SoaAlloc: Accelerating Single-Method Multiple-Objects Applications on GPUs

09/20/2018
by   Matthias Springer, et al.
0

We propose SoaAlloc, a dynamic object allocator for Single-Method Multiple-Objects applications in CUDA. SoaAlloc is the first allocator for GPUs that (a) arranges allocations in a SIMD-friendly Structure of Arrays (SOA) data layout, (b) provides a do-all operation for maximizing the benefit of SOA, and (c) is on par with state-of-the-art memory allocators for raw (de)allocation time. Our benchmarks show that the SOA layout leads to significantly better memory bandwidth utilization, resulting in a 2x speedup of application code.

READ FULL TEXT

page 1

page 2

page 3

research
10/28/2018

SoaAlloc: A Lock-free Hierarchical Bitmap-based Object Allocator for GPUs

Designing dynamic memory allocators for GPUs is challenging because appl...
research
01/11/2019

Automatic acceleration of Numpy applications on GPUs and multicore CPUs

Frameworks like Numpy are a popular choice for application developers fr...
research
10/28/2018

DynaSOAr: A Parallel Memory Allocator for Object-oriented Programming on GPUs with Efficient Memory Access

Object-oriented programming has long been regarded as too inefficient fo...
research
04/17/2021

Ripple : Simplified Large-Scale Computation on Heterogeneous Architectures with Polymorphic Data Layout

GPUs are now used for a wide range of problems within HPC. However, maki...
research
02/11/2022

Increasing FPGA Accelerators Memory Bandwidth with a Burst-Friendly Memory Layout

Offloading compute-intensive kernels to hardware accelerators relies on ...
research
06/15/2019

Accelerating Concurrent Heap on GPUs

Priority queue, often implemented as a heap, is an abstract data type th...
research
01/12/2023

Neural Shadow Mapping

We present a neural extension of basic shadow mapping for fast, high qua...

Please sign up or login with your details

Forgot password? Click here to reset