m (→Definition) Tag: Source edit |
m (→Definition) Tag: Source edit |
||
(One intermediate revision by the same user not shown) | |||
Line 1: | Line 1: | ||
== Definition == |
== Definition == |
||
− | '''Supercomputing systems''' employ a large number of interdependent [[processor]]s, which are the core unit of a [[computer]] that gathers [[instruction]]s and [[data]]. These [[processor]]s are mounted onto a [[computer chip]], which is the portion of the system that carries out the [[instruction]]s of a [[computer program]]. These [[ |
+ | '''Supercomputing systems''' employ a large number of interdependent [[processor]]s, which are the core unit of a [[computer]] that gathers [[instruction]]s and [[data]]. These [[processor]]s are mounted onto a [[computer chip]], which is the portion of the system that carries out the [[instruction]]s of a [[computer program]]. These [[computer chip]]s are inserted onto a [[computer card]], which also holds [[memory]] for the [[computer chip]]s to use. A number of [[computer card]]s are attached to a [[node card]], which have one or more [[processor]]s with a common [[memory]] and are connected by high-speed [[interconnection network]]s. Each [[node card]] is inserted into a single cabinet, and that [[configuration]] is repeated many times to build a single supercomputing system. |
Each supercomputing system has a peak performance, which is the maximum rate of [[floating-point operations per second]] ([[FLOPS]]) that the [[system]] can sustain. |
Each supercomputing system has a peak performance, which is the maximum rate of [[floating-point operations per second]] ([[FLOPS]]) that the [[system]] can sustain. |
Latest revision as of 21:01, 8 June 2021
Definition[]
Supercomputing systems employ a large number of interdependent processors, which are the core unit of a computer that gathers instructions and data. These processors are mounted onto a computer chip, which is the portion of the system that carries out the instructions of a computer program. These computer chips are inserted onto a computer card, which also holds memory for the computer chips to use. A number of computer cards are attached to a node card, which have one or more processors with a common memory and are connected by high-speed interconnection networks. Each node card is inserted into a single cabinet, and that configuration is repeated many times to build a single supercomputing system.
Each supercomputing system has a peak performance, which is the maximum rate of floating-point operations per second (FLOPS) that the system can sustain.
Figure 1 shows the common components of a supercomputing system.