Academic Journal
Προβολή σε EDS

Memory hierarchy in AI systems: dual contexts, estimation modelling, and optimisation strategies.

Λεπτομέρειες βιβλιογραφικής εγγραφής
Τίτλος: Memory hierarchy in AI systems: dual contexts, estimation modelling, and optimisation strategies.
Συγγραφείς: Brahma, Manmohan1 manmohan.brahma@gmail.com, Patra, G. K.1,2
Πηγή: Current Science (00113891). 7/10/2026, Vol. 131 Issue 1, p64-76. 13p.
Θεματικοί όροι: *Computer memory management, *Estimation theory, *Artificial intelligence, *Systems design
Περίληψη: The rapid growth of large-scale artificial intelligence (AI) models has placed memory management at the centre of both hardware and software design challenges. The present study investigates and introduces the Memory Capacity Estimation Model (MCEM): a parametric, device-agnostic, and precision-aware analytical framework for predicting peak memory utilisation across central processing units, graphics processing units, and neural processing units. MCEM formalises a dual-context memory taxonomy that distinguishes training (bandwidth-bound, writedominated) from inference (latency-sensitive, key-value cache-dominated) based on dataflow characteristics, temporal write/read ratios, and reuse distance. Validated against open benchmarks from MLPerf, Hugging Face, and advanced micro devices Instinct MI300X hardware reports, MCEM achieves estimation errors of 0.73%-2.7% across five diverse model-device configurations. The framework is shown to reduce memory overprovisioning by 30%-40% relative to naive sizing, with implications for sustainable, cost-effective AI deployment at cloud, high-performance computing, and edge scales. [ABSTRACT FROM AUTHOR]
Βάση Δεδομένων: Academic Search Index
FullText Text:
  Availability: 0
Header DbId: asx
DbLabel: Academic Search Index
An: 195357105
RelevancyScore: 1431
AccessLevel: 6
PubType: Academic Journal
PubTypeId: academicJournal
PreciseRelevancyScore: 1430.76843261719
IllustrationInfo
Items – Name: Title
  Label: Title
  Group: Ti
  Data: Memory hierarchy in AI systems: dual contexts, estimation modelling, and optimisation strategies.
– Name: Author
  Label: Authors
  Group: Au
  Data: <searchLink fieldCode="AR" term="%22Brahma%2C+Manmohan%22">Brahma, Manmohan</searchLink><relatesTo>1</relatesTo><i> manmohan.brahma@gmail.com</i><br /><searchLink fieldCode="AR" term="%22Patra%2C+G%2E+K%2E%22">Patra, G. K.</searchLink><relatesTo>1,2</relatesTo>
– Name: TitleSource
  Label: Source
  Group: Src
  Data: <searchLink fieldCode="JN" term="%22Current+Science+%2800113891%29%22">Current Science (00113891)</searchLink>. 7/10/2026, Vol. 131 Issue 1, p64-76. 13p.
– Name: Subject
  Label: Subject Terms
  Group: Su
  Data: *<searchLink fieldCode="DE" term="%22Computer+memory+management%22">Computer memory management</searchLink><br />*<searchLink fieldCode="DE" term="%22Estimation+theory%22">Estimation theory</searchLink><br />*<searchLink fieldCode="DE" term="%22Artificial+intelligence%22">Artificial intelligence</searchLink><br />*<searchLink fieldCode="DE" term="%22Systems+design%22">Systems design</searchLink>
– Name: Abstract
  Label: Abstract
  Group: Ab
  Data: The rapid growth of large-scale artificial intelligence (AI) models has placed memory management at the centre of both hardware and software design challenges. The present study investigates and introduces the Memory Capacity Estimation Model (MCEM): a parametric, device-agnostic, and precision-aware analytical framework for predicting peak memory utilisation across central processing units, graphics processing units, and neural processing units. MCEM formalises a dual-context memory taxonomy that distinguishes training (bandwidth-bound, writedominated) from inference (latency-sensitive, key-value cache-dominated) based on dataflow characteristics, temporal write/read ratios, and reuse distance. Validated against open benchmarks from MLPerf, Hugging Face, and advanced micro devices Instinct MI300X hardware reports, MCEM achieves estimation errors of 0.73%-2.7% across five diverse model-device configurations. The framework is shown to reduce memory overprovisioning by 30%-40% relative to naive sizing, with implications for sustainable, cost-effective AI deployment at cloud, high-performance computing, and edge scales. [ABSTRACT FROM AUTHOR]
PLink https://search.ebscohost.com/login.aspx?direct=true&site=eds-live&db=asx&AN=195357105
RecordInfo BibRecord:
  BibEntity:
    Identifiers:
      – Type: doi
        Value: 10.18520/cs/v131/i1/64-76
    Languages:
      – Code: eng
        Text: English
    PhysicalDescription:
      Pagination:
        PageCount: 13
        StartPage: 64
    Subjects:
      – SubjectFull: Computer memory management
        Type: general
      – SubjectFull: Estimation theory
        Type: general
      – SubjectFull: Artificial intelligence
        Type: general
      – SubjectFull: Systems design
        Type: general
    Titles:
      – TitleFull: Memory hierarchy in AI systems: dual contexts, estimation modelling, and optimisation strategies.
        Type: main
  BibRelationships:
    HasContributorRelationships:
      – PersonEntity:
          Name:
            NameFull: Brahma, Manmohan
      – PersonEntity:
          Name:
            NameFull: Patra, G. K.
    IsPartOfRelationships:
      – BibEntity:
          Dates:
            – D: 10
              M: 07
              Text: 7/10/2026
              Type: published
              Y: 2026
          Identifiers:
            – Type: issn-print
              Value: 00113891
          Numbering:
            – Type: volume
              Value: 131
            – Type: issue
              Value: 1
          Titles:
            – TitleFull: Current Science (00113891)
              Type: main
ResultId 1