Workshop: Thirteenth Workshop on Accelerator Programming and Directives (WACCPD 2026)
Authors: Alexander Strack (University of Stuttgart, Louisiana State University) and Marcel Graf, Alexander Van Craen, and Dirk Pflüger (University of Stuttgart)
Abstract: Modern multi-vendor GPU environments force a choice between portable code and native performance. GPRat is an open-source Python library for exact Gaussian process regression built on the HPX asynchronous many-task runtime, tiled algorithms, and CPU/CUDA BLAS backends. We contribute a UXL Foundation oneMath backend with a hybrid HPX/SYCL design, extending GPRat to AMD and Intel GPUs. The new SYCL backend comes within \qty{3}{\percent} of native CUDA on an NVIDIA A30 at the largest problem size and outperforms the GPflow and GPyTorch libraries by up to $1.75\times$ and $2.17\times$ on the A30 and up to $1.79\times$ and $4.62\times$ on an AMD Instinct MI210. On an Intel Arc B580, a consumer-class card with FP64 throughput below our CPU's, the unmodified SYCL backend trails GPRat's CPU backend. Overall, oneMath's dispatch to vendor-native BLAS lets a single SYCL codebase match CUDA on NVIDIA and outperform established Python GP frameworks on AMD, without vendor-specific code.
Back to Thirteenth Workshop on Accelerator Programming and Directives (WACCPD 2026) Archive Listing
Back to Full Workshop Archive Listing