Performance Papers of the Week

We present Ripple, a CPU-GPU co-scheduler that combines model-derived GPU memory volumes with CPU completion feedback to regulate interference.

Although prior work has extensively studied LLM routing and serving efficiency, the relative importance of fleet composition and routing under resource constraints remains less understood. We investigate this question using a controlled four-GPU serving environment with heterogeneous clinical NLP workloads spanning short factual verification, biomedical semantic matching, and long-context clinical summarization.

This paper presents Token-as-a-Utility, a model-agnostic architecture integrating inference metering, service-level-aware provisioning, and inter-operator settlement.

From these, we obtain the SIMPLE Inference Model to Predict LLMs Energy (SIMPLE), a four-parameter law that separates prefill from decode, and decode into KV-cache cost and next-token computation.

Hardware profilers report how busy each unit was, not where each warp's time went; intra-kernel tracers can show it, but only with hand-placed probes and a recompilation that alters the very kernel being measured. We present Warpline, which gives agents this view automatically while leaving the kernel essentially unchanged.

We introduce HPC-Bench, a performance-grounded benchmark for evaluating LLMs as optimizers of computational kernels in realistic HPC workloads, including 114 workloads across 16 HPC computational motifs.

We present GC-MoE, an open-source FP8/FP4 MoE inference operator for NVIDIA Hopper and Blackwell GPUs that uses CUDA Green Contexts to split the device into a small communication partition and a large compute partition.

This separation risks removing geometric structures that appear redundant under coarse metrics but are critical for precise localization, while leaving hardware-level efficiencies in the retained geometry unexploited. We present FieldRT, an algorithm-hardware co-design methodology that uses a common analysis of geometric sensitivity and data lifetime to derive both the deployment profile and its spatial realization.

We present CrowdForge, an AI- and MR-mediated paradigm where audiences collectively control generative music and virtual environments during a live performance featuring a guitarist and DJ.

We present TAILOR, a framework that guides LLM synthesis using the Conditional Value at Risk (CVaR) of the cost difference between each candidate and an existing baseline on the same workload.

A 6G service crosses a chain of stages from different vendors, and its contract binds class-conditioned guarantees at the terminal interface: in the packet relay chain studied here, a priority class is owed a timely-delivery rate and a loss ceiling. We define trustworthiness as conformance of the delivered per-class distribution of outcomes to the contract and develop the measurement structure this forces.

The window size and budget are found on a per-layer basis using a multi-objective search algorithm.

We propose Feasibility-Preserving Transactional Refinement (FPTR), which always keeps a validated feasible allocation, the incumbent.

Those studies rarely state which blocks the drive chooses to clean. Varying only this choice in simulation, we show it decides how much a label is worth: label value peaks when the drive searches moderately for blocks holding little valid data, falls when it always takes the block with the least, and turns negative when it ignores how much valid data blocks hold.

We propose bypassing policy development and instead asking pretrained models to make resource-management decisions directly, given the current system state, desired outcome, and available choices.

We present KOSMOS, a library for self-scheduling persistent kernels whose workgroups use queues ordered by consumer demand to switch between communication and computation roles as data become ready, with no scheduler, per-call cost model, or tuned split.

This memory grows with the number and size of the experts, so it rises as MoE models scale. We introduce Sonar-MoE, which updates each expert tensor during the backward pass and frees its gradient right away.

In this work, we conduct a large-scale, cross-platform comparative analysis of eight publicly available FaaS workload traces from major cloud providers (Alibaba, Azure, Huawei, IBM) and a scientific computing platform (Globus).

In this paper, we identify that the performance gap stems primarily from frequent data type conversions between quantized and floating-point values, and dynamic and inefficient memory access patterns caused by the growing Key-Value (KV) cache during autoregressive decoding.