性能

Linear layers hold most of a transformer's parameters. We replace each linear layer with one that stores $K$ out of $mn$ two-dimensional DCT coefficients per weight matrix and reconstructs the full matrix through an inverse DCT at every…

On-device running Large Language Models (LLMs) is nowadays a critical enabler towards preserving user privacy. We observe that the attention operator falls back from the special-purpose NPU to the general-purpose CPU/GPU because of…

We present memshare\footnote{The Software package is published as a CRAN package under https://CRAN.R-project.org/package=memshare, a package that enables shared memory multicore computation in R by allocating buffers in C++ shared memory…

An $N$-point FFT admits many valid implementations that differ in radix choice, stage ordering, and register-blocking strategy. These alternatives use different SIMD instruction mixes with different latencies, yet produce the same…

One of the most popular and basic principles in programming is the DRY principle (don't repeat yourself). According to it, code duplication should be avoided within a single application. Instead of duplicating it, the code can be exported…

We present the first kernel-fused SAR Range Doppler pipeline on any GPU platform. By fusing FFT, matched-filter multiply, and IFFT into a single Metal compute dispatch -- keeping all intermediate data in 32\,KiB on-chip memory -- we process…

Modern blockchain ecosystems comprise many heterogeneous networks, creating a growing need for interoperability. Cross-chain bridges provide the core infrastructure for this interoperability by enabling verifiable state transitions that…

The exponential growth in data has intensified the demand for computational power to train large-scale deep learning models. However, the rapid growth in model size and complexity raises concerns about equal and fair access to computational…

The fused multiply-add (FMA) instruction enables the radix-2 FFT butterfly to be computed in 6~FMA operations -- the proven minimum. The classical factorization by Linzer and Feig~\cite{linzer1993} precomputes the ratio $\cot\theta =…

Performance diagnosis in production-scale AI training is challenging because subtle OS-level issues can trigger cascading GPU delays and network slowdowns, degrading training efficiency across thousands of GPUs. Existing profiling tools are…

Low-altitude wireless networks (LAWN) require drones to follow specific trajectories controlled by ground base stations (GBSs). However, given complex low-altitude channel conditions and limited spectrum and power resources, sensing errors…

Scaling laws relate model quality to compute budget (FLOPs), but practitioners face wall-clock time constraints, not compute budgets. We study optimal model sizing under fixed time budgets from 5 minutes to 24 hours on consumer GPUs (RTX…

We study the multiserver-job setting in the load-focused multilevel scaling limit, where system load approaches capacity much faster than the growth of the number of servers $n$. We consider the ``1 and $n$'' system, where each job requires…

The Embedded Trace Macrocell (ETM) is a standard component of Arm's CoreSight architecture, present in a wide range of platforms and primarily designed for tracing and debugging. In this work, we demonstrate that it can be repurposed to…

Ranking methods or models based on their performance is of prime importance but is tricky because performance is fundamentally multidimensional. In the case of classification, precision and recall are scores with probabilistic…

As AI accelerators gain prominence, their potential for traditional scientific computing workloads remains unclear. This paper explores Tenstorrent's Wormhole architecture, a spatial computing platform designed for neural network…

Frequently, multiple entities (methods, algorithms, procedures, solutions, etc.) can be developed for a common task and applied across various domains that differ in the distribution of scenarios encountered. For example, in computer…

High-speed packet processing on multicore CPUs places extreme demands on memory allocators. In systems like DPDK, fixed-size memory pools back packet buffers (mbufs) to avoid costly dynamic allocation. However, even DPDK's optimized mempool…

The key-value (KV) cache has become the dominant contributor to memory consumption in large language model (LLM) inference. Although offloading KVCache from GPU high-bandwidth memory (HBM) to CPU DRAM alleviates device memory pressure, DRAM…

Incident management is essential to maintain the reliability and availability of cloud computing services. Cloud vendors typically disclose incident reports to the public, summarizing the failures and recovery process to help minimize their…