Journal: Visual Computing for Industry, Biomedicine and Art

Article Title: Optimizing photoacoustic image reconstruction using cross-platform parallel computation

doi: 10.1186/s42492-018-0002-5

Figure Lengend Snippet: Overall flowchart of a MEXCUDA function

Article Snippet: From this source code, we create the compiled MEXCUDA function by using the mexcuda command in MATLAB.

Techniques:

Journal: Visual Computing for Industry, Biomedicine and Art

Article Title: Optimizing photoacoustic image reconstruction using cross-platform parallel computation

doi: 10.1186/s42492-018-0002-5

Figure Lengend Snippet: Example process flow of the heterogeneous computing. In this figure, we create a sample MEXCUDA function for calculating the radius matrix of a scanned region to each transducer elements. The host code (CPU) is in charge of initialization and finalization blocks such as reading the input from MATLAB, performing condition checks, and allocating memory for input and output data. Device code in GPU is responsible for computing the radius matrix in parallel from required inputs (parallel computation block)

Article Snippet: From this source code, we create the compiled MEXCUDA function by using the mexcuda command in MATLAB.

Techniques: Blocking Assay

Journal: Neuroinformatics

Article Title: Fast Estimation of L1-Regularized Linear Models in the Mass-Univariate Setting

doi: 10.1007/s12021-020-09489-1

Figure Lengend Snippet: Overview over hardware and software requirements of the different functions for estimating L1-regularized linear models. The lasso function is part of the Statistics and Machine Learning Toolbox and only included here for benchmarking purposes (see main text). The lasso_mex, lasso_mexcuda and lasso_gpu functions introduced here are specifically optimized for the mass-univariate analysis approach and depend on different hardware and software configurations

Article Snippet: Moreover, the GPU-accelerated functions require either the CUDA toolkit (lasso_mexcuda) or Matlab’s Parallel Computing Toolbox (lasso_gpu).

Techniques: Software

Journal: Visual Computing for Industry, Biomedicine and Art

Article Title: Optimizing photoacoustic image reconstruction using cross-platform parallel computation

doi: 10.1186/s42492-018-0002-5

Figure Lengend Snippet: Overall flowchart of a MEXCUDA function

Article Snippet: In details, MEXCUDA is the expansion of MATLAB mex function that utilizes C/C++ for execution using C++ MEX API.

Techniques:

Journal: Visual Computing for Industry, Biomedicine and Art

Article Title: Optimizing photoacoustic image reconstruction using cross-platform parallel computation

doi: 10.1186/s42492-018-0002-5

Figure Lengend Snippet: Example process flow of the heterogeneous computing. In this figure, we create a sample MEXCUDA function for calculating the radius matrix of a scanned region to each transducer elements. The host code (CPU) is in charge of initialization and finalization blocks such as reading the input from MATLAB, performing condition checks, and allocating memory for input and output data. Device code in GPU is responsible for computing the radius matrix in parallel from required inputs (parallel computation block)

Article Snippet: In details, MEXCUDA is the expansion of MATLAB mex function that utilizes C/C++ for execution using C++ MEX API.

Techniques: Blocking Assay

Journal: Visual Computing for Industry, Biomedicine and Art

Article Title: Optimizing photoacoustic image reconstruction using cross-platform parallel computation

doi: 10.1186/s42492-018-0002-5

Figure Lengend Snippet: Overall flowchart of a MEXCUDA function

Article Snippet: We first need to generate a MEXCUDA function before calling it in MATLAB.

Techniques:

Journal: Visual Computing for Industry, Biomedicine and Art

Article Title: Optimizing photoacoustic image reconstruction using cross-platform parallel computation

doi: 10.1186/s42492-018-0002-5

Figure Lengend Snippet: Example process flow of the heterogeneous computing. In this figure, we create a sample MEXCUDA function for calculating the radius matrix of a scanned region to each transducer elements. The host code (CPU) is in charge of initialization and finalization blocks such as reading the input from MATLAB, performing condition checks, and allocating memory for input and output data. Device code in GPU is responsible for computing the radius matrix in parallel from required inputs (parallel computation block)

Article Snippet: We first need to generate a MEXCUDA function before calling it in MATLAB.

Techniques: Blocking Assay