Search for: All records

Fixed-point iteration over parallel relational algebra (RA) operators underpins a broad class of symbolic AI and recursive analytic workloads, including graph reachability, points-to analysis, and rule-based inference. At scale, each iteration materializes substantial intermediate and output relations across distributed memory, causing persistence I/O to become a dominant contributor to end-to-end cost. This I/O is inherently write-dominated: every iteration emits a delta of newly derived tuples, while the runtime must periodically checkpoint accumulated relation state to stable storage for fault tolerance and out-of-core execution. This paper presents a case study in optimizing checkpoints for parallel iterated relational algebra, comparing an existing relation-dumping pipeline against a structured, performance-tuned alternative. The baseline stores relations as opaque binary dumps via POSIX positioned writes or collective MPIIO with minimal file-system hinting.We extend this pipeline with a parallel HDF5 backend that stores relations as schema-aware datasets and writes rank partitions using HDF5 hyperslabs with collective transfers. To avoid the performance cliffs associated with default parallel HDF5 settings, we apply a systematic size-aware tuning policy that configures both HDF5 property lists and MPI-IO hints. The optimized pathway uses a two-phase workflow that creates the file and dataset first and then performs the parallel write. It also applies output-regime-dependent alignment and buffering, configures metadata caching and sieve buffering, selects an appropriate dataset layout, and enforces collective dataset transfers. Across final write sizes from 15 GB to 30 GB, the tuned HDF5 path consistently outperforms the MPI-IO baselines. At higher MPI process counts, it is often 35%-58% faster while producing self-describing outputs. 

Visualizing the large-scale datasets output by HPC resources presents a difficult challenge, as the memory and compute power required become prohibitively expensive for end user systems. Novel view synthesis techniques can address this by producing a small, interactive model of the data, requiring only a set of training images to learn from. While these models allow accessible visualization of large data and complex scenes, they do not provide the interactions needed for scientific volumes, as they do not support interactive selection of transfer functions and lighting parameters. To address this, we introduce Volume Encoding Gaussians (VEG), a 3D Gaussian-based representation for volume visualization that supports arbitrary color and opacity mappings. Unlike prior 3D Gaussian Splatting (3DGS) methods that store color and opacity for each Gaussian, VEG decouple the visual appearance from the data representation by encoding only scalar values, enabling transfer function-agnostic rendering of 3DGS models. To ensure complete scalar field coverage, we introduce an opacity-guided training strategy, using differentiable rendering with multiple transfer functions to optimize our data representation. This allows VEG to preserve fine features across a dataset’s full scalar range while remaining independent of any specific transfer function. Across a diverse set of volume datasets, we demonstrate that our method outperforms the state-of-the-art on transfer functions unseen during training, while requiring a fraction of the memory and training time. 

Abstract PurposeTo examine the effect of incorporating self‐supervised denoising as a pre‐processing step for training deep learning (DL) based reconstruction methods on data corrupted by Gaussian noise. K‐space data employed for training are typically multi‐coil and inherently noisy. Although DL‐based reconstruction methods trained on fully sampled data can enable high reconstruction quality, obtaining large, noise‐free datasets is impractical. MethodsWe leverage Generalized Stein's Unbiased Risk Estimate (GSURE) for denoising. We evaluate two DL‐based reconstruction methods: Diffusion Probabilistic Models (DPMs) and Model‐Based Deep Learning (MoDL). We evaluate the impact of denoising on the performance of these DL‐based methods in solving accelerated multi‐coil magnetic resonance imaging (MRI) reconstruction. The experiments were carried out on T2‐weighted brain and fat‐suppressed proton‐density knee scans. ResultsWe observed that self‐supervised denoising enhances the quality and efficiency of MRI reconstructions across various scenarios. Specifically, employing denoised images rather than noisy counterparts when training DL networks results in lower normalized root mean squared error (NRMSE), higher structural similarity index measure (SSIM) and peak signal‐to‐noise ratio (PSNR) across different SNR levels, including 32, 22, and 12 dB for T2‐weighted brain data, and 24, 14, and 4 dB for fat‐suppressed knee data. ConclusionWe showed that denoising is an essential pre‐processing technique capable of improving the efficacy of DL‐based MRI reconstruction methods under diverse conditions. By refining the quality of input data, denoising enables training more effective DL networks, potentially bypassing the need for noise‐free reference MRI scans. 

Modern Datalog engines (e.g., LogicBlox, Soufflé, ddlog) enable their users to write declarative queries which com- pute recursive deductions over extensional facts, leaving high-performance operationalization (query planning, semi- naïve evaluation, and parallelization) to the engine. Such engines form the backbone of modern high-throughput ap- plications in static analysis, network monitoring, and social- media mining. In this paper, we present a methodology for implementing a modern in-memory Datalog engine on data center GPUs, allowing us to achieve significant (up to 45×) gains compared to Soufflé (a modern CPU-based en- gine) on context-sensitive points-to analysis of PostgreSQL. We present GPUlog, a Datalog engine backend that imple- ments iterated relational algebra kernels over a novel range- indexed data structure we call the hash-indexed sorted ar- ray (HISA). HISA combines the algorithmic benefits of in- cremental range-indexed relations with the raw computa- tion throughput of operations over dense data structures. Our experiments show that GPUlog is significantly faster than CPU-based Datalog engines while achieving a favorable memory footprint compared to contemporary GPU-based joins.