Systems & Networks · Computer Science & AI
Performance
The newest papers in Performance that ReadSearch has read, newest first. Each links to its own page, with its authors, date, where it was published and the original.
158 papers, from 4 September 2026 to 5 November 2026: 158 arXiv preprints.
Get the few papers worth your time in Performance every morning.
Follow Performance and ReadSearch reads its new papers each day, then brings you the few worth reading, each with its key finding in a sentence.
Newest papers, page 5
CXDVirt: Low Latency Kernel Module Based CXL-SSD Emulation
CXL-SSDs promise memory-semantic access to NAND-scale capacity through a small in-device DRAM cache, but scarce hardware makes emulation central to CXL-SSD research.Request Order Matters: Cache-History Sensitivity in Selective KV-Cache Reuse for Rolling Agents
Long-running agents repeatedly call an LLM while retaining most of their document window, evicting old documents, and appending new ones.Fine-Tuning a 3B-Parameter LLM on a Smartphone: Characterizing Sustained Training
Multi-billion-parameter LLMs now run on phones for inference, and training them on the device would personalize them without user data leaving the phone.Evaluating Inference Compute for Generative AI: A Framework for Enterprise Workloads
LLM deployment is shifting from single-turn completion to agentic trajectories in which a model plans, calls tools, reads results and reasons at test time before acting.Characterizing Parallelism Strategies in LLM Inference: Fundamental Compute-Communication Trade-offs
Large Language Model (LLM) inference has become the dominant workload in modern AI systems, requiring serving infrastructures to maximize throughput while meeting strict latency Service-Level…PhaseGate: Phase-Aware CPU Retrieval Scheduling for On-Device LLMs on Unified Memory
On-device assistants run GPU-based LLM inference alongside CPU retrieval on unified-memory systems.AID: A Framework for AI Infrastructure Dynamics
A useful model of AI inference infrastructure must specify the system state, the information available to an observer, and the decisions the model is intended to support.Network Speed Scaling with Competitive Ratios Independent of the Network
In network speed scaling, jobs arrive over time at a network of servers whose speeds can be tuned, every job must be processed by the servers along one of its allowed routes, and the goal is to…TreeWalker: Partial Evaluation for Grouped Tree-Ensemble Inference
Many inference workloads evaluate a trained tree ensemble on row groups that share feature values: discrete-time survival models expand each patient into $G$ time steps, click-through-rate models…How Do Coding Agents Optimize Software and Report Performance Validation? A Large-Scale Empirical Study of Open-Source Pull Requests
Software performance optimization requires identifying bottlenecks and improvement opportunities, and empirically evaluating the effects and trade-offs across performance objectives.Catscan: Visualizing Pipelines of CPU Performance Simulation
Processor pipeline visualization tools are routine inside industry CPU teams, but few of them are described or released publicly.Activation Sparsity with Weight Approximation for Faster LLM Decoding on Offloaded Weights
Deploying LLMs on consumer-grade GPUs with insufficient memory to hold their weights can result in prohibitively slow inference, because decoding repeatedly transfers offloaded weights from system…FFASR: Benchmarking Far-Field Automatic Speech Recognition using High-Fidelity Simulated RIRs
Far-field automatic speech recognition(ASR) degrades under reverberation, noise, and talker motion, yet the benchmarks that drive model selection emphasize close-microphone speech.Working Around the Compute Ceiling: Byte-Exact Memory in Galahad Makes LLM Reading a One-Time Cost LLM Reading a One-Time Cost
A transformer language model performs a bounded amount of computation per token, and recent work by Vishal Sikka, former CEO of Infosys, argues that this bound limits which tasks a model can carry…AgBench: Agentic AI Benchmarks for Personal AI Devices
Agentic AI systems increasingly rely on cloud-hosted large language models for planning, tool use, and iterative execution, raising concerns about API cost and data exposure.DPS: Dual-Mode Precision LLM Serving with Semi-Unified Memory
Existing LLM serving systems virtualize and optimize KV-cache memory, but treat model-weight memory as fixed throughout execution.Accelerator Choice Is Not Enough: AlphaFold2 Inference on Cloud TPUs
AlphaFold2 is written in JAX, so the same inference code compiles and runs unchanged on CPUs, GPUs and Google Cloud TPUs.Beneath the Tokens: A Performance Engineering Study of Multi-Token Prediction in GPU-Accelerated LLM Inference
Autoregressive large language model inference repeatedly invokes the target model to generate one token at a time, making generation sensitive to GPU memory movement and sequential execution.Beyond Energy: When Sustainability Dimensions Reshape LLM Serving Decisions
Large language model (LLM) serving has environmental impacts across energy consumption, carbon emission, water consumption, and biodiversity loss.Hardware-Aware Features for CUTLASS Kernel Selection
GPU libraries such as CUTLASS expose tens of thousands of semantically equivalent kernels for a single operation, making exhaustive autotuning expensive and execution-free selection difficult.