NUMAlink

Last updated

NUMAlink is a system interconnect developed by Silicon Graphics (SGI) for use in its distributed shared memory ccNUMA computer systems. NUMAlink was originally developed by SGI for their Origin 2000 and Onyx2 systems. At the time of these systems' introduction, it was branded as "CrayLink" during SGI's brief ownership of Cray Research. [1]

Contents

Hewlett Packard Enterprise entered an original equipment manufacturer (OEM) arrangement with Silicon Graphics International (SGI) to use Numalink as the foundation in some mission critical servers. [2]

GenerationIntroductionBandwidth [lower-roman 1] Systems
Stanford DASH~19900.12 GB/s [lower-roman 2]
NUMAlink 219960.8 GB/s Origin 200, Origin 2000, Onyx2
NUMAlink 320001.6 GB/s Origin 3000, Altix 3000
NUMAlink 420043.2 GB/s Altix 4000
NUMAlink 520097.5 GB/s Altix UV
NUMAlink 620126.7 GB/s SGI UV 2000, SGI UV 3000, SGI UV 30
NUMAlink 7201414.9 GB/sHPE Integrity MC990 X/SGI UV 300, SGI UV 30EX
NUMAlink 8201713.3 GB/sHPE Superdome Flex
Notes
  1. ↑ Bandwidth per port in each direction (each NUMAlink port is a dual simplex channel).
  2. ↑ DASH used separate request and reply mesh networks, so this bandwidth number is not directly comparable to NUMAlink.

There was no NUMAlink 1, as SGI's engineers deemed the system interconnect used in the Stanford DASH to be the first generation NUMAlink interconnect. NUMAlink 2 (branded as CrayLink) was announced in October 1996 for the Onyx2 visualization systems, the Origin 200 and the Origin 2000 servers and supercomputers. [1] The NUMAlink 2 interface is the Hub ASIC. NUMAlink 2 is capable of 1.6 GB/s of peak bandwidth through two 800 MB/s, PECL 400 MHz 16-bit unidirectional links. [3]

NUMAlink 3 is the third generation of the interconnect, introduced in 2000 and used in the Origin 3000 and Altix 3000. NUMAlink 3 is capable of 3.2 GB/s of peak bandwidth through two 1.6 GB/s unidirectional links. [3] The name NUMAflex reflects the modular design approach around this time. [4]

NUMAlink 4 is the fourth generation of the interconnect, introduced in 2004 and used in the Altix 4000. NUMAlink 4 is capable of 6.4 GB/s of peak bandwidth through two 3.2 GB/s unidirectional links. [3]

NUMAlink 5 is the fifth generation of the interconnect, introduced in 2009 and used in the Altix UV series. NUMAlink 5 is capable of 15 GB/s of peak bandwidth through two 7.5 GB/s unidirectional links. [5]

NUMAlink 6 is the sixth generation of the interconnect, introduced in 2012 and used in the SGI UV 2000, SGI UV 3000, SGI UV 30. NUMAlink 6 is capable of 6.7 GB/s of bidirectional peak bandwidth for up to 256 socket system and 64TB of coherent shared memory. [6] [7]

NUMAlink 7 is the seventh generation of the interconnect, introduced in 2014 and used in the HPE Integrity MC990 X/SGI UV 300, SGI UV 30EX, SGI UV 300H, SGI UV 300RL. NUMAlink 7 is capable of 14.94 GB/s of bidirectional peak bandwidth for up to 64 socket system and 64TB of coherent shared memory. [8] [9] [10]

NUMAlink 8 is the eighth generation of the interconnect, introduced in 2017 and used in the HPE Superdome Flex. NUMAlink 8 provides 13.3 GB/s of bandwidth per port [11] and systems using it are capable of 853.33 GB/s of bisection peak bandwidth (64 links are cut) across a 32 socket system with up to 48 TB of coherent shared memory. [12] [13] [14]

See also

Related Research Articles

<span class="mw-page-title-main">IRIX</span> Computer operating system

IRIX is a discontinued operating system developed by Silicon Graphics (SGI) to run on the company's proprietary MIPS workstations and servers. It is based on UNIX System V with BSD extensions. In IRIX, SGI originated the XFS file system and the industry-standard OpenGL graphics API.

<span class="mw-page-title-main">Itanium</span> Family of 64-bit Intel microprocessors

Itanium is a discontinued family of 64-bit Intel microprocessors that implement the Intel Itanium architecture. The Itanium architecture originated at Hewlett-Packard (HP), and was later jointly developed by HP and Intel. Launched in June 2001, Intel initially marketed the processors for enterprise servers and high-performance computing systems. In the concept phase, engineers said "we could run circles around PowerPC...we could kill the x86." Early predictions were that IA-64 would expand to the lower-end servers, supplanting Xeon, and eventually penetrate into the personal computers, eventually to supplant reduced instruction set computing (RISC) and complex instruction set computing (CISC) architectures for all general-purpose applications.

<span class="mw-page-title-main">Silicon Graphics</span> 1981–2009 American computing company

Silicon Graphics, Inc. was an American high-performance computing manufacturer, producing computer hardware and software. Founded in Mountain View, California, in November 1981 by James Clark, its initial market was 3D graphics computer workstations, but its products, strategies and market positions developed significantly over time.

<span class="mw-page-title-main">Non-uniform memory access</span> Computer memory design used in multiprocessing

Non-uniform memory access (NUMA) is a computer memory design used in multiprocessing, where the memory access time depends on the memory location relative to the processor. Under NUMA, a processor can access its own local memory faster than non-local memory. NUMA is beneficial for workloads with high memory locality of reference and low lock contention, because a processor may operate on a subset of memory mostly or entirely within its own cache node, reducing traffic on the memory bus.

HyperTransport (HT), formerly known as Lightning Data Transport, is a technology for interconnection of computer processors. It is a bidirectional serial/parallel high-bandwidth, low-latency point-to-point link that was introduced on April 2, 2001. The HyperTransport Consortium is in charge of promoting and developing HyperTransport technology.

Cray Inc., a subsidiary of Hewlett Packard Enterprise, is an American supercomputer manufacturer headquartered in Seattle, Washington. It also manufactures systems for data storage and analytics. Several Cray supercomputer systems are listed in the TOP500, which ranks the most powerful supercomputers in the world.

XIO is a packet-based, high-performance computer bus employed by the SGI Origin 2000, Octane, Altix, Fuel and Tezro machines. The XIO forms a bus between high-performance system devices and the memory controller.

<span class="mw-page-title-main">Altix</span> Supercomputer family

Altix is a line of server computers and supercomputers produced by Silicon Graphics, based on Intel processors. It succeeded the MIPS/IRIX-based Origin 3000 servers.

Fireplane is a computer internal interconnect created by Sun Microsystems.

<span class="mw-page-title-main">Cray T3E</span>

The Cray T3E was Cray Research's second-generation massively parallel supercomputer architecture, launched in late November 1995. The first T3E was installed at the Pittsburgh Supercomputing Center in 1996. Like the previous Cray T3D, it was a fully distributed memory machine using a 3D torus topology interconnection network. The T3E initially used the DEC Alpha 21164 (EV5) microprocessor and was designed to scale from 8 to 2,176 Processing Elements (PEs). Each PE had between 64 MB and 2 GB of DRAM and a 6-way interconnect router with a payload bandwidth of 480 MB/s in each direction. Unlike many other MPP systems, including the T3D, the T3E was fully self-hosted and ran the UNICOS/mk distributed operating system with a GigaRing I/O subsystem integrated into the torus for network, disk and tape I/O.

<span class="mw-page-title-main">R10000</span> MIPS microprocessor

The R10000, code-named "T5", is a RISC microprocessor implementation of the MIPS IV instruction set architecture (ISA) developed by MIPS Technologies, Inc. (MTI), then a division of Silicon Graphics, Inc. (SGI). The chief designers are Chris Rowen and Kenneth C. Yeager. The R10000 microarchitecture is known as ANDES, an abbreviation for Architecture with Non-sequential Dynamic Execution Scheduling. The R10000 largely replaces the R8000 in the high-end and the R4400 elsewhere. MTI was a fabless semiconductor company; the R10000 was fabricated by NEC and Toshiba. Previous fabricators of MIPS microprocessors such as Integrated Device Technology (IDT) and three others did not fabricate the R10000 as it was more expensive to do so than the R4000 and R4400.

<span class="mw-page-title-main">HPE Superdome</span> Series of server computers

The HPE Superdome is a high-end server computer designed and manufactured by Hewlett Packard Enterprise. The product's most recent version, "Superdome 2," was released in 2010 supporting 2 to 32 sockets and 4 TB of memory. The Superdome used PA-RISC processors when it debuted in 2000. Since 2002, a second version of the machine based on Itanium 2 processors has been marketed as the HP Integrity Superdome.

The Challenge, code-named Eveready and Terminator, is a family of server computers and supercomputers developed and manufactured by Silicon Graphics in the early to mid-1990s that succeeded the earlier Power Series systems. The Challenge was later succeeded by the NUMAlink-based Origin 200 and Origin 2000 in 1996.

<span class="mw-page-title-main">SGI Origin 2000</span> Series of server computers

The SGI Origin 2000 is a family of mid-range and high-end server computers developed and manufactured by Silicon Graphics (SGI). They were introduced in 1996 to succeed the SGI Challenge and POWER Challenge. At the time of introduction, these ran the IRIX operating system, originally version 6.4 and later, 6.5. A variant of the Origin 2000 with graphics capability is known as the Onyx2. An entry-level variant based on the same architecture but with a different hardware implementation is known as the Origin 200. The Origin 2000 was succeeded by the Origin 3000 in July 2000, and was discontinued on June 30, 2002.

<span class="mw-page-title-main">SGI Origin 200</span> Entry-level server by Silicon Graphics

The SGI Origin 200, code named Speedo, was an entry-level server computer developed and manufactured by SGI, introduced in October 1996 to accompany their mid-range and high-end Origin 2000. It is based on the same architecture as the Origin 2000 but has an unrelated hardware implementation. At the time of introduction, these systems ran the IRIX 6.4, and later, the IRIX 6.5 operating systems. The Origin 200 was discontinued on 30 June 2002.

The Origin 3000 and the Onyx 3000 is a family of mid-range and high-end computers developed and manufactured by SGI. The Origin 3000 is a server, and the Onyx 3000 is a visualization system. Both systems were introduced in July 2000 to succeed the Origin 2000 and the Onyx2 respectively. These systems ran the IRIX 6.5 Advanced Server Environment operating system. Entry-level variants of these systems based on the same architecture but with a different hardware implementation are known as the Origin 300 and Onyx 300. The Origin 3000 was succeeded by the Altix 3000 in 2004 and the last model was discontinued on 29 December 2006, while the Onyx 3000 was succeeded by the Onyx4 and the Itanium-based Prism in 2004 and the last model was discontinued on 25 March 2005.

The National Center for Computational Sciences (NCCS) is a United States Department of Energy (DOE) Leadership Computing Facility that houses the Oak Ridge Leadership Computing Facility (OLCF), a DOE Office of Science User Facility charged with helping researchers solve challenging scientific problems of global interest with a combination of leading high-performance computing (HPC) resources and international expertise in scientific computing.

<span class="mw-page-title-main">Silicon Graphics International</span> Former computer hardware and software company

Silicon Graphics International Corp. was an American manufacturer of computer hardware and software, including high-performance computing systems, x86-based servers for datacenter deployment, and visualization products. The company was founded as Rackable Systems in 1999, but adopted the "SGI" name in 2009 after acquiring Silicon Graphics Inc. out of bankruptcy.

Coherent Accelerator Processor Interface (CAPI), is a high-speed processor expansion bus standard for use in large data center computers, initially designed to be layered on top of PCI Express, for directly connecting central processing units (CPUs) to external accelerators like graphics processing units (GPUs), ASICs, FPGAs or fast storage. It offers low latency, high speed, direct memory access connectivity between devices of different instruction set architectures.

References

  1. 1 2 "Silicon Graphics and Cray Research Unveil Modular Origin Server Family: High-Bandwidth Systems Revolutionize Computer Buying Economics With Seamless Scalability". Press release. October 7, 1996. Archived from the original on July 7, 1997. Retrieved September 21, 2013.
  2. ↑ "SGI Enters OEM Agreement with Hewlett Packard Enterprise (HPE) to Deliver SGI UV Technology through HPE Mission Critical Solutions - SGI Blog". 9 February 2016. Archived from the original on 4 November 2016. Retrieved 11 August 2016.
  3. 1 2 3 "SGI® NUMAlink™ Industry Leading Interconnect Technology" (PDF). White paper. April 13, 2005. Archived from the original (PDF) on March 28, 2006. Retrieved September 21, 2013.
  4. ↑ John Mashey (August 30, 2000). "NUMAflex Modular Design Approach: A Revolution in Evolution" . Retrieved September 28, 2016.
  5. ↑ "SGI Altix UV" (PDF). www.sgi.com. Silicon Graphics International. 2009. Retrieved 2009-11-18.
  6. ↑ "SGI UV 2000 Datasheet" (PDF). www.sgi.com. Silicon Graphics International. 2015. Archived from the original (PDF) on 2017-05-17. Retrieved 2016-02-10.
  7. ↑ "SGI UV 3000, UV 30 Datasheet" (PDF). www.sgi.com. Silicon Graphics International. 2015. Archived from the original (PDF) on 2017-06-11. Retrieved 2016-02-10.
  8. ↑ "SGI UV 300, UV 30EX Datasheet" (PDF). www.sgi.com. Silicon Graphics International. 2015. Retrieved 2016-02-10.
  9. ↑ "SGI UV 300H Datasheet" (PDF). www.sgi.com. Silicon Graphics International. 2015. Retrieved 2016-02-10.
  10. ↑ "SGI UV 300RL Datasheet" (PDF). www.sgi.com. Silicon Graphics International. 2015. Retrieved 2016-02-10.
  11. ↑ "Endeavour Configuration Details". NASA Advanced Supercomputing (NAS) Division. 2021-08-05. Retrieved 2023-10-28. A Superdome Flex ASIC (also known as NUMAlink 8) provides 16 flex grid ports, each capable of 13.3 GB/s data rates for maximum flex grid bandwidth. The total bi-sectional crossbar grid bandwidth for a 32-socket Superdome Flex server is more than 850 GB/s.
  12. ↑ "HPE's Superdome Gets An SGI NUMAlink Makeover". www.nextplatform.com. The Next Platform. 2017. Retrieved 2017-11-14.
  13. ↑ "HPE Superdome Flex Server - Product documentation". hpe.com. HPE. 2017. Retrieved 2017-11-14.
  14. ↑ "HPE High Performance Computing & AI Solutions" (PDF). www.hpe.com. Hewlett Packard Enterprise. 2017. Retrieved 2018-03-24.
This page is based on this Wikipedia article
Text is available under the CC BY-SA 4.0 license; additional terms may apply.
Images, videos and audio are available under their respective licenses.