Understanding Data Movement in AMD Multi-GPU Systems with Infinity Fabric Article Swipe
Gabin Schieffer
,
R. S. Shi
,
Stefano Markidis
,
Andreas Herten
,
Jennifer Faj
,
Ivy Peng
·
YOU?
·
· 2024
· Open Access
·
· DOI: https://doi.org/10.48550/arxiv.2410.00801
YOU?
·
· 2024
· Open Access
·
· DOI: https://doi.org/10.48550/arxiv.2410.00801
Modern GPU systems are constantly evolving to meet the needs of computing-intensive applications in scientific and machine learning domains. However, there is typically a gap between the hardware capacity and the achievable application performance. This work aims to provide a better understanding of the Infinity Fabric interconnects on AMD GPUs and CPUs. We propose a test and evaluation methodology for characterizing the performance of data movements on multi-GPU systems, stressing different communication options on AMD MI250X GPUs, including point-to-point and collective communication, and memory allocation strategies between GPUs, as well as the host CPU. In a single-node setup with four GPUs, we show that direct peer-to-peer memory accesses between GPUs and utilization of the RCCL library outperform MPI-based solutions in terms of memory/communication latency and bandwidth. Our test and evaluation method serves as a base for validating memory and communication strategies on a system and improving applications on AMD multi-GPU computing systems.
Related Topics
Concepts
Metadata
- Type
- preprint
- Language
- en
- Landing Page
- http://arxiv.org/abs/2410.00801
- https://arxiv.org/pdf/2410.00801
- OA Status
- green
- Related Works
- 10
- OpenAlex ID
- https://openalex.org/W4403821443
All OpenAlex metadata
Raw OpenAlex JSON
- OpenAlex ID
-
https://openalex.org/W4403821443Canonical identifier for this work in OpenAlex
- DOI
-
https://doi.org/10.48550/arxiv.2410.00801Digital Object Identifier
- Title
-
Understanding Data Movement in AMD Multi-GPU Systems with Infinity FabricWork title
- Type
-
preprintOpenAlex work type
- Language
-
enPrimary language
- Publication year
-
2024Year of publication
- Publication date
-
2024-10-01Full publication date if available
- Authors
-
Gabin Schieffer, R. S. Shi, Stefano Markidis, Andreas Herten, Jennifer Faj, Ivy PengList of authors in order
- Landing page
-
https://arxiv.org/abs/2410.00801Publisher landing page
- PDF URL
-
https://arxiv.org/pdf/2410.00801Direct link to full text PDF
- Open access
-
YesWhether a free full text is available
- OA status
-
greenOpen access status per OpenAlex
- OA URL
-
https://arxiv.org/pdf/2410.00801Direct OA link when available
- Concepts
-
Infinity, Computer science, Movement (music), Parallel computing, Computer graphics (images), Computational science, Mathematics, Physics, Mathematical analysis, AcousticsTop concepts (fields/topics) attached by OpenAlex
- Cited by
-
0Total citation count in OpenAlex
- Related works (count)
-
10Other works algorithmically related by OpenAlex
Full payload
| id | https://openalex.org/W4403821443 |
|---|---|
| doi | https://doi.org/10.48550/arxiv.2410.00801 |
| ids.doi | https://doi.org/10.48550/arxiv.2410.00801 |
| ids.openalex | https://openalex.org/W4403821443 |
| fwci | |
| type | preprint |
| title | Understanding Data Movement in AMD Multi-GPU Systems with Infinity Fabric |
| biblio.issue | |
| biblio.volume | |
| biblio.last_page | |
| biblio.first_page | |
| topics[0].id | https://openalex.org/T10054 |
| topics[0].field.id | https://openalex.org/fields/17 |
| topics[0].field.display_name | Computer Science |
| topics[0].score | 0.9857000112533569 |
| topics[0].domain.id | https://openalex.org/domains/3 |
| topics[0].domain.display_name | Physical Sciences |
| topics[0].subfield.id | https://openalex.org/subfields/1708 |
| topics[0].subfield.display_name | Hardware and Architecture |
| topics[0].display_name | Parallel Computing and Optimization Techniques |
| topics[1].id | https://openalex.org/T10715 |
| topics[1].field.id | https://openalex.org/fields/17 |
| topics[1].field.display_name | Computer Science |
| topics[1].score | 0.9718999862670898 |
| topics[1].domain.id | https://openalex.org/domains/3 |
| topics[1].domain.display_name | Physical Sciences |
| topics[1].subfield.id | https://openalex.org/subfields/1705 |
| topics[1].subfield.display_name | Computer Networks and Communications |
| topics[1].display_name | Distributed and Parallel Computing Systems |
| topics[2].id | https://openalex.org/T10829 |
| topics[2].field.id | https://openalex.org/fields/17 |
| topics[2].field.display_name | Computer Science |
| topics[2].score | 0.9700999855995178 |
| topics[2].domain.id | https://openalex.org/domains/3 |
| topics[2].domain.display_name | Physical Sciences |
| topics[2].subfield.id | https://openalex.org/subfields/1705 |
| topics[2].subfield.display_name | Computer Networks and Communications |
| topics[2].display_name | Interconnection Networks and Systems |
| is_xpac | False |
| apc_list | |
| apc_paid | |
| concepts[0].id | https://openalex.org/C7321624 |
| concepts[0].level | 2 |
| concepts[0].score | 0.6869779825210571 |
| concepts[0].wikidata | https://www.wikidata.org/wiki/Q205 |
| concepts[0].display_name | Infinity |
| concepts[1].id | https://openalex.org/C41008148 |
| concepts[1].level | 0 |
| concepts[1].score | 0.5792186260223389 |
| concepts[1].wikidata | https://www.wikidata.org/wiki/Q21198 |
| concepts[1].display_name | Computer science |
| concepts[2].id | https://openalex.org/C2780226923 |
| concepts[2].level | 2 |
| concepts[2].score | 0.566778302192688 |
| concepts[2].wikidata | https://www.wikidata.org/wiki/Q929848 |
| concepts[2].display_name | Movement (music) |
| concepts[3].id | https://openalex.org/C173608175 |
| concepts[3].level | 1 |
| concepts[3].score | 0.43940433859825134 |
| concepts[3].wikidata | https://www.wikidata.org/wiki/Q232661 |
| concepts[3].display_name | Parallel computing |
| concepts[4].id | https://openalex.org/C121684516 |
| concepts[4].level | 1 |
| concepts[4].score | 0.3649468421936035 |
| concepts[4].wikidata | https://www.wikidata.org/wiki/Q7600677 |
| concepts[4].display_name | Computer graphics (images) |
| concepts[5].id | https://openalex.org/C459310 |
| concepts[5].level | 1 |
| concepts[5].score | 0.3433522880077362 |
| concepts[5].wikidata | https://www.wikidata.org/wiki/Q117801 |
| concepts[5].display_name | Computational science |
| concepts[6].id | https://openalex.org/C33923547 |
| concepts[6].level | 0 |
| concepts[6].score | 0.2134753167629242 |
| concepts[6].wikidata | https://www.wikidata.org/wiki/Q395 |
| concepts[6].display_name | Mathematics |
| concepts[7].id | https://openalex.org/C121332964 |
| concepts[7].level | 0 |
| concepts[7].score | 0.15324661135673523 |
| concepts[7].wikidata | https://www.wikidata.org/wiki/Q413 |
| concepts[7].display_name | Physics |
| concepts[8].id | https://openalex.org/C134306372 |
| concepts[8].level | 1 |
| concepts[8].score | 0.1173655092716217 |
| concepts[8].wikidata | https://www.wikidata.org/wiki/Q7754 |
| concepts[8].display_name | Mathematical analysis |
| concepts[9].id | https://openalex.org/C24890656 |
| concepts[9].level | 1 |
| concepts[9].score | 0.09442448616027832 |
| concepts[9].wikidata | https://www.wikidata.org/wiki/Q82811 |
| concepts[9].display_name | Acoustics |
| keywords[0].id | https://openalex.org/keywords/infinity |
| keywords[0].score | 0.6869779825210571 |
| keywords[0].display_name | Infinity |
| keywords[1].id | https://openalex.org/keywords/computer-science |
| keywords[1].score | 0.5792186260223389 |
| keywords[1].display_name | Computer science |
| keywords[2].id | https://openalex.org/keywords/movement |
| keywords[2].score | 0.566778302192688 |
| keywords[2].display_name | Movement (music) |
| keywords[3].id | https://openalex.org/keywords/parallel-computing |
| keywords[3].score | 0.43940433859825134 |
| keywords[3].display_name | Parallel computing |
| keywords[4].id | https://openalex.org/keywords/computer-graphics |
| keywords[4].score | 0.3649468421936035 |
| keywords[4].display_name | Computer graphics (images) |
| keywords[5].id | https://openalex.org/keywords/computational-science |
| keywords[5].score | 0.3433522880077362 |
| keywords[5].display_name | Computational science |
| keywords[6].id | https://openalex.org/keywords/mathematics |
| keywords[6].score | 0.2134753167629242 |
| keywords[6].display_name | Mathematics |
| keywords[7].id | https://openalex.org/keywords/physics |
| keywords[7].score | 0.15324661135673523 |
| keywords[7].display_name | Physics |
| keywords[8].id | https://openalex.org/keywords/mathematical-analysis |
| keywords[8].score | 0.1173655092716217 |
| keywords[8].display_name | Mathematical analysis |
| keywords[9].id | https://openalex.org/keywords/acoustics |
| keywords[9].score | 0.09442448616027832 |
| keywords[9].display_name | Acoustics |
| language | en |
| locations[0].id | pmh:oai:arXiv.org:2410.00801 |
| locations[0].is_oa | True |
| locations[0].source.id | https://openalex.org/S4306400194 |
| locations[0].source.issn | |
| locations[0].source.type | repository |
| locations[0].source.is_oa | True |
| locations[0].source.issn_l | |
| locations[0].source.is_core | False |
| locations[0].source.is_in_doaj | False |
| locations[0].source.display_name | arXiv (Cornell University) |
| locations[0].source.host_organization | https://openalex.org/I205783295 |
| locations[0].source.host_organization_name | Cornell University |
| locations[0].source.host_organization_lineage | https://openalex.org/I205783295 |
| locations[0].license | |
| locations[0].pdf_url | https://arxiv.org/pdf/2410.00801 |
| locations[0].version | submittedVersion |
| locations[0].raw_type | text |
| locations[0].license_id | |
| locations[0].is_accepted | False |
| locations[0].is_published | False |
| locations[0].raw_source_name | |
| locations[0].landing_page_url | http://arxiv.org/abs/2410.00801 |
| locations[1].id | doi:10.48550/arxiv.2410.00801 |
| locations[1].is_oa | True |
| locations[1].source.id | https://openalex.org/S4306400194 |
| locations[1].source.issn | |
| locations[1].source.type | repository |
| locations[1].source.is_oa | True |
| locations[1].source.issn_l | |
| locations[1].source.is_core | False |
| locations[1].source.is_in_doaj | False |
| locations[1].source.display_name | arXiv (Cornell University) |
| locations[1].source.host_organization | https://openalex.org/I205783295 |
| locations[1].source.host_organization_name | Cornell University |
| locations[1].source.host_organization_lineage | https://openalex.org/I205783295 |
| locations[1].license | |
| locations[1].pdf_url | |
| locations[1].version | |
| locations[1].raw_type | article |
| locations[1].license_id | |
| locations[1].is_accepted | False |
| locations[1].is_published | |
| locations[1].raw_source_name | |
| locations[1].landing_page_url | https://doi.org/10.48550/arxiv.2410.00801 |
| indexed_in | arxiv, datacite |
| authorships[0].author.id | https://openalex.org/A5092658397 |
| authorships[0].author.orcid | https://orcid.org/0009-0003-6504-7109 |
| authorships[0].author.display_name | Gabin Schieffer |
| authorships[0].author_position | first |
| authorships[0].raw_author_name | Schieffer, Gabin |
| authorships[0].is_corresponding | False |
| authorships[1].author.id | https://openalex.org/A5107859614 |
| authorships[1].author.orcid | |
| authorships[1].author.display_name | R. S. Shi |
| authorships[1].author_position | middle |
| authorships[1].raw_author_name | Shi, Ruimin |
| authorships[1].is_corresponding | False |
| authorships[2].author.id | https://openalex.org/A5085178088 |
| authorships[2].author.orcid | https://orcid.org/0000-0003-0639-0639 |
| authorships[2].author.display_name | Stefano Markidis |
| authorships[2].author_position | middle |
| authorships[2].raw_author_name | Markidis, Stefano |
| authorships[2].is_corresponding | False |
| authorships[3].author.id | https://openalex.org/A5028313433 |
| authorships[3].author.orcid | https://orcid.org/0000-0002-7150-2505 |
| authorships[3].author.display_name | Andreas Herten |
| authorships[3].author_position | middle |
| authorships[3].raw_author_name | Herten, Andreas |
| authorships[3].is_corresponding | False |
| authorships[4].author.id | https://openalex.org/A5091933605 |
| authorships[4].author.orcid | https://orcid.org/0000-0002-7733-6229 |
| authorships[4].author.display_name | Jennifer Faj |
| authorships[4].author_position | middle |
| authorships[4].raw_author_name | Faj, Jennifer |
| authorships[4].is_corresponding | False |
| authorships[5].author.id | https://openalex.org/A5037069204 |
| authorships[5].author.orcid | https://orcid.org/0000-0003-4158-3583 |
| authorships[5].author.display_name | Ivy Peng |
| authorships[5].author_position | last |
| authorships[5].raw_author_name | Peng, Ivy |
| authorships[5].is_corresponding | False |
| has_content.pdf | True |
| has_content.grobid_xml | True |
| is_paratext | False |
| open_access.is_oa | True |
| open_access.oa_url | https://arxiv.org/pdf/2410.00801 |
| open_access.oa_status | green |
| open_access.any_repository_has_fulltext | False |
| created_date | 2025-10-10T00:00:00 |
| display_name | Understanding Data Movement in AMD Multi-GPU Systems with Infinity Fabric |
| has_fulltext | True |
| is_retracted | False |
| updated_date | 2025-11-06T06:51:31.235846 |
| primary_topic.id | https://openalex.org/T10054 |
| primary_topic.field.id | https://openalex.org/fields/17 |
| primary_topic.field.display_name | Computer Science |
| primary_topic.score | 0.9857000112533569 |
| primary_topic.domain.id | https://openalex.org/domains/3 |
| primary_topic.domain.display_name | Physical Sciences |
| primary_topic.subfield.id | https://openalex.org/subfields/1708 |
| primary_topic.subfield.display_name | Hardware and Architecture |
| primary_topic.display_name | Parallel Computing and Optimization Techniques |
| related_works | https://openalex.org/W2306553012, https://openalex.org/W4251417359, https://openalex.org/W4229993731, https://openalex.org/W2381059884, https://openalex.org/W2372756075, https://openalex.org/W1979193983, https://openalex.org/W2998581040, https://openalex.org/W2220405813, https://openalex.org/W2610708416, https://openalex.org/W2374111230 |
| cited_by_count | 0 |
| locations_count | 2 |
| best_oa_location.id | pmh:oai:arXiv.org:2410.00801 |
| best_oa_location.is_oa | True |
| best_oa_location.source.id | https://openalex.org/S4306400194 |
| best_oa_location.source.issn | |
| best_oa_location.source.type | repository |
| best_oa_location.source.is_oa | True |
| best_oa_location.source.issn_l | |
| best_oa_location.source.is_core | False |
| best_oa_location.source.is_in_doaj | False |
| best_oa_location.source.display_name | arXiv (Cornell University) |
| best_oa_location.source.host_organization | https://openalex.org/I205783295 |
| best_oa_location.source.host_organization_name | Cornell University |
| best_oa_location.source.host_organization_lineage | https://openalex.org/I205783295 |
| best_oa_location.license | |
| best_oa_location.pdf_url | https://arxiv.org/pdf/2410.00801 |
| best_oa_location.version | submittedVersion |
| best_oa_location.raw_type | text |
| best_oa_location.license_id | |
| best_oa_location.is_accepted | False |
| best_oa_location.is_published | False |
| best_oa_location.raw_source_name | |
| best_oa_location.landing_page_url | http://arxiv.org/abs/2410.00801 |
| primary_location.id | pmh:oai:arXiv.org:2410.00801 |
| primary_location.is_oa | True |
| primary_location.source.id | https://openalex.org/S4306400194 |
| primary_location.source.issn | |
| primary_location.source.type | repository |
| primary_location.source.is_oa | True |
| primary_location.source.issn_l | |
| primary_location.source.is_core | False |
| primary_location.source.is_in_doaj | False |
| primary_location.source.display_name | arXiv (Cornell University) |
| primary_location.source.host_organization | https://openalex.org/I205783295 |
| primary_location.source.host_organization_name | Cornell University |
| primary_location.source.host_organization_lineage | https://openalex.org/I205783295 |
| primary_location.license | |
| primary_location.pdf_url | https://arxiv.org/pdf/2410.00801 |
| primary_location.version | submittedVersion |
| primary_location.raw_type | text |
| primary_location.license_id | |
| primary_location.is_accepted | False |
| primary_location.is_published | False |
| primary_location.raw_source_name | |
| primary_location.landing_page_url | http://arxiv.org/abs/2410.00801 |
| publication_date | 2024-10-01 |
| publication_year | 2024 |
| referenced_works_count | 0 |
| abstract_inverted_index.a | 23, 39, 54, 95, 133, 142 |
| abstract_inverted_index.In | 94 |
| abstract_inverted_index.We | 52 |
| abstract_inverted_index.as | 88, 90, 132 |
| abstract_inverted_index.in | 13, 119 |
| abstract_inverted_index.is | 21 |
| abstract_inverted_index.of | 10, 42, 63, 112, 121 |
| abstract_inverted_index.on | 47, 66, 73, 141, 147 |
| abstract_inverted_index.to | 6, 37 |
| abstract_inverted_index.we | 101 |
| abstract_inverted_index.AMD | 48, 74, 148 |
| abstract_inverted_index.GPU | 1 |
| abstract_inverted_index.Our | 126 |
| abstract_inverted_index.and | 15, 29, 50, 56, 79, 82, 110, 124, 128, 138, 144 |
| abstract_inverted_index.are | 3 |
| abstract_inverted_index.for | 59, 135 |
| abstract_inverted_index.gap | 24 |
| abstract_inverted_index.the | 8, 26, 30, 43, 61, 91, 113 |
| abstract_inverted_index.CPU. | 93 |
| abstract_inverted_index.GPUs | 49, 109 |
| abstract_inverted_index.RCCL | 114 |
| abstract_inverted_index.This | 34 |
| abstract_inverted_index.aims | 36 |
| abstract_inverted_index.base | 134 |
| abstract_inverted_index.data | 64 |
| abstract_inverted_index.four | 99 |
| abstract_inverted_index.host | 92 |
| abstract_inverted_index.meet | 7 |
| abstract_inverted_index.show | 102 |
| abstract_inverted_index.test | 55, 127 |
| abstract_inverted_index.that | 103 |
| abstract_inverted_index.well | 89 |
| abstract_inverted_index.with | 98 |
| abstract_inverted_index.work | 35 |
| abstract_inverted_index.CPUs. | 51 |
| abstract_inverted_index.GPUs, | 76, 87, 100 |
| abstract_inverted_index.needs | 9 |
| abstract_inverted_index.setup | 97 |
| abstract_inverted_index.terms | 120 |
| abstract_inverted_index.there | 20 |
| abstract_inverted_index.Fabric | 45 |
| abstract_inverted_index.MI250X | 75 |
| abstract_inverted_index.Modern | 0 |
| abstract_inverted_index.better | 40 |
| abstract_inverted_index.direct | 104 |
| abstract_inverted_index.memory | 83, 106, 137 |
| abstract_inverted_index.method | 130 |
| abstract_inverted_index.serves | 131 |
| abstract_inverted_index.system | 143 |
| abstract_inverted_index.between | 25, 86, 108 |
| abstract_inverted_index.latency | 123 |
| abstract_inverted_index.library | 115 |
| abstract_inverted_index.machine | 16 |
| abstract_inverted_index.options | 72 |
| abstract_inverted_index.propose | 53 |
| abstract_inverted_index.provide | 38 |
| abstract_inverted_index.systems | 2 |
| abstract_inverted_index.However, | 19 |
| abstract_inverted_index.Infinity | 44 |
| abstract_inverted_index.accesses | 107 |
| abstract_inverted_index.capacity | 28 |
| abstract_inverted_index.domains. | 18 |
| abstract_inverted_index.evolving | 5 |
| abstract_inverted_index.hardware | 27 |
| abstract_inverted_index.learning | 17 |
| abstract_inverted_index.systems, | 68 |
| abstract_inverted_index.systems. | 151 |
| abstract_inverted_index.MPI-based | 117 |
| abstract_inverted_index.computing | 150 |
| abstract_inverted_index.different | 70 |
| abstract_inverted_index.improving | 145 |
| abstract_inverted_index.including | 77 |
| abstract_inverted_index.movements | 65 |
| abstract_inverted_index.multi-GPU | 67, 149 |
| abstract_inverted_index.solutions | 118 |
| abstract_inverted_index.stressing | 69 |
| abstract_inverted_index.typically | 22 |
| abstract_inverted_index.achievable | 31 |
| abstract_inverted_index.allocation | 84 |
| abstract_inverted_index.bandwidth. | 125 |
| abstract_inverted_index.collective | 80 |
| abstract_inverted_index.constantly | 4 |
| abstract_inverted_index.evaluation | 57, 129 |
| abstract_inverted_index.outperform | 116 |
| abstract_inverted_index.scientific | 14 |
| abstract_inverted_index.strategies | 85, 140 |
| abstract_inverted_index.validating | 136 |
| abstract_inverted_index.application | 32 |
| abstract_inverted_index.methodology | 58 |
| abstract_inverted_index.performance | 62 |
| abstract_inverted_index.single-node | 96 |
| abstract_inverted_index.utilization | 111 |
| abstract_inverted_index.applications | 12, 146 |
| abstract_inverted_index.peer-to-peer | 105 |
| abstract_inverted_index.performance. | 33 |
| abstract_inverted_index.communication | 71, 139 |
| abstract_inverted_index.interconnects | 46 |
| abstract_inverted_index.understanding | 41 |
| abstract_inverted_index.characterizing | 60 |
| abstract_inverted_index.communication, | 81 |
| abstract_inverted_index.point-to-point | 78 |
| abstract_inverted_index.computing-intensive | 11 |
| abstract_inverted_index.memory/communication | 122 |
| cited_by_percentile_year | |
| countries_distinct_count | 0 |
| institutions_distinct_count | 6 |
| citation_normalized_percentile |