FSZ: Breaking the Prediction-Throughput Trade-off in GPU Lossy Compression

2026-07-20T08:52:18Zbfcdad65eb2fac7f7b6d09829b412ac3b1bcb13cfd2001c8f6ba9d457b2b4846
ADAscaleByzantine fault toleranceFSZGPU collectivesGPU lossy compressionHonest Quorum ProblemIPU renderingJoyNexusLLM agent toolingMCP gatewayModel Context Protocol (MCP)NCCLSQUIROaccess controldependency-aware autoscalingisolationlow-latency networkingmicroservice placementmulti-tenant trainingpost-quantum considerationsquantum-classical schedulingserverless autoscalingsession affinity

What happened

A collection of recent systems and ML infrastructure papers from arXiv covering GPU-side algorithmic advances (FSZ lossy compression; low-latency GPU collectives), large-scale agent/tool access and gateway design for LLMs (MCP gateway), multi-tenant training and service architectures for vision-language-action models (JoyNexus), cloud/serverless autoscaling and microservice placement frameworks (dependency-aware autoscaling; ADASCALE), an IPU-based 3D Gaussian renderer, and scheduler designs for quantum-classical and security-constrained workloads (SQUIRO). Several works address multi-tenant/​

Why it matters

A reviewed impact interpretation has not been published for this record.

Evidence and limitations

Source ID
arxiv_cs_dc
Record identifier
bfcdad65eb2fac7f7b6d09829b412ac3b1bcb13cfd2001c8f6ba9d457b2b4846
Enrichment time
2026-07-20T08:52:18Z
AI-assisted enrichment
Yes

This record may overlap with other records. Its enrichment can be incomplete or wrong, and machine assistance was used. Validate consequential decisions against the linked source and your own environment.