当前位置 : 首页 > 市场合作 > 项目信息市场合作
Market cooperation| 发包方名称 | National University Corporation - Ibaraki |
| 发包方国别(国家/地区) | 日本 |
| 发包方座机 | |
| 发包方传真 | |
| 发包方邮箱 | |
| 发包方地址 | Associate Head, Center for Computational Sciences, University of Tsukuba, 1-1-1 Tennodai Tsukuba-shi 305-8577 Japan |
| 发包方网址 | |
| 发包方联系人姓名 | Akira Shigeta |
| 发包方联系人座机 | 029-853-6486 |
| 发包方联系人手机 | |
| 发包方联系人邮箱 | |
| 项目名称 | Big Memory Supercomputer System 1 Set |
| 项目类型 | - |
| 项目预算¥ | |
| 项目投标截止时间 | 2021-04-28 |
| 项目开始时间 | |
| 项目结束时间 | |
| 服务商资质要求 | |
| 要求高匹配度成功案例个数 | |
| 项目交付地点 | |
| 是否准许转包或分包 | |
| 项目信息详述 |
⑴ Classification of the products to be procured : 14
⑵ Nature and quantity of the products to be purchased : Big Memory Supercomputer System 1 Set
⑶ Type of the procurement : Purchase
⑷ Basic requirements of the procurement : A Big Memory Supercomputer System consists of parallel processing servers, a parallel file system, login servers, and other hardware and software for stable and efficient operation of the whole system.
A Parallel processing servers must satisfy the following conditions.
a Parallel processing servers consist of compute nodes having identical specification.
b A compute node consists of CPU part and GPU (Graphics Processing Unit) part, which is connected by an interconnection network.
c CPU part consists of one or two sockets of general-purpose CPU, main memory, persistent memory and other peripherals. GPU part consists of one or two GPUs that the same number of CPUs.
d Theoretical peak performance of each CPU must be 3TFLOPS or higher in 64bit double precision floating point arithmetic.
e Memory capacity of each CPU must be 128GiByte or larger, and the theoretical peak bandwidth for each CPU must be 280GByte/sec or higher.
f Persistent memory capacity of each CPU must be 1TiByte or larger, and the theoretical peak bandwidth of random accesses must be 24GByte/sec or higher.
g Theoretical peak performance of each GPU must be equal to or higher than 19.5TFLOPS in double precision floating point arithmetic.
h Each GPU must have 64GiByte or large capacity of device memory, and the theoretical peak bandwidth must be 2TByte/sec or higher.
i Each compute node must be equipped with NVMe SSD with 2TByte or larger capacity for each CPU. It must have 3DW/D (Drive Write per Day) or higher durability.
j Each compute node must be connected with a full bisection network interconnect, each speed must be 200Gbps or higher for each CPU.
k The aggregated peak performance of double precision floating point of CPU part and GPU part must be 0.9PFLOPS or higher and 5.8PFLOPS or higher, respectively.
l The aggregated memory capacity of CPU part must be 38 TiByte or larger, and the aggregated persistent memory capacity of CPU part must be 300 TiByte or larger.
B A parallel file system must satisfy the following conditions.
a The parallel file system must be connected to the interconnection network of all compute nodes.
b The capacity of user available space must be 9PByte or larger.
c The theoretical peak bandwidth must be 80GByte/sec or higher.
d Each device of the parallel file system must be swappable without stopping the entire system.
C Login servers must satisfy the following conditions.
a Login servers must have 3 nodes or more.
b Login servers must be connected to the interconnection network of all compute nodes.
c Each login server must have a network interface for external network at the rate of 100Gbps or higher, and must be connected to an external network router.
d The external network router must have 1 port or more of 100Gbit Ethernet and 3 ports or more of 10Gbit Ethernet.
e If it is possible to connect to the existing Cygnus supercomputer by the InfiniBand network, the proposal is welcome.
D The software available on all the compute nodes must satisfy the following conditions.
a The system must be operated by Linux operating system.
b On each compute node, Fortran, C and C++ including automatic parallelization feature and OpenMP must be supported for CPU part.
c On each compute node, CUDA (including CUDA translator) and OpenACC or OpenMP must be supported for GPU part.
d MPI 3.0 or later must be provided for parallel processing over all compute nodes.
e Highly parallelized numerical and artificial intelligence library must be provided.
f Batch-job scheduling system to manage all the compute nodes must be provided.
E The power consumption of parallel processing servers, the parallel file system, login servers and all the cooling system must be equal to or less than 300kVA in total. To support the continuous operation of parallel processing servers and the parallel file system which are fully powered, power supply unit capacity, cooling system, floor plan, and all the environments must be carefully considered.
|
版权所有:天津市数字贸易综合服务平台 备案号:津B2-20080229-9