平成  年  月  日 月曜日 官 報 (号外政府調達第  号) 3 説明書の交付 本公表に基づき応募する供給 者に対して導入説明書を交付する。 ¥ 交付期間 平成22年9月13日から平成22年 11月5日まで。 ] 交付場所 上記2]に同じ。 4 説明会の開催 本公表に基づく導入説明会を 開催する。 ¥ 開催日時 平成22年9月29日15時00分 ] 開催場所 国立遺伝学研究所本館2階会議 室 5 その他 この導入計画の詳細は導入説明書に よる。なお、本公表内容は予定であり、変更す ることがあり得る。 6 Summary ¥ Classification of the products to be procured : 14 ] Nature and quantity of the products to be rent : Supercomputer system 1 Set ^ Type of the procurement : Rent _ Basic requirements of the procurement : A This system should be mainly composed of a supercomputer, large capacity storage systems, Web servers, RDB servers and network devices. B The supercomputer should be composed of several nodes as distributed memory parallel computers. The supercomputer should be composed of three types (thin, medium, fat) of nodes having different number of CPUs, memory sizes, and local disk sizes. Theoretical peak performance on double‐precision floating point of each node must be 144 GFLOPS or higher. The theoretical peak performance of the supercomputer except GPGPU must be 172 TFLOPS or higher as a whole. C Each node runs the same version of Linux operating system. All the CPUs in the nodes must support ISA (instruction set architecture) of both x86‘32 and x86‘64. The injection bandwidth for communication between any two nodes should be equal to or faster than 40Gbps. And bisection bandwidth of the network must be (injection bandwidth × number of nodes)/2 or greater. Each fat node must facilitate 32 cores or more x86‘64 CPUs, and 1TB or more memory. There must be 8TB or more local disk per node. There must be two or more fat nodes in the supercomputer. E Each medium node must facilitate 12 cores or more x86‘64 CPUs, 96GB or more memory and 400 GB or more SSD. The read performance of the SSD must be 400MB/s or greater, and the write performance must be 140MB/s or greater. At least 30% of all compute nodes shall be the medium nodes. F Each thin node must facilitate 12 cores or more x86‘64 CPUs, 48GB or more memory, 400GB or more SSD. The read performance of the SSD must be 400MB/s or greater, and the write performance must be 140MB/s or greater. About 15% of the thin nodes must facilitate 1TB or more SSD. G About 15% of the thin nodes must facilitate at least one GPGPU. The peak performance on double‐precision floating point of each chip must be 1 TFLOPS or higher, and the peak performance on single‐precision must be 500GFLOPS or higher. Each GPGPU accelerator must be connected to independent PCI Express generation‐2×16 channels with full bandwidth. To provide for future extension of the supercomputer system, it is preferable that all thin nodes and all medium nodes must facilitate two or more slots for GPGPU cards. H Large capacity storage system must facilitate 10 PB or greater effective capacity under triple‐redundant or greater technologies such as RAID6 for data integrity protection. Physical data access from all node‐connecting switches to all disk controllers must be redundant (except the data path from disk controllers to physical disks). Part of the large capacity storage system can be accessed by all nodes via highly reliable and highly performing, parallel, shared file system. All the file system must facilitate appropriate redundancy which enables dynamic failover. D I Physical I/O performance between nodes and the shared file system must be 120 GB/s or higher in total of all nodes. This is the performance in which the effect of all the other components(including secondary storage controllers, networks) on the performance are taken into account. The physical I/O performance without them must be 240GB/s or higher. J We request submission for materials with allowing the proposals of storage configuration part of which utilizes hierarchical storage management system and compensates less than 50% of storage capacity with MAID or the tape devices. In this case, the transfer rate from the MAID or the tape devices to the shared file system must be 5GB/s or higher. K Part of thin nodes must be provided with 4 or more 20TB or greater RAID6 systems, to accommodate currently operating distributed RDB systems for DDBJ database construction. L Part of thin nodes should be used for currently available web services of DDBJ, including DDBJ home page, search programs and WebAPIs, which requires independent separate security management @which must facilitate two or more 20TB or greater RAID6 disks. M The network devices must be able to connect to SINET4 with 40GB/s or higher bandwidth. N There must be high‐speed data transfer facilities such as Aspera. O Propose power saving mechanisms which enable the annual averaged PUE (power usage effectiveness) to be 1.4 or less. P The above mentioned supercomputer system must be equipped with UPS devices with sufficient capacity. ‘ Time limit for the submission of the requested material : 5 : 15 PM 5, November, 2010 a Contact point for the notice : Masashi Uematsu, Supplies Team, Research Promotion Section, National Institute of Genetics, Research Organization of Information and Systems, 1111 Yata Mishima‐shi 411‘8540 Japan, TEL 055‘981‘6730 意見招請に関する公示 次のとおり調達特定役務の仕様書案の作成が完 了したので、 仕様書案に対する意見を招請します。 平成 22 年9月 13 日 国立教育政策研究所総務部長 上口 孝之 ◎調達機関番号 016 ◎所在地番号 13 〇第1号 1 調達内容 ¥ 品目分類番号 76 ] 購入等特定役務及び数量 A.平成23年度全国学力・学習状況調査問 題用紙(小学校)の印刷 一式 B.平成23 年度全国学力・学習状況調査問題用紙(中学 校)の印刷 一式 2 意見の提出方法 ¥ 意見の提出期限 平成22年10月4日11時00 分(郵送の場合は必着のこと。 ) ] 提出先 〒100‘8951 東京都千代田区霞が 関3‘2‘2 国立教育政策研究所総務部会 計課管理係 宇都宮 隆 電話 03 ‘ 6733 ‘ 6918 3 仕様書案の交付 ¥ 交付期間 平成22年9月13日から平成22年 10月4日まで。 ] 交付場所 上記2]に同じ。 4 仕様書案の説明会 ¥ 開催日時 平成22年9月16日14時00分 ] 開催場所 国立教育政策研究所第2特別会 議室 5 Summary ¥ Classification of the services to be procured : 76 ] Nature and quantity of the services to be required : A. Print of Nationwide Assessment of Academic Ability testbooklet for Elementary school in 2011 fiscal year 1 Set B.Print of Nationwide Assessment of Academic Ability testbooklet for Junior high school in 2011 fiscal year 1 Set