topology detection
play

topology detection Screenshot (und Demo?) hwloc /proc/cpuinfo - PowerPoint PPT Presentation

May 03, 2023 •33 likes •323 views

topology detection Screenshot (und Demo?) hwloc /proc/cpuinfo processor : 0 vendor_id : GenuineIntel model name : Intel(R) Xeon(R) CPU E5-2620 0 @ 2.00GHz cache size : 15360 KB physical id : 0 siblings :

  1. topology detection

  2. ● Screenshot (und Demo?) hwloc

  3. /proc/cpuinfo processor : 0 vendor_id : GenuineIntel model name : Intel(R) Xeon(R) CPU E5-2620 0 @ 2.00GHz cache size : 15360 KB physical id : 0 siblings : 12 core id : 0 cpu cores : 6 flags : fpu vme de pse … processor : 1 vendor_id : GenuineIntel model name : Intel(R) Xeon(R) CPU E5-2620 0 @ 2.00GHz cache size : 15360 KB physical id : 0 siblings : 12 core id : 1 cpu cores : 6 flags : fpu vme de pse …

  4. /proc/cpuinfo processor : 0 vendor_id : GenuineIntel model name : Intel(R) Xeon(R) CPU E5-2620 0 @ 2.00GHz cache size : 15360 KB physical id : 0 siblings : 12 core id : 0 cpu cores : 6 flags : fpu vme de pse … processor : 1 vendor_id : GenuineIntel model name : Intel(R) Xeon(R) CPU E5-2620 0 @ 2.00GHz cache size : 15360 KB physical id : 0 siblings : 12 core id : 1 cpu cores : 6 flags : fpu vme de pse …

  5. numactl --hardware node 0 cpus: 0 1 2 3 4 5 12 13 14 15 16 17 node 0 size: 32140 MB node 0 free: 28841 MB node 1 cpus: 6 7 8 9 10 11 18 19 20 21 22 23 node 1 size: 32251 MB node 1 free: 30743 MB node distances: node 0 1 0: 10 20 1: 20 10

  6. numactl --hardware node 0 cpus: 0 1 2 3 4 5 12 13 14 15 16 17 node 0 size: 32140 MB node 0 free: 28841 MB node 1 cpus: 6 7 8 9 10 11 18 19 20 21 22 23 node 1 size: 32251 MB node 1 free: 30743 MB node distances: node 0 1 0: 10 20 1: 20 10

  7. numactl --hardware [...] node distances: node 0 1 2 3 4 5 6 7 0: 10 12 17 17 19 19 19 19 1: 12 10 17 17 19 19 19 19 2: 17 17 10 12 19 19 19 19 3: 17 17 12 10 19 19 19 19 4: 19 19 19 19 10 12 17 17 5: 19 19 19 19 12 10 17 17 6: 19 19 19 19 17 17 10 12 7: 19 19 19 19 17 17 12 10

  8. ● /sys/devices/system/node/node* ● /sys/devices/system/node/node*/distance ● /sys/devices/system/node/node*/cpu*

  9. Information sources ● ACPI – System Resource Affinity Table (SRAT) ● NUMA nodes (proximity domains), RAM, Interrupt controllers, – System Locality Distance Information Table (SLIT) ● only relative information about "distances"

  10. estimate hardware topology [...] node distances: node 0 1 2 3 4 5 6 7 0: 10 12 17 17 19 19 19 19 1: 12 10 17 17 19 19 19 19 2: 17 17 10 12 19 19 19 19 3: 17 17 12 10 19 19 19 19 4: 19 19 19 19 10 12 17 17 5: 19 19 19 19 12 10 17 17 6: 19 19 19 19 17 17 10 12 7: 19 19 19 19 17 17 12 10

  11. estimate hardware topology [...] node distances: node 0 1 2 3 4 5 6 7 0: 10 12 17 17 19 19 19 19 1: 12 10 17 17 19 19 19 19 2: 17 17 10 12 19 19 19 19 3: 17 17 12 10 19 19 19 19 4: 19 19 19 19 10 12 17 17 5: 19 19 19 19 12 10 17 17 6: 19 19 19 19 17 17 10 12 7: 19 19 19 19 17 17 12 10 0 1 2 3 4 5 6 7

  12. estimate hardware topology [...] node distances: node 0 1 2 3 4 5 6 7 0: 10 12 17 17 19 19 19 19 1: 12 10 17 17 19 19 19 19 2: 17 17 10 12 19 19 19 19 3: 17 17 12 10 19 19 19 19 4: 19 19 19 19 10 12 17 17 5: 19 19 19 19 12 10 17 17 6: 19 19 19 19 17 17 10 12 7: 19 19 19 19 17 17 12 10 0 2 4 6 1 3 5 7

  13. estimate hardware topology [...] node distances: node 0 1 2 3 4 5 6 7 0: 10 12 17 17 19 19 19 19 1: 12 10 17 17 19 19 19 19 2: 17 17 10 12 19 19 19 19 3: 17 17 12 10 19 19 19 19 4: 19 19 19 19 10 12 17 17 5: 19 19 19 19 12 10 17 17 6: 19 19 19 19 17 17 10 12 7: 19 19 19 19 17 17 12 10 0 2 4 6 1 3 5 7

  14. estimate hardware topology [...] node distances: node 0 1 2 3 4 5 6 7 0: 10 12 17 17 19 19 19 19 1: 12 10 17 17 19 19 19 19 2: 17 17 10 12 19 19 19 19 3: 17 17 12 10 19 19 19 19 4: 19 19 19 19 10 12 17 17 5: 19 19 19 19 12 10 17 17 6: 19 19 19 19 17 17 10 12 7: 19 19 19 19 17 17 12 10 ? 0 2 1 3 4 6 5 7

  15. actual hardware topology? ● Wasn't there only max 4 QPI links per CPU? ● $cat /sys/devices/virtual/dmi/id/product_name ProLiant DL980 G7

  16. actual hardware topology? ● Wasn't there only max 4 QPI links per CPU? ● $cat /sys/devices/virtual/dmi/id/product_name ProLiant DL980 G7

  17. actual hardware topology [...] node distances: node 0 1 2 3 4 5 6 7 0: 10 12 17 17 19 19 19 19 1: 12 10 17 17 19 19 19 19 2: 17 17 10 12 19 19 19 19 3: 17 17 12 10 19 19 19 19 4: 19 19 19 19 10 12 17 17 5: 19 19 19 19 12 10 17 17 6: 19 19 19 19 17 17 10 12 7: 19 19 19 19 17 17 12 10

  18. Absolute measurements ● mlc -e

  19. Absolute measurements Raw Bandwidth values (MB/s) Socket 0 1 2 3 4 5 6 7 0 25368,9 10882,8 11203,2 11201,7 11179,1 11175,9 11175,6 11177,8 1 10858,8 25524,8 11184,3 11186,9 11168,0 11183,7 11175,1 11165,0 2 11165,7 11211,5 25605,7 10877,3 11182,6 11173,5 11175,9 11182,9 3 11181,9 11207,4 10876,2 25558,8 11176,5 11167,7 11169,2 11174,5 4 11119,3 11166,1 11165,6 11169,0 25540,0 10862,3 11186,2 11190,0 5 11140,0 11179,1 11182,0 11172,1 10867,8 25553,2 11187,6 11192,8 6 11153,3 11191,1 11193,1 11189,3 11212,0 11208,4 25633,1 10900,8 7 11165,8 11196,0 11195,1 11187,2 11210,4 11210,6 10920,2 25634,7

  20. Absolute measurements Raw Bandwidth values (MB/s) → non-direct connections very similar speeds Socket 0 1 2 3 4 5 6 7 0 25368,9 10882,8 11203,2 11201,7 11179,1 11175,9 11175,6 11177,8 1 10858,8 25524,8 11184,3 11186,9 11168,0 11183,7 11175,1 11165,0 2 11165,7 11211,5 25605,7 10877,3 11182,6 11173,5 11175,9 11182,9 3 11181,9 11207,4 10876,2 25558,8 11176,5 11167,7 11169,2 11174,5 4 11119,3 11166,1 11165,6 11169,0 25540,0 10862,3 11186,2 11190,0 5 11140,0 11179,1 11182,0 11172,1 10867,8 25553,2 11187,6 11192,8 6 11153,3 11191,1 11193,1 11189,3 11212,0 11208,4 25633,1 10900,8 7 11165,8 11196,0 11195,1 11187,2 11210,4 11210,6 10920,2 25634,7

  21. Absolute measurements Idle latency Soc 0 1 2 3 4 5 6 7 0 37.30 49.90 74.70 73.50 84.40 82.60 80.30 82.00 1 49.40 36.30 73.80 76.90 85.00 82.30 84.40 86.10 2 75.00 75.60 35.00 46.40 75.30 75.40 78.10 81.80 3 69.60 67.60 47.30 35.10 83.50 81.70 78.20 81.50 4 77.10 78.60 77.50 78.90 34.80 48.90 70.00 75.30 5 79.80 76.50 79.80 80.90 46.20 35.10 69.20 69.80 6 75.70 74.10 80.00 77.10 67.60 67.80 35.10 46.50 7 83.70 84.00 84.00 82.70 69.90 70.30 47.40 34.90 Normalisiert: SLIT (expected) vs measured 10 10 12 11,1 17 22,7 19 25,5

  22. Absolute measurements Idle latency Soc 0 1 2 3 4 5 6 7 0 37.30 49.90 74.70 73.50 84.40 82.60 80.30 82.00 1 49.40 36.30 73.80 76.90 85.00 82.30 84.40 86.10 2 75.00 75.60 35.00 46.40 75.30 75.40 78.10 81.80 3 69.60 67.60 47.30 35.10 83.50 81.70 78.20 81.50 4 77.10 78.60 77.50 78.90 34.80 48.90 70.00 75.30 5 79.80 76.50 79.80 80.90 46.20 35.10 69.20 69.80 6 75.70 74.10 80.00 77.10 67.60 67.80 35.10 46.50 7 83.70 84.00 84.00 82.70 69.90 70.30 47.40 34.90 SLIT (expected) vs measured (fastest normalized to 10) 10 10 12 11,1 Close...ish ? 17 22,7 19 25,5

  23. NUMA and IO ● not only memory is non-uniform ● IO controllers are connected to only some nodes

  24. NUMA and IO ● PCIe 4.0: 2 GB/s per lane vs QPI: 12,8 GB/s per link ● 40 GbE cards, access to GPU memory utilize these speeds ● example applications: network processing, clusters

  27. Identifying IO nodes ● lspci -tv, then look up in /sys/bus/... ● Why is our machine structured like it is?

  28. future work ● do extended measurements on 8-node machine ● utilize QPI link information for discovery

  29. http://www.acpi.info/DOWNLOADS/ACPIspec40a.pdf ● http://h50146.www5.hp.com/products/software/oe/linux/mainstream/sup ● port/whitepaper/pdfs/c03261871.pdf http://on-demand.gputechconf.com/gtc/2014/presentations/S4589- ● openmpi-rdma-support-cuda.pdf man numa, numactl, lstopo ● http://www.ntop.org/pf_ring/not-all-servers-are-alike-with-pf_ring-zcdna- ● part-3/

Recommend

1 Star topology Bus topology popular through mid 90s Now star topology prevails

1 Star topology Bus topology popular through mid 90s Now star topology prevails

The Data Link Layer Last time Our goals: link layer services understand principles error detection, correction behind data link layer services: multiple access protocols and LANs error detection, correction link layer

876 views • 17 slides
Web Spam Know Your Neighbors: Web Spam Detection using the Web Topology Presenter: Sadia Masood

Web Spam Know Your Neighbors: Web Spam Detection using the Web Topology Presenter: Sadia Masood

Seminar: Future Of Web Search University of Saarland Web Spam Know Your Neighbors: Web Spam Detection using the Web Topology Presenter: Sadia Masood Tutor : Klaus Berberich Date : 17-Jan-2008 The Agenda Focus of the First Paper

1.22k views • 53 slides
WHAT DO WE KNOW ABOUT THE TOPOLOGY? WHAT DO WE KNOW ABOUT THE TOPOLOGY? Number of nodes and

WHAT DO WE KNOW ABOUT THE TOPOLOGY? WHAT DO WE KNOW ABOUT THE TOPOLOGY? Number of nodes and

TxProbe: Discovering Bitcoins Network Topology Using Orphan Transactions Sergi Delgado-Segura , Surya Bakshi, Cristina Prez-Sol, James Litton, Andrew Pachulski, Andrew Miller and Bobby Bhattacharjee sr_gi WHAT DO WE KNOW ABOUT THE TOPOLOGY?

1.84k views • 162 slides
Combinatorics and topology of toric arrangements II. Topology of arrangements in the complex torus

Combinatorics and topology of toric arrangements II. Topology of arrangements in the complex torus

Combinatorics and topology of toric arrangements II. Topology of arrangements in the complex torus (An invitation to combinatorial algebraic topology) Emanuele Delucchi (SNSF / Universit e de Fribourg) Toblach/Dobbiaco February 23, 2017

543 views • 42 slides
Order Topology Definition Let ( X , < ) be an ordered set. Then the order topology on X is the

Order Topology Definition Let ( X , < ) be an ordered set. Then the order topology on X is the

Order Topology Definition Let ( X , < ) be an ordered set. Then the order topology on X is the topology generated by the basis consisting of unions of sets of the form 1 Open intervals of the form ( a , b ) with a < b in X . 2 If X has a

62 views • 3 slides
Topological data analysis and topology-based visualization Leila De Floriani Topology-based

Topological data analysis and topology-based visualization Leila De Floriani Topology-based

Topological data analysis and topology-based visualization Leila De Floriani Topology-based methods v Why topology? topology deals with qualitative geometric information , thus v providing a structural description of data topological

298 views • 10 slides
& OTHER QUESTIONS: EQUAL A COFFEE MUG? WHAT IS TOPOLOGY? SARAH BLACKWELL WHAT IS TOPOLOGY?

& OTHER QUESTIONS: EQUAL A COFFEE MUG? WHAT IS TOPOLOGY? SARAH BLACKWELL WHAT IS TOPOLOGY?

WHY DOES A DONUT & OTHER QUESTIONS: EQUAL A COFFEE MUG? WHAT IS TOPOLOGY? SARAH BLACKWELL WHAT IS TOPOLOGY? WHAT DOES IT MEAN FOR TWO THINGS TO BE THE SAME? 1 = 7 (mod 6) color versus colour congruent or similar triangles

542 views • 30 slides
ALM API for Topology Management ALM API for Topology Management and Network Layer Transparent

ALM API for Topology Management ALM API for Topology Management and Network Layer Transparent

ALM API for Topology Management ALM API for Topology Management and Network Layer Transparent Multimedia Transport <draft lim irtf sam alm api 00.txt > p 71 ST IETF SAM RG MEETING, PHILADELPHIA. Lim Boon Ping

231 views • 19 slides
WHAT DO WE KNOW ABOUT THE TOPOLOGY? WHAT DO WE KNOW ABOUT THE TOPOLOGY? Number of nodes and

WHAT DO WE KNOW ABOUT THE TOPOLOGY? WHAT DO WE KNOW ABOUT THE TOPOLOGY? Number of nodes and

# SFBW19 TxProbe: Discovering Bitcoins Network Topology Using Orphan Transactions Sergi Delgado-Segura , Surya Bakshi, Cristina Prez-Sol, James Litton, Andrew Pachulski, Andrew Miller and Bobby Bhattacharjee sr_gi WHAT DO WE KNOW ABOUT THE

1.96k views • 168 slides
1 Ring Topology Ring Topology In a ring network, every device has exactly two neighbours

1 Ring Topology Ring Topology In a ring network, every device has exactly two neighbours

What is a Topology? Network topologies describe the ways in which the elements of a network are mapped. They describe the physical and logical arrangement of the network nodes. The physical topology of a network refers to the

279 views • 4 slides
Hard Problems in 3-Manifold Topology School on Low-Dimensional Geometry and Topology: Discrete

Hard Problems in 3-Manifold Topology School on Low-Dimensional Geometry and Topology: Discrete

Hard Problems in 3-Manifold Topology School on Low-Dimensional Geometry and Topology: Discrete and Algorithmic Aspects Arnaud de Mesmay 1 Yoav Rieck 2 Eric Sedgwick 3 Martin Tancer 4 1 CNRS, GIPSA-Lab 2 University of Arkansas 3 DePaul

1.37k views • 83 slides
Detection of neutral particles detection of neutrons detection of neutrinons detection of low

Detection of neutral particles detection of neutrons detection of neutrinons detection of low

Detection of neutral particles detection of neutrons detection of neutrinons detection of low energy photons (detection of high energy photons calorimeters) Peter Krizan, Neutron and neutrino detection Detection of neutral particles

1.21k views • 67 slides
On Evolution of On Evolution of C2 Netw ork Topology C2 Netw ork Topology IC ICCR CRTS 2010

On Evolution of On Evolution of C2 Netw ork Topology C2 Netw ork Topology IC ICCR CRTS 2010

On Evolution of On Evolution of C2 Netw ork Topology C2 Netw ork Topology IC ICCR CRTS 2010 TS 2010 Paper 082 National Key Laboratory of Science and Technology on C4ISR Dr. Songhua Huang NRIEE, CETC, China Introduction C2 netw orks

458 views • 16 slides
The Smooth Rectangular Peg Problem Kyle Hayden September 23, 2020 Topology Topology concerns

The Smooth Rectangular Peg Problem Kyle Hayden September 23, 2020 Topology Topology concerns

The Smooth Rectangular Peg Problem Kyle Hayden September 23, 2020 Topology Topology concerns the properties of objects/spaces that are preserved un- der continuous deformations, such as stretching, twisting, crumpling and bending, but not

242 views • 20 slides
Topology API Use Cases

Topology API Use Cases

Topology API Use Cases dra$-amante-i2rs-topology-use-cases-01 Shane Amante Jan Medved, jmedved@cisco.com Stefano Previdi, sprevidi@cisco.com Victor

424 views • 14 slides
Pizzas, Bagels, Pretzels, and Euler's Magical ---- an informal introduction to topology What

Pizzas, Bagels, Pretzels, and Euler's Magical ---- an informal introduction to topology What

Pizzas, Bagels, Pretzels, and Euler's Magical ---- an informal introduction to topology What is topology? Given a set X , a topology on X is a collection T of subsets of X , satisfying the following axioms: 1. The empty set and X are in T . 2.

946 views • 38 slides
Multiple Source Multiple Destination Topology Inference Destination Topology Inference using

Multiple Source Multiple Destination Topology Inference Destination Topology Inference using

Multiple Source Multiple Destination Topology Inference Destination Topology Inference using Network Coding Pegah Sattari EECS, UC Irvine Joint work with Athina Markopoulou, at UCI, Christina Fragouli at EPFL Lausanne Christina Fragouli,

503 views • 36 slides
I2RS Service Topology Draft-hares-i2rs-service-topo-dm-05 I2RS Service Topology Model Why

I2RS Service Topology Draft-hares-i2rs-service-topo-dm-05 I2RS Service Topology Model Why

I2RS Service Topology Draft-hares-i2rs-service-topo-dm-05 I2RS Service Topology Model Why discuss today? Completes the generic topology model I2RS Service Model Aligns with L3SM? Recent Design Team draft yang model

463 views • 12 slides
Metric Spaces Definition If d is a metric on X , then the metric topology on X induced by d is

Metric Spaces Definition If d is a metric on X , then the metric topology on X induced by d is

Metric Spaces Definition If d is a metric on X , then the metric topology on X induced by d is the topology generated by the basis = { B d ( X , ) : x X and > 0 } . The set X endowed with this topology is called the metric space (

52 views • 4 slides
NETWORK LAYER INTERNET TOPOLOGY CONSTRUCTION M. Engin Tozal & Kamil Sarac The University of

NETWORK LAYER INTERNET TOPOLOGY CONSTRUCTION M. Engin Tozal & Kamil Sarac The University of

1 NETWORK LAYER INTERNET TOPOLOGY CONSTRUCTION M. Engin Tozal & Kamil Sarac The University of Texas at Dallas AGENDA 2 Internet Topology and its Representation Existing Topology Representations Subnet Level Topologies

594 views • 38 slides
Hard Problems in 3-Manifold Topology Einstein Workshop on Discrete Geometry and Topology Arnaud de

Hard Problems in 3-Manifold Topology Einstein Workshop on Discrete Geometry and Topology Arnaud de

Hard Problems in 3-Manifold Topology Einstein Workshop on Discrete Geometry and Topology Arnaud de Mesmay 1 Yoav Rieck 2 Eric Sedgwick 3 Martin Tancer 4 1 CNRS, GIPSA-Lab 2 University of Arkansas 3 DePaul University 4 Charles University Eric

862 views • 60 slides
Ch01. Point-Set Topology and Calculus Ping Yu Faculty of Business and Economics The University

Ch01. Point-Set Topology and Calculus Ping Yu Faculty of Business and Economics The University

Ch01. Point-Set Topology and Calculus Ping Yu Faculty of Business and Economics The University of Hong Kong Ping Yu (HKU) Topology and Calculus 1 / 51 Sets and Set Operations 1 Functions 2 Point-Set Topology in the Euclidean Space 3

953 views • 52 slides
Router Microarchitecture and Scalability of Ring Topology in On-Chip Networks John Kim, Hanjoon

Router Microarchitecture and Scalability of Ring Topology in On-Chip Networks John Kim, Hanjoon

Router Microarchitecture and Scalability of Ring Topology in On-Chip Networks John Kim, Hanjoon Kim Department of Computer Science KAIST Ring Router Microarchitecture NoCArc09 1 Topology Topology efficiently exploits the available

366 views • 25 slides
Geometry and Topology, Lecture 4 The fundamental group and covering spaces Text: Andrew Ranicki

Geometry and Topology, Lecture 4 The fundamental group and covering spaces Text: Andrew Ranicki

1 Geometry and Topology, Lecture 4 The fundamental group and covering spaces Text: Andrew Ranicki (Edinburgh) Pictures: Julia Collins (Edinburgh) 8th November, 2007 2 The method of algebraic topology Algebraic topology uses algebra to

383 views • 36 slides

More recommend