Searcharxiv⌕ Search

arXiv subjects

Victor Lee

Publications and source records attributed to Victor Lee.

13 recordsLinked to original sources
arXiv2026

Themis: Software-Defined Hardware Prefetching

Data cache misses represent a significant portion of stall cycles in datacenter workloads. Hardware prefetchers that reduce such stalls by fetching data ahead of time have become increasingly sophisticated. However, to achieve high coverage, they have to prefetch aggressively, generating many inaccurate accesses that waste memory bandwidth. This is problematic in datacenter environments where memory bandwidth is a limited resource due to high multi-tenancy. We observe that for datacenter workloads, inaccurate prefetches can be effectively filtered on a data page granularity, without sacrificing prefetch coverage. However, storing per-page metadata about prefetch usefulness in hardware is costly, so we propose a novel hardware-software interface for data prefetching: The software directs the hardware on where to prefetch, and the hardware identifies and issues prefetches in the regions of interest. We propose Themis, a profile-guided hardware prefetching solution that implements this new interface. Themis utilizes page-level hints stored in page-table entries to disable the prefetcher for certain data pages at runtime. Themis requires no binary or ISA changes and can be used to optimize processes without disrupting their execution. Themis is also orthogonal to existing works on prefetching and can be applied to optimize any hardware prefetcher. Our results show that Themis is able to achieve around 40% reduction in useless prefetch requests, resulting in speedup for all the evaluated prefetchers for datacenter workloads, including 4.1% for BOP, 3.1% for SPP+PPF, and 1.4% for Pythia.

cs.AR↗
arXiv2025

Concorde: Fast and Accurate CPU Performance Modeling with Compositional Analytical-ML Fusion

Cycle-level simulators such as gem5 are widely used in microarchitecture design, but they are prohibitively slow for large-scale design space explorations. We present Concorde, a new methodology for learning fast and accurate performance models of microarchitectures. Unlike existing simulators and learning approaches that emulate each instruction, Concorde predicts the behavior of a program based on compact performance distributions that capture the impact of different microarchitectural components. It derives these performance distributions using simple analytical models that estimate bounds on performance induced by each microarchitectural component, providing a simple yet rich representation of a program's performance characteristics across a large space of microarchitectural parameters. Experiments show that Concorde is more than five orders of magnitude faster than a reference cycle-level simulator, with about 2% average Cycles-Per-Instruction (CPI) prediction error across a range of SPEC, open-source, and proprietary benchmarks. This enables rapid design-space exploration and performance sensitivity analyses that are currently infeasible, e.g., in about an hour, we conducted a first-of-its-kind fine-grained performance attribution to different microarchitectural components across a diverse set of programs, requiring nearly 150 million CPI evaluations.

cs.AR↗
arXiv2022

Alternately Optimized Graph Neural Networks

Graph Neural Networks (GNNs) have greatly advanced the semi-supervised node classification task on graphs. The majority of existing GNNs are trained in an end-to-end manner that can be viewed as tackling a bi-level optimization problem. This process is often inefficient in computation and memory usage. In this work, we propose a new optimization framework for semi-supervised learning on graphs. The proposed framework can be conveniently solved by the alternating optimization algorithms, resulting in significantly improved efficiency. Extensive experiments demonstrate that the proposed method can achieve comparable or better performance with state-of-the-art baselines while it has significantly better computation and memory efficiency.

cs.LG↗
arXiv2020

Efficient Probabilistic Inference in the Quest for Physics Beyond the Standard Model

We present a novel probabilistic programming framework that couples directly to existing large-scale simulators through a cross-platform probabilistic execution protocol, which allows general-purpose inference engines to record and control random number draws within simulators in a language-agnostic way. The execution of existing simulators as probabilistic programs enables highly interpretable posterior inference in the structured model defined by the simulator code base. We demonstrate the technique in particle physics, on a scientifically accurate simulation of the tau lepton decay, which is a key ingredient in establishing the properties of the Higgs boson. Inference efficiency is achieved via inference compilation where a deep recurrent neural network is trained to parameterize proposal distributions and control the stochastic simulator in a sequential importance sampling scheme, at a fraction of the computational cost of a Markov chain Monte Carlo baseline.

cs.LG↗
arXiv2019

DisCo: Physics-Based Unsupervised Discovery of Coherent Structures in Spatiotemporal Systems

Extracting actionable insight from complex unlabeled scientific data is an open challenge and key to unlocking data-driven discovery in science. Complementary and alternative to supervised machine learning approaches, unsupervised physics-based methods based on behavior-driven theories hold great promise. Due to computational limitations, practical application on real-world domain science problems has lagged far behind theoretical development. We present our first step towards bridging this divide - DisCo - a high-performance distributed workflow for the behavior-driven local causal state theory. DisCo provides a scalable unsupervised physics-based representation learning method that decomposes spatiotemporal systems into their structurally relevant components, which are captured by the latent local causal state variables. Complex spatiotemporal systems are generally highly structured and organize around a lower-dimensional skeleton of coherent structures, and in several firsts we demonstrate the efficacy of DisCo in capturing such structures from observational and simulated scientific data. To the best of our knowledge, DisCo is also the first application software developed entirely in Python to scale to over 1000 machine nodes, providing good performance along with ensuring domain scientists' productivity. We developed scalable, performant methods optimized for Intel many-core processors that will be upstreamed to open-source Python library packages. Our capstone experiment, using newly developed DisCo workflow and libraries, performs unsupervised spacetime segmentation analysis of CAM5.1 climate simulation data, processing an unprecedented 89.5 TB in 6.6 minutes end-to-end using 1024 Intel Haswell nodes on the Cori supercomputer obtaining 91% weak-scaling and 64% strong-scaling efficiency.

physics.comp-ph↗
arXiv2019

Towards Unsupervised Segmentation of Extreme Weather Events

Extreme weather is one of the main mechanisms through which climate change will directly impact human society. Coping with such change as a global community requires markedly improved understanding of how global warming drives extreme weather events. While alternative climate scenarios can be simulated using sophisticated models, identifying extreme weather events in these simulations requires automation due to the vast amounts of complex high-dimensional data produced. Atmospheric dynamics, and hydrodynamic flows more generally, are highly structured and largely organize around a lower dimensional skeleton of coherent structures. Indeed, extreme weather events are a special case of more general hydrodynamic coherent structures. We present a scalable physics-based representation learning method that decomposes spatiotemporal systems into their structurally relevant components, which are captured by latent variables known as local causal states. For complex fluid flows we show our method is capable of capturing known coherent structures, and with promising segmentation results on CAM5.1 water vapor data we outline the path to extreme weather identification from unlabeled climate model simulation data.

physics.comp-ph↗
arXiv2019

Etalumis: Bringing Probabilistic Programming to Scientific Simulators at Scale

Probabilistic programming languages (PPLs) are receiving widespread attention for performing Bayesian inference in complex generative models. However, applications to science remain limited because of the impracticability of rewriting complex scientific simulators in a PPL, the computational cost of inference, and the lack of scalable implementations. To address these, we present a novel PPL framework that couples directly to existing scientific simulators through a cross-platform probabilistic execution protocol and provides Markov chain Monte Carlo (MCMC) and deep-learning-based inference compilation (IC) engines for tractable inference. To guide IC inference, we perform distributed training of a dynamic 3DCNN--LSTM architecture with a PyTorch-MPI-based framework on 1,024 32-core CPU nodes of the Cori supercomputer with a global minibatch size of 128k: achieving a performance of 450 Tflop/s through enhancements to PyTorch. We demonstrate a Large Hadron Collider (LHC) use-case with the C++ Sherpa simulator and achieve the largest-scale posterior inference in a Turing-complete PPL.

cs.LG↗
arXiv2019

TigerGraph: A Native MPP Graph Database

We present TigerGraph, a graph database system built from the ground up to support massively parallel computation of queries and analytics. TigerGraph's high-level query language, GSQL, is designed for compatibility with SQL, while simultaneously allowing NoSQL programmers to continue thinking in Bulk-Synchronous Processing (BSP) terms and reap the benefits of high-level specification. GSQL is sufficiently high-level to allow declarative SQL-style programming, yet sufficiently expressive to concisely specify the sophisticated iterative algorithms required by modern graph analytics and traditionally coded in general-purpose programming languages like C++ and Java. We report very strong scale-up and scale-out performance over a benchmark we published on GitHub for full reproducibility.

cs.DB↗
arXiv2018

CosmoFlow: Using Deep Learning to Learn the Universe at Scale

Deep learning is a promising tool to determine the physical model that describes our universe. To handle the considerable computational cost of this problem, we present CosmoFlow: a highly scalable deep learning application built on top of the TensorFlow framework. CosmoFlow uses efficient implementations of 3D convolution and pooling primitives, together with improvements in threading for many element-wise operations, to improve training performance on Intel(C) Xeon Phi(TM) processors. We also utilize the Cray PE Machine Learning Plugin for efficient scaling to multiple nodes. We demonstrate fully synchronous data-parallel training on 8192 nodes of Cori with 77% parallel efficiency, achieving 3.5 Pflop/s sustained performance. To our knowledge, this is the first large-scale science application of the TensorFlow framework at supercomputer scale with fully-synchronous training. These enhancements enable us to process large 3D dark matter distribution and predict the cosmological parameters $Ω_M$, $σ_8$ and n$_s$ with unprecedented accuracy.

astro-ph.CO↗
arXiv2018

Evolutionary strategy for inverse charge measurements of dielectric particles

We report a computational strategy to obtain the charges of individual dielectric particles from experimental observation of their interactions as a function of time. This strategy uses evolutionary optimization to minimize the difference between trajectories extracted from experiment and simulated trajectories based on many-particle force fields. The force fields include both Coulombic interactions and dielectric polarization effects that arise due to particle-particle charge mismatch and particle-environment dielectric contrast. The strategy was applied to systems of free falling charged granular particles in vacuum, where electrostatic interactions are the only driving forces that influence the particles' motion. We show that when the particles' initial positions and velocities are known, the optimizer requires only an initial and final particle configuration of a short trajectory in order to accurately infer the particles' charges; when the initial velocities are unknown and only the initial positions are given, the optimizer can learn from multiple frames along the trajectory to determine the particles' initial velocities and charges. While the results presented here offer a proof-of-concept demonstration of the proposed ideas, the proposed strategy could be extended to more complex systems of electrostatically charged granular matter.

cond-mat.soft↗
arXiv2018

Collisional Charging of Individual Sub-Millimeter Particles: Using Ultrasonic Levitation to Initiate and Track Charge Transfer

Electrostatic charging of insulating fine particles can be responsible for numerous phenomena ranging from lightning in volcanic plumes to dust explosions. However, even basic aspects of how fine particles become charged are still unclear. Studying particle charging is challenging because it usually involves the complexities associated with many particle collisions. To address these issues we introduce a method based on acoustic levitation, which makes it possible to initiate sequences of repeated collisions of a single sub-millimeter particle with a flat plate, and to precisely measure the particle charge in-situ after each collision. We show that collisional charge transfer between insulators is dependent on the hydrophobicity of the contacting surfaces. We use glass, which we modify by attaching nonpolar molecules to the particle, the plate, or both. We find that hydrophilic surfaces develop significant positive charges after contacting hydrophobic surfaces. Moreover, we demonstrate that charging between a hydrophilic and a hydrophobic surface is suppressed in an acidic environment and enhanced in a basic one. Application of an electric field during each collision is found to modify the charge transfer, again depending on surface hydrophobicity. We discuss these results within the context of contact charging due to ion transfer and show that they lend strong support to OH- ions as the charge carriers.

cond-mat.mtrl-sci↗
arXiv2014

Size-dependent same-material tribocharging in insulating grains

Observations of flowing granular matter have suggested that same-material tribocharging de- pends on particle size, rendering large grains positive and small ones negative. Models assuming the transfer of trapped electrons can explain this, but so far have not been validated. Tracking individual grains in an electric field, we show quantitatively that charge is transferred based on size between materially identical grains. However, the surface density of trapped electrons, measured independently by thermoluminescence techniques, is orders of magnitude too small to account for the scale of charge transferred. This suggests that another negatively charged species, such as ions, is responsible.

cond-mat.soft↗