TY  - EJOU
AU  - Chu, Xinyue 
AU  - Gao, Jiaquan 
AU  - Sheng, Bo 

TI  - Efficient Concurrent L1-Minimization Solvers on GPUs
T2  - Computer Systems Science and Engineering

PY  - 2021
VL  - 38
IS  - 3
SN  - 

AB  - Given that the concurrent L1-minimization (L1-min) problem is often required in some real applications, we investigate how to solve it in parallel on GPUs in this paper. First, we propose a novel self-adaptive warp implementation of the matrix-vector multiplication (<i>Ax</i>) and a novel self-adaptive thread implementation of the matrix-vector multiplication (<i>A</i><sup><i>T</i></sup><i>x</i>), respectively, on the GPU. The vector-operation and inner-product decision trees are adopted to choose the optimal vector-operation and inner-product kernels for vectors of any size. Second, based on the above proposed kernels, the iterative shrinkage-thresholding algorithm is utilized to present two concurrent L1-min solvers from the perspective of the streams and the thread blocks on a GPU, and optimize their performance by using the new features of GPU such as the shuffle instruction and the read-only data cache. Finally, we design a concurrent L1-min solver on multiple GPUs. The experimental results have validated the high effectiveness and good performance of our proposed methods.
KW  - Concurrent L1-minimization problem; dense matrix-vector multiplication; fast iterative shrinkage-thresholding algorithm; CUDA; GPUs

DO  - 10.32604/csse.2021.017144