Xeon Phi

From Wikipedia, the free encyclopedia
  (Redirected from Intel MIC)
Jump to: navigation, search
Not to be confused with the ATI Xenos, or Xenon.
The Tianhe-2 supercomputer uses Xeon Phi processors
Xeon Phi
Instruction set x64

Xeon Phi is a main/co processor designed, manufactured, marketed, and sold by Intel. It is primarily focused toward the supercomputer and enterprise markets (including servers, and workstations). When used in a supercomputer, server, or workstation environment it will usually work alongside an existing Intel Xeon processor or processor(s). Unlike Intel Core i7, Intel Xeon CPUs that are reliant on the LGA 2011-3 CPU socket, Xeon Phi CPUs require PCI Express v3 x16 in either x8 or x16 mode. The default in most cases is "auto switching".

The product falls into the Many Integrated Core (MIC) architecture and prototype products codenamed Knights Ferry were announced and released to developers in 2010. The Knights Corner product was announced in 2011 and uses a 22 nm process. A second generation product codenamed Knights Landing using a 14 nm process was announced in June 2013.

In September 2011, the Texas Advanced Computing Center (TACC) announced it would use Knights Corner cards in their 10 petaFLOPS "Stampede" supercomputer, providing 8 petaFLOPS of computing power.

At the International Supercomputing Conference (2012, Hamburg), Intel announced the branding of the processor product family as Intel Xeon Phi.[1]

In November 2012, Intel formally announced the first products citing claims of CPU-like versatile programmability, high performance and power efficiency.[2] The Green 500 list placed a system using these new products as the most power efficient computer in the world.[3]

In June 2013, the Tianhe-2 supercomputer at the National Supercomputing Center in Guangzhou (NSCC-GZ) was announced[4] as the world's fastest supercomputer. It utilizes Intel Xeon Phi coprocessors and Ivy Bridge-EP Xeon processors to achieve 33.86 petaFLOPS.[5]

Competitors include Nvidia's Tesla-branded product lines.



The Larrabee microarchitecture (in development since 2006[6]) introduced very wide (512-bit) SIMD units to a x86 architecture based processor design, extended to a cache-coherent multiprocessor system connected via a ring bus to memory; each core was capable of four-way multithreading. Due to the design being intended for GPU as well as general purpose computing the Larrabee chips also included specialised hardware for texture sampling.[7][8] The project to produce a retail GPU product directly from the Larrabee research project was terminated in May 2010.[9]

Another contemporary Intel research project implementing x86 architecture on a many-multicore processor was the 'Single-chip Cloud Computer' (prototype introduced 2009[10]), a design mimicking a cloud computing computer datacentre on a single chip with multiple independent cores: the prototype design included 48 cores per chip with hardware support for selective frequency and voltage control of cores to maximize energy efficiency, and incorporated a mesh network for interchip messaging. The design lacked cache-coherent cores and focused on principles that would allow the design to scale to many more cores.[11]

The Teraflops Research Chip (prototype unveiled 2007[12]) is an experimental 80-core chip with two floating point units per core, implementing a 96-bit VLIW architecture instead of the x86 architecture.[13] The project investigated intercore communication methods, per-chip power management, and achieved 1.01 TFLOPS at 3.16 GHz consuming 62 W of power.[14][15]

Knights Ferry[edit]

Intel's MIC prototype board, named Knights Ferry, incorporating a processor codenamed Aubrey Isle was announced May 31, 2010. The product was stated to be a derivative of the Larrabee project and other Intel research including the Single-chip Cloud Computer.[16][17]

The development product was offered as a PCIe card with 32 in-order cores at up to 1.2 GHz with four threads per core, 2 GB GDDR5 memory,[18] and 8 MB coherent L2 cache (256 KB per core with 32 KB L1 cache), and a power requirement of ~300 W,[18] built at a 45 nm process.[19] In the Aubrey Isle core a 1,024-bit ring bus (512-bit bi-directional) connects processors to main memory.[20] Single board performance has exceeded 750 GFLOPS.[19] The prototype boards only support single precision floating point instructions.[21]

Initial developers included CERN, Korea Institute of Science and Technology Information (KISTI) and Leibniz Supercomputing Centre. Hardware vendors for prototype boards included IBM, SGI, HP, Dell and others.[22]

Knights Corner[edit]

The Knights Corner product line is made at a 22 nm process size, using Intel's Tri-gate technology with more than 50 cores per chip, and is Intel's first many-cores commercial product.[16][19]

In June 2011, SGI announced a partnership with Intel to utilize the MIC architecture in its high performance computing products.[23] In September 2011, it was announced that the Texas Advanced Computing Center (TACC) will use Knights Corner cards in their 10 petaFLOPS "Stampede" supercomputer, providing 8 petaFLOPS of compute power.[24] According to "Stampede: A Comprehensive Petascale Computing Environment" the "second generation Intel (Knights Landing) MICs will be added when they become available, increasing Stampede's aggregate peak performance to at least 15 PetaFLOPS."[25]

On November 15, 2011, Intel showed an early silicon version of a Knights Corner processor.[26][27]

On June 5, 2012, Intel released open source software and documentation regarding Knights Corner.[28]

On June 18, 2012, Intel announced at the 2012 Hamburg International Supercomputing Conference that Xeon Phi will be the brand name used for all products based on their Many Integrated Core architecture.[1][29][30][31][32][33][34] In June 2012, Cray announced it would be offering 22 nm 'Knight's Corner' chips (branded as 'Xeon Phi') as a co-processor in its 'Cascade' systems.[35][36]

In June 2012, ScaleMP announced it will provide its virtualization software to allow using 'Knight's Corner' chips (branded as 'Xeon Phi') as main processor transparent extension. The virtualization software will allow 'Knight's Corner' to run legacy MMX/SSE code and access unlimited amount of (host) memory without need for code changes.[37] An important component of the Intel Xeon Phi coprocessor’s core is its vector processing unit (VPU).[38] The VPU features a novel 512-bit SIMD instruction set, officially known as Intel® Initial Many Core Instructions (Intel® IMCI). Thus, the VPU can execute 16 single-precision (SP) or 8 double-precision (DP) operations per cycle. The VPU also supports Fused Multiply-Add (FMA) instructions and hence can execute 32 SP or 16 DP floating point operations per cycle. It also provides support for integers. The VPU also features an Extended Math Unit (EMU) that can execute operations such as reciprocal, square root, and log, thereby allowing these operations to be executed in a vector fashion with high bandwidth. The EMU operates by calculating polynomial approximations of these functions

On November 12, 2012, Intel announced two Xeon Phi coprocessor families using the 22 nm process size: the Xeon Phi 3100 and the Xeon Phi 5110P.[39][40][41] The Xeon Phi 3100 will be capable of more than 1 teraFLOPS of double precision floating point instructions with 240 GB/sec memory bandwidth at 300 W.[39][40][41] The Xeon Phi 5110P will be capable of 1.01 teraFLOPS of double precision floating point instructions with 320 GB/sec memory bandwidth at 225 W.[39][40][41] The Xeon Phi 7120P will be capable of 1.2 teraFLOPS of double precision floating point instructions with 352 GB/sec memory bandwidth at 300 W.

On June 17, 2013, the Tianhe-2 supercomputer was announced[4] by TOP500 as the world's fastest. It used Intel Ivy Bridge Xeon and Xeon Phi processors to achieve 33.86 petaFLOPS. According to the TOP500 list, Tianhe-2 was the world's fastest supercomputer since its introduction in June 2013 through the most recent list in November 2015.[42]

Knights Landing[edit]

Code name for the second generation MIC architecture product from Intel.[25] Intel officially first revealed details of its second generation Intel Xeon Phi products on June 17, 2013.[5] Intel said that the next generation of Intel MIC Architecture-based products will be available in two forms, as a coprocessor or a host processor (CPU), and be manufactured using Intel's 14nm process technology. Knights Landing products will include integrated on-package memory for significantly higher memory bandwidth.

Knights Landing will be built using up to 72 Airmont (Atom) cores with four threads per core,[43][44] using LGA 3647 socket[45] supporting for up to 384 GB of "far" DDR4 RAM and 8–16 GB of stacked "near" 3D MCDRAM, a version of High Bandwidth Memory. Each core will have two 512-bit vector units and will support AVX-512 SIMD instructions, specifically the Intel AVX-512 Foundational Instructions (AVX-512F) with Intel AVX-512 Conflict Detection Instructions (AVX-512CD), Intel AVX-512 Exponential and Reciprocal Instructions (AVX-512ER), and Intel AVX-512 Prefetch Instructions (AVX-512PF).[46]

The National Energy Research Scientific Computing Center announced that Phase 2 of its newest supercomputing system "Cori" would use Knights Landing Xeon Phi coprocessors.[47]

On June 20, 2016, Intel launched the Intel Xeon Phi product family x200 based on the Knights Landing architecture, stressing its applicability to not just traditional simulation workloads, but also to machine learning.[48][49] The model lineup announced at launch included only Xeon Phi of bootable form-factor, but two versions of it: standard processors and processors with integrated Intel Omni-Path architecture fabric.[50] The latter is denoted by the suffix F in the model number. Integrated fabric is expected to provide better the latency lower cost than discrete high-performance network cards.[48]

Knights Hill[edit]

Knights Hill is the codename for the third-generation MIC architecture, for which Intel announced the first details at SC14. It will be manufactured in a 10 nm process.[51]

In April 2015, the United States Department of Energy announced that a supercomputer named Aurora will be deployed at Argonne National Laboratory[52] based upon the "third-generation Intel Xeon Phi" processor.[53]

Knights Mill[edit]

Knights Mill is Intel's codename for a Xeon Phi product specialized in deep learning.[54] While little is known about Knights Mill yet, it has been announced that it will improve efficiency. It is also expected to support reduced variable precision which have been used to accelerate machine learning in other products, such as half-precision floating-point variables in Nvidia's Tesla.


The cores of Knights Corner are based on a modified version of P54C design, used in the original Pentium.[55] The basis of the Intel MIC architecture is to leverage x86 legacy by creating a x86-compatible multiprocessor architecture that can utilize existing parallelization software tools.[19] Programming tools include OpenMP, OpenCL,[56] Cilk/Cilk Plus and specialised versions of Intel's Fortran, C++[57] and math libraries.[58]

Design elements inherited from the Larrabee project include x86 ISA, 4-way SMT per core, 512-bit SIMD units, 32 KB L1 instruction cache, 32 KB L1 data cache, coherent L2 cache (512 KB per core[59]), and ultra-wide ring bus connecting processors and memory.

The Knights Corner instruction set documentation is available from Intel.[60][61][62]


An empirical performance and programmability study has been performed by researchers,[63] in which the authors claim that achieving high performance with Xeon Phi still needs help from programmers and that merely relying on compilers with traditional programming models is still far from reality. However, research in various domains, such as life sciences,[64] deep learning[65] and computer-aided engineering[66] demonstrated that exploiting both the thread- and SIMD-parallelism of Xeon Phi achieves significant speed-ups.


See also[edit]


  1. ^ a b Radek (June 18, 2012). "Chip Shot: Intel Names the Technology to Revolutionize the Future of HPC - Intel® Xeon® Phi™ Product Family". Intel. Retrieved December 12, 2012. 
  2. ^ "Intel Delivers New Architecture for Discovery with Intel® Xeon Phi™ Coprocessors". Intel. Retrieved June 21, 2013. 
  3. ^ "The Green500 List - November 2012". Green500. Retrieved June 21, 2013. 
  4. ^ a b "TOP500 - June 2013". TOP500 - June 2013. TOP500. Retrieved June 18, 2013. 
  5. ^ a b "Intel Powers the World's Fastest Supercomputer, Reveals New and Future High Performance Computing Technologies". Retrieved June 21, 2013. 
  6. ^ Charlie Demerjian (July 3, 2006), "New from Intel: It's Mini-Cores!", www.theinquirer.net, The Inquirer 
  7. ^ Seiler, L.; Cavin, D.; Espasa, E.; Grochowski, T.; Juan, M.; Hanrahan, P.; Carmean, S.; Sprangle, A.; Forsyth, J.; Abrash, R.; Dubey, R.; Junkins, E.; Lake, T.; Sugerman, P. (August 2008). "Larrabee: A Many-Core x86 Architecture for Visual Computing" (PDF). ACM Transactions on Graphics. Proceedings of ACM SIGGRAPH 2008. 27 (3): 18:11–18:11. doi:10.1145/1360612.1360617. ISSN 0730-0301. Retrieved 2008-08-06. 
  8. ^ Tom Forsyth, "SIMD Programming with Larrabee" (PDF), www.stanford.edu, Intel 
  9. ^ Ryan Smith (May 25, 2010), "Intel Kills Larrabee GPU, Will Not Bring a Discrete Graphics Product to Market\", www.anandtech.com, AnandTech 
  10. ^ Tony Bradley (December 3, 2009), "Intel 48-Core "Single-Chip Cloud Computer" Improves Power Efficiency", www.pcworld.com, PCWorld 
  11. ^ "Intel Research : Single-Chip Cloud Computer", techresearch.intel.com, Intel 
  12. ^ Ben Ames (February 11, 2007), "Intel Tests Chip Design With 80-Core Processor", www.pcworld.com, IDG News 
  13. ^ "Intel Details 80-Core Teraflops Research Chip - X-bit labs". xbitlabs.com. Retrieved August 27, 2015. 
  14. ^ "Intel's Teraflops Research Chip" (PDF), download.intel.com, Intel 
  15. ^ Anton Shilov (February 12, 2007), "Intel Details 80-Core Teraflops Research Chip", www.xbitlabs.com, Xbit laboratories 
  16. ^ a b Rupert Goodwins (June 1, 2010), "Intel unveils many-core Knights platform for HPC", www.zdnet.co.uk, ZDNet 
  17. ^ "Intel News Release : Intel Unveils New Product Plans dor High-Performance Computing", www.intel.com, Intel, May 31, 2010 
  18. ^ a b Mike Giles (June 24, 2010), "Runners and riders in GPU steeplechase" (PDF), people.maths.ox.ac.uk, pp. 8–10 
  19. ^ a b c d Gareth Halfacree (June 20, 2011), "Intel pushes for HPC space with Knights Corner", www.thinq.co.uk, Net Communities Limited, UK 
  20. ^ "Intel Many Integrated Core Architecture" (PDF), www.many-core.group.cam.ac.uk, Intel, December 2010 
  21. ^ Rick Merritt (June 20, 2011), "OEMs show systems with Intel MIC chips", www.eetimes.com, EE Times 
  22. ^ Tom R. Halfhill (July 18, 2011), "Intel Shows MIC Progress", www.linleygroup.com, The Linley Group 
  23. ^ Andrea Petrou (June 20, 2011), "SGI wants Intel for super supercomputer", news.techeye.net 
  24. ^ ""Stampede's" Comprehensive Capabilities to Bolster U.S. Open Science Computational Resources", www.tacc.utexas.edu, Texas Advanced Computing Center, September 22, 2011 
  25. ^ a b "Stampede: A Comprehensive Petascale Computing Environment" (PDF). IEEE Cluster 2011 Special Topic. Retrieved November 16, 2011. 
  26. ^ Yam, Marcus (November 16, 2011), "Intel's Knights Corner: 50+ Core 22nm Co-processor", www.tomshardware.com, Tom's Hardware, retrieved November 16, 2011 
  27. ^ Sylvie Barak (November 16, 2011), "Intel unveils 1 TFLOP/s Knights Corner", www.eetimes.com, EE Times, retrieved November 16, 2011 
  28. ^ James Reinders (June 5, 2012), Knights Corner: Open source software stack, Intel 
  29. ^ Prickett Morgan, Timothy (June 18, 2012), "Intel slaps Xeon Phi brand on MIC coprocessors", 222.theregister.co.uk 
  30. ^ Intel Corporation (June 18, 2012), "Latest Intel(R) Xeon(R) Processors E5 Product Family Achieves Fastest Adoption of New Technology on Top500 List", www.marketwatch.com, Intel(R) Xeon(R) Phi(TM) is the new brand name for all future Intel(R) Many Integrated Core Architecture based products targeted at HPC, enterprise, datacenters and workstations. The first Intel(R) Xeon(R) Phi(TM) product family member is scheduled for volume production by the end of 2012 
  31. ^ Raj Hazra (June 18, 2012). "Intel Xeon Phi coprocessors accelerate the pace of discovery and innovation". Intel. Retrieved December 12, 2012. 
  32. ^ Rick Merritt (June 18, 2012). "Cray will use Intel MIC, branded Xeon Phi". EETimes. Retrieved December 12, 2012. 
  33. ^ Terrence O'Brien (June 18, 2012). "Intel christens its 'Many Integrated Core' products Xeon Phi, eyes exascale milestone". Engadget. Retrieved December 12, 2012. 
  34. ^ Jeffrey Burt (June 18, 2012). "Intel Wraps Xeon Phi Branding Around MIC Coprocessors". EWeek. Retrieved December 12, 2012. 
  35. ^ Merritt, Rick (June 8, 2012), "Cray will use Intel MIC, branded Xeon Phi", www.eetimes.com 
  36. ^ Latif, Lawrence (June 19, 2012), "Cray to support Intel's Xeon Phi in Cascade clusters", www.theinquirer.net 
  37. ^ "ScaleMP vSMP Foundation to Support Intel Xeon Phi", www.ScaleMP.com, ScaleMP, June 20, 2012 
  38. ^ https://software.intel.com/en-us/articles/intel-xeon-phi-coprocessor-codename-knights-corner
  39. ^ a b c IntelPR (November 12, 2012). "Intel Delivers New Architecture for Discovery with Intel® Xeon Phi™ Coprocessors". Intel. Retrieved December 12, 2012. 
  40. ^ a b c Agam Shah (November 12, 2012). "Intel ships 60-core Xeon Phi processor". Computerworld. Retrieved December 12, 2012. 
  41. ^ a b c Johan De Gelas (November 14, 2012). "The Xeon Phi at work at TACC". AnandTech. Retrieved December 12, 2012. 
  42. ^ "Tianhe-2 (MilkyWay-2)". Top500.org. November 14, 2015. Retrieved May 6, 2016. 
  43. ^ "Intel Xeon Phi 'Knights Landing' Features Integrated Memory With 500 GB/s Bandwidth and DDR4 Memory Support - Architecture Detailed". WCCFtech. Retrieved August 27, 2015. 
  44. ^ Sebastian Anthony (November 26, 2013), Intel unveils 72-core x86 Knights Landing CPU for exascale supercomputing, ExtremeTech 
  45. ^ Tom's Hardware: Intel Xeon Phi Knights Landing Now Shipping; Omni Path Update, Too. June 20, 2016
  46. ^ James Reinders (July 23, 2013), AVX-512 Instructions, Intel 
  47. ^ http://www.nersc.gov/users/computational-systems/cori
  48. ^ a b 2016 ISC High Performance: Intel's Rajeeb Hazra Delivers Keynote Address
  49. ^ How Intel® Xeon Phi™ Processors Benefit Machine Learning/Deep Learning Apps and Frameworks
  50. ^ Introducing the Intel® Xeon Phi™ Processor – Your Path to Deeper Insight
  51. ^ Eric Gardner (November 25, 2014), What public disclosures has Intel made about Knights Landing?, Intel Corporation 
  52. ^ ALCF staff (April 9, 2015), Introducing Aurora 
  53. ^ ALCF staff (April 9, 2015), Aurora 
  54. ^ Smith, Ryan (17 August 2016). "Intel Announces Knight's Mill: A Xeon Phi for Deep Learning". Anandtech. Retrieved 17 August 2016. 
  55. ^ Hruska, Joel (July 30, 2012). "Intel's 50-core champion: In-depth on Xeon Phi". ExtremeTech. Ziff Davis, Inc. Retrieved December 2, 2012. 
  56. ^ Rick Merritt (June 20, 2011), "OEMs show systems with Intel MIC chips", www.eetimes.com, EE Times 
  57. ^ Efficient Hybrid Execution of C++ Applications using Intel(R) Xeon Phi(TM) Coprocessor, November 23, 2012, arXiv:1211.5530free to read 
  58. ^ "News Fact Sheet: Intel Many Integrated Core (Intel MIC) Architecture ISC'11 Demos and Performance Description" (PDF), newsroom.intel.com, Intel, June 20, 2011, archived from the original (PDF) on 24 March 2012 
  59. ^ Tesla vs. Xeon Phi vs. Radeon. A Compiler Writer’s Perspective // The Portland Group (PGI), CUG 2013 Proceedings
  60. ^ "Intel® Many Integrated Core Architecture (Intel MIC Architecture) - RESOURCES (including downloads)". Intel. Retrieved January 6, 2014. 
  61. ^ "Intel Xeon Phi Coprocessor Instruction Set Architecture Reference Manual" (PDF). Intel. September 7, 2012. Retrieved January 6, 2014. 
  62. ^ "Intel® Developer Zone: Intel Xeon Phi Coprocessor". Intel. Retrieved January 6, 2014. 
  63. ^ Fang, Jianbin; Sips, Henk; Zhang, Lilun; Xu, Chuanfu; Yonggang, Che; Varbanescu, Ana Lucia (2014). "Test-Driving Intel Xeon Phi" (PDF). Retrieved December 30, 2013. 
  64. ^ Accelerating DNA Sequence Analysis using Intel Xeon Phi, June 29, 2015, arXiv:1506.08612free to read 
  65. ^ The Potential of the Intel Xeon Phi for Supervised Deep Learning, June 30, 2015, arXiv:1506.09067free to read 
  66. ^ Margetts, L.; Arregui-Mena, J.D.; Hewitt, W.T.; Mason, L. (2–3 June 2016). Parallel finite element analysis using the Intel Xeon Phi. Emerging Technology Conference EMiT2016. Barcelona, Spain. 
  67. ^ Jon Stokes (June 20, 2011). "Intel takes wraps off 50-core supercomputing processor plans". Ars Technica. 
  68. ^ Johan De Gelas. "Conclusions - Intel's Xeon E5-2600 V2: 12-core Ivy Bridge EP for Servers". anandtech.com. Retrieved August 27, 2015. 
  69. ^ a b Johan De Gelas (September 11, 2012). "Intel's Xeon Phi in 10 Petaflops supercomputer". AnandTech. Retrieved December 12, 2012. 

External links[edit]