← 研踪 PaperTrailarXiv:2505.02977 · v4
Tackling Parallelization Challenges of Randomized Preconditioners With Dependency Tracking
首次提交 2025-05-06 03:06
版本更新 2026-11-19 00:59
雷达观测 2026-11-19 15:04
摘要
Incomplete Cholesky (IC) preconditioners are the standard black-box, memory-efficient first resort for large graph Laplacian systems, but classical IC can break down on irregular graphs. Approximate Cholesky (AC) addresses this with randomized dropping that is provably breakdown-free. Its stochastic fill-in pattern, however, resists static parallelization, creating a "setup bottleneck." We present ParAC, which parallelizes the sequential rchol algorithm on CPUs and GPUs using dynamic dependency tracking and architecture-aware memory management, without altering its numerics. Total time-to-solution speedups over deterministic IC reach on CPU and on GPU. Algebraic multigrid (AMG) remains preferable on structured PDE discretizations; on irregular graph Laplacians where AMG struggles, ParAC attains up to and . On 200 adversarial SDD systems, ParAC converges on every instance, whereas the AMG solvers fail on 15 and 19, on CPU and GPU, respectively.
观测到的版本
相邻记录可能跳过中间版本;首次发现 v2 及以上也会标为版本更新。
v4 · 2026-11-19 00:59 · 最新观测版本
Tackling Parallelization Challenges of Randomized Preconditioners With Dependency Tracking
Incomplete Cholesky (IC) preconditioners are the standard black-box, memory-efficient first resort for large graph Laplacian systems, but classical IC can break down on irregular graphs. Approximate Cholesky (AC) addresses this with randomized dropping that is provably breakdown-free. Its stochastic fill-in pattern, however, resists static parallelization, creating a "setup bottleneck." We present ParAC, which parallelizes the sequential rchol algorithm on CPUs and GPUs using dynamic dependency tracking and architecture-aware memory management, without altering its numerics. Total time-to-solution speedups over deterministic IC reach on CPU and on GPU. Algebraic multigrid (AMG) remains preferable on structured PDE discretizations; on irregular graph Laplacians where AMG struggles, ParAC attains up to and . On 200 adversarial SDD systems, ParAC converges on every instance, whereas the AMG solvers fail on 15 and 19, on CPU and GPU, respectively.
查看此版本解读 →查看此版本原文 ↗