LLMSpace is the first framework to jointly model operational and embodied carbon for LLM inference on LEO satellites, incorporating radiation-hardened hardware, peripheral systems, and workload patterns such as prefill-decode behavior.
Trends in ai supercomputers,
7 Pith papers cite this work. Polarity classification is still indexing.
7
Pith papers citing it
verdicts
UNVERDICTED 7representative citing papers
A single global merge at the final step of decentralized SGD matches the convergence rate of parallel SGD while improving test accuracy under high data heterogeneity.
StickyInvoc introduces sticky tasks that load LLM model state once and invocation tasks that reuse it, yielding 3.6x speedup on a 150k-inference workflow.
CSA-UD is a communication-semantic-aware unreliable datagram RDMA loss recovery mechanism that improves QP scalability and reduces 99th percentile flow completion times in hyperscale AI training collectives.
Introduces Switching Efficiency (η) decomposed into data, routing efficiency, and port utilization factors to analyze and improve communication bottlenecks in AI data center networks for LLM training.
U.S. operators control 48% of non-U.S. data center projects by investment value, limiting digital sovereignty for host nations and offering the U.S. an additional governance tool for deployed AI infrastructure.
MRLS leaf-spine networks deliver 50% higher throughput than Fat-Tree and 100% higher than Dragonfly for All2All collectives with 100k endpoints via simulation evaluation.
citing papers explorer
-
StickyInvoc: Rethinking Task Models for High-throughput Workflows in the LLM Era
StickyInvoc introduces sticky tasks that load LLM model state once and invocation tasks that reuse it, yielding 3.6x speedup on a 150k-inference workflow.