|
==Computers==
To reflect NERSC's mission to support scientific research, the center names its major systems after scientists. The center is located in Shyh Wang Hall, one of the nation's most energy-efficient{{Citation needed|date=January 2022}} supercomputer facilities<ref>https://cs.lbl.gov/news-media/news/2020/less-is-more-lbnl-breaks-new-ground-in-data-center-optimization/</ref>. The building was financed by the [[University of California]] which manages Berkeley Lab for the [[United States Department of Energy|U.S. Department of Energy (DOE)]]. The utility infrastructure and computer systems are provided by DOE.
The newest supercomputer [[Perlmutter (supercomputer)|Perlmutter]], is named in honor of [[Saul Perlmutter]], an astrophysicist at Berkeley Lab who shared the 2011 [[Nobel Prize in Physics]] for his contributions to research showing that the expansion of the universe is accelerating. It is a [[Cray]] system based on the Shasta architecture, with [[Zen 3]] based [[Epyc|AMD Epyc]] [[Central processing unit|CPUs]] ("Milan") and [[Ampere (microarchitecture)|NVIDIA Ampere]] [[Graphics processing unit|GPUs]].<ref>{{Cite web|url=https://www.nersc.gov/systems/perlmutter/|title = Perlmutter}}</ref>
Another NERSC supercomputer is CORICori, named in honor of [[Gerty Cori]], a [[biochemist]] who was the first American woman to receive a [[Nobel Prize]] in science. Cori is a [[Cray XC40]] system with 622,336 Intel processor cores and a theoretical peak performance of 30 [[petaflop]]/s (30 quadrillion operations per second). Cori was delivered in two phases. The first phase — also known as the Data Partition — was installed in late 2015 and comprises 12 cabinets and more than 1,600 [[Xeon|Intel Xeon]] "Haswell" compute nodes. ItThe wassecond customizedphase<ref>{{Cite toweb|url=http://www.nersc.gov/users/computational-systems/cori/configuration/cori-intel-xeon-phi-nodes/|title=Cori supportIntel dataXeon Phi (KNL) Nodes|website=www.nersc.gov|language=en|access-intensivedate=2018-02-09}}</ref> scienceof Cori, installed in summer 2016,<ref>{{Cite web|url=http://www.nersc.gov/news-publications/nersc-news/nersc-center-news/2016/cori-supercomputer-now-fully-installed-at-berkeley-lab/|title=Cori Supercomputer Now Fully Installed at Berkeley Lab|website=www.nersc.gov|language=en|access-date=2018-02-09}}</ref> added another 52 cabinets and themore analysisthan of9,300 largenodes with second-generation [[DataIntel set|datasetsXeon Phi]] throughprocessors a(code-named combinationKnights ofLanding, hardwareor andKNL softwarefor configurationsshort), andmaking queueCori policiesthe largest{{Citation needed|date=January 2022}} supercomputing system for open science based on KNL processors.
The second phase<ref>{{Cite web|url=http://www.nersc.gov/users/computational-systems/cori/configuration/cori-intel-xeon-phi-nodes/|title=Cori Intel Xeon Phi (KNL) Nodes|website=www.nersc.gov|language=en|access-date=2018-02-09}}</ref> of Cori, installed in summer 2016,<ref>{{Cite web|url=http://www.nersc.gov/news-publications/nersc-news/nersc-center-news/2016/cori-supercomputer-now-fully-installed-at-berkeley-lab/|title=Cori Supercomputer Now Fully Installed at Berkeley Lab|website=www.nersc.gov|language=en|access-date=2018-02-09}}</ref> added another 52 cabinets and more than 9,300 nodes with second-generation [[Intel Xeon Phi]] processors (code-named Knights Landing, or KNL for short), making Cori the largest{{Citation needed|date=January 2022}} supercomputing system for open science based on KNL processors. With 68 active physical cores on each KNL and 32 on each Haswell processor, Cori has almost 700,000 [[processor core]]s. The two phases of Cori are integrated via the Cray Aries interconnect, which has a dragonfly [[network topology]] that provides scalable [[Bandwidth (computing)|bandwidth]].
Cori features a Burst Buffer based on the Cray DataWarp technology. The Burst Buffer, a 1.5 PB layer of [[Non-volatile random-access memory|NVRAM]] storage, sits between compute node memory and Cori's 30-[[petabyte]] Lustre parallel scratch [[file system]]. The burst buffer provides about 1.5 TB/sec of [[Input/output|I/O]] bandwidth, more than twice that of the scratch file system. NERSC has also added software-defined [[wikt:networking|networking]] features to Cori to more efficiently move data in and out of the system, giving users end-to-end connectivity and bandwidth for real-time data analysis, and a real-time queue for time-sensitive analyses of data.
|
