone in my team who is more of a server guy than myself.. maa team loney servers kuda build chestaru.. i used to submit the Requests for the physical servers and all this kind of stuff sometime ago..ala edho konceh, telusu..
harshakudaravally Posted February 1, 2013 Author Report Posted February 1, 2013 [quote name='PMR aka OM' timestamp='1359748841' post='1303206365'] ledhu baa nenu application implement chestaanu kani. motham server team kuda ma daniloney unnnaru.. will try to check with someone in my team who is more of a server guy than myself.. maa team loney servers kuda build chestaru.. i used to submit the Requests for the physical servers and all this kind of stuff sometime ago..ala edho konceh, telusu.. [/quote] Hey can u please check my PM
Recommended Posts