Scaling Analysis of LLM Hardware Infrastructure through Uniform Estimates
International Journal of Creative and Open Research in Engineering and Management · 21.08.2026 · 02 (08) · 1-9
DOI: 10.55041/ijcope.v2i8.175 ↗Kurz gefasst
Fachartikel in der Zeitschrift International Journal of Creative and Open Research in Engineering and Management, online erschienen am 21.08.2026. Verfasst von VA Sharma.
Abstract (Originalsprache)
As a methodological illustration we indicate how the same asymptotic and summation techniques supply analytic tools for the derivation of rigorous scaling laws that appear in performance models of GPU clusters, data-centre resource allocation, and the hardware infrastructure supply chains that underpin the training of large language models (LLMs). In this expanded version we develop these links further: the uniform bounds on the Macdonald kernel furnish a priori capacity and latency estimates for large GPU fleets; the Watson-type asymptotic describe the continuous-to-discrete transition that occurs when the number of accelerators tends to infinity; and the Poisson-type identities convert continuous resource integrals into discrete sums over GPU ranks, data-centre racks, or nodes of a global semiconductor supply chain. These analytic devices are shown to be formally analogous to the resilience and digital-twin frameworks that appear in the contemporary literature on infrastructure supply chains. Keywords Kontorovich–Lebedev transform; Watson lemma; asymptotic expansions; index transforms; scaling analysis; GPU clusters; data-centre infrastructure; LLM training systems: infrastructure supply chain; digital twins; supply-chain resilience.
Zitieren
APA
Sharma, V. (2026). Scaling Analysis of LLM Hardware Infrastructure through Uniform Estimates. International Journal of Creative and Open Research in Engineering and Management, 02(08), 1-9. https://doi.org/10.55041/ijcope.v2i8.175
Vancouver
Sharma V. Scaling Analysis of LLM Hardware Infrastructure through Uniform Estimates. International Journal of Creative and Open Research in Engineering and Management. 2026;02(08):1-9. doi:10.55041/ijcope.v2i8.175
BibTeX
@article{Sharma2026,
author = {Sharma, VA},
title = {Scaling Analysis of LLM Hardware Infrastructure through Uniform Estimates},
journal = {International Journal of Creative and Open Research in Engineering and Management},
year = {2026},
volume = {02},
number = {08},
pages = {1--9},
doi = {10.55041/ijcope.v2i8.175}
}
EUVIMED