1 / 20

GPU acceleration in Matlab

GPU acceleration in Matlab. Jan Kamenick ý. UTIA Friday seminar 9.11.2012. GPU acceleration. CPU fast general-purpose GPU highly parallel handles specific tasks with large amount of data m emory transfers needed. GPU acceleration in Matlab. Build-in functions

rusti
Download Presentation

GPU acceleration in Matlab

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript

  1. GPU acceleration in Matlab Jan Kamenický UTIA Friday seminar 9.11.2012

  2. GPU acceleration • CPU • fast • general-purpose • GPU • highly parallel • handles specific tasks with large amount of data • memory transfers needed

  3. GPU acceleration in Matlab • Build-in functions • many Matlab functions support GPU acceleration natively • arrayfun • specific element-wise processing • CUDA kernels • write “.cu” files • compile to “.ptx” (parallel thread execution) • run using feval

  4. Prerequisites • Matlab 2010b or newer • Parallel Computing Toolbox ver

  5. Prerequisites >> ver ------------------------------------------------------------------------------------- MATLAB Version 7.13.0.564 (R2011b) MATLAB License Number: XXXXXX Operating System: Microsoft Windows 7 Version 6.1 (Build 7601: Service Pack 1) Java VM Version: Java 1.6.0_17-b04 with Sun Microsystems Inc. Java HotSpot(TM) 64-Bit Server VM mixed mode ------------------------------------------------------------------------------------- MATLAB                                                Version 7.13       (R2011b) Simulink                                              Version 7.8        (R2011b) Computer Vision System Toolbox                        Version 4.1        (R2011b) Curve Fitting Toolbox                                 Version 3.2        (R2011b) DSP System Toolbox                                    Version 8.1        (R2011b) Data Acquisition Toolbox                              Version 3.0        (R2011b) Filter Design HDL Coder                               Version 2.9        (R2011b) Fixed-Point Toolbox                                   Version 3.4        (R2011b) Global Optimization Toolbox                           Version 3.2        (R2011b) Image Acquisition Toolbox                             Version 4.2        (R2011b) Image Processing Toolbox                              Version 7.3        (R2011b) MATLAB Compiler                                       Version 4.16       (R2011b) MATLAB Distributed Computing Server                   Version 5.2        (R2011b) Neural Network Toolbox                                Version 7.0.2      (R2011b) Optimization Toolbox                                  Version 6.1        (R2011b) Parallel Computing Toolbox                            Version 5.2        (R2011b) Partial Differential Equation Toolbox                 Version 1.0.19     (R2011b) Signal Processing Toolbox                             Version 6.16       (R2011b) Simulink 3D Animation                                 Version 6.0        (R2011b) Statistics Toolbox                                    Version 7.6        (R2011b) Symbolic Math Toolbox                                 Version 5.7        (R2011b) Wavelet Toolbox                                       Version 4.8        (R2011b)

  6. Prerequisites >> ver------------------------------------------------------------------------------------- MATLAB Version 7.13.0.564 (R2011b)MATLAB License Number: XXXXXX Operating System: Microsoft Windows 7 Version 6.1 (Build 7601: Service Pack 1) Java VM Version: Java 1.6.0_17-b04 with Sun Microsystems Inc. Java HotSpot(TM) 64-Bit Server VM mixed mode ------------------------------------------------------------------------------------- MATLAB                                                Version 7.13       (R2011b) Simulink                                              Version 7.8        (R2011b) Computer Vision System Toolbox                        Version 4.1        (R2011b) Curve Fitting Toolbox                                 Version 3.2        (R2011b) DSP System Toolbox                                    Version 8.1        (R2011b) Data Acquisition Toolbox                              Version 3.0        (R2011b) Filter Design HDL Coder                               Version 2.9        (R2011b) Fixed-Point Toolbox                                   Version 3.4        (R2011b) Global Optimization Toolbox                           Version 3.2        (R2011b) Image Acquisition Toolbox                             Version 4.2        (R2011b) Image Processing Toolbox                              Version 7.3        (R2011b) MATLAB Compiler                                       Version 4.16       (R2011b) MATLAB Distributed Computing Server                   Version 5.2        (R2011b) Neural Network Toolbox                                Version 7.0.2      (R2011b) Optimization Toolbox                                  Version 6.1        (R2011b) Parallel Computing Toolbox                            Version 5.2        (R2011b)Partial Differential Equation Toolbox                 Version 1.0.19     (R2011b) Signal Processing Toolbox                             Version 6.16       (R2011b) Simulink 3D Animation                                 Version 6.0        (R2011b) Statistics Toolbox                                    Version 7.6        (R2011b) Symbolic Math Toolbox                                 Version 5.7        (R2011b) Wavelet Toolbox                                       Version 4.8        (R2011b)

  7. Prerequisites • Matlab 2010b or newer • Parallel Computing Toolbox ver • NVIDIA GPU with CUDA version 1.3 or higher gpuDevice

  8. Prerequisites >> gpuDevice ans = parallel.gpu.CUDADevicehandle  Package: parallel.gpu   Properties:                       Name: 'GeForce GTX 285'                      Index: 1          ComputeCapability: '1.3'             SupportsDouble: 1              DriverVersion: 5         MaxThreadsPerBlock: 512           MaxShmemPerBlock: 16384         MaxThreadBlockSize: [512 512 64]                MaxGridSize: [65535 65535]                  SIMDWidth: 32                TotalMemory: 2.1475e+009                 FreeMemory: 1.9656e+009        MultiprocessorCount: 30               ClockRateKHz: 1476000                ComputeMode: 'Default'       GPUOverlapsTransfers: 1     KernelExecutionTimeout: 1           CanMapHostMemory: 1            DeviceSupported: 1             DeviceSelected: 1 Methods, Events, Superclasses

  9. Prerequisites >> gpuDevice ans = parallel.gpu.CUDADevicehandle  Package: parallel.gpu   Properties:                       Name: 'GeForce GTX 285'                      Index: 1          ComputeCapability: '1.3'            SupportsDouble: 1              DriverVersion: 5         MaxThreadsPerBlock: 512           MaxShmemPerBlock: 16384         MaxThreadBlockSize: [512 512 64]                MaxGridSize: [65535 65535]                  SIMDWidth: 32                TotalMemory: 2.1475e+009                 FreeMemory: 1.9656e+009        MultiprocessorCount: 30               ClockRateKHz: 1476000                ComputeMode: 'Default'       GPUOverlapsTransfers: 1     KernelExecutionTimeout: 1           CanMapHostMemory: 1            DeviceSupported: 1             DeviceSelected: 1 Methods,Events,Superclasses

  10. Basic usage • Send data to GPU • either allocate there or transfer from workspace • Run Matlab functions • GPU acceleration is used automatically • Retrieve the output data

  11. GPUArray class parallel.gpu.GPUArray • main data class for GPU computations • stored in the GPU memory • create directly using static methods • copy from existing data gpuArray(img)

  12. GPUArray class • Supported data types: (u)int8, (u)int16, (u)int32, (u)int64, single, double, logical • determine the type using classUnderlying(gpuVar) • Retrieve the data using workspaceVar = gather(gpuVar)

  13. GPU accelerated Matlab functions (2012b) methods(‘parallel.gpu.GPUArray’)

  14. GPU accelerated Matlab functions (2012b)

  15. Simple example • Solve system of linear equations (Ax = b) A = gpuArray(A); b = gpuArray(b); x = A\b; x = gather(x);

  16. Simple example M = fft2(msk); • Compute convolution using FFT img = gpuArray(img); msk = padarray(msk,size(img)-size(msk),0,'post'); msk= gpuArray(msk); I = fft2(img); M = fft2(msk,size(img,1),size(img,2)); res = real(ifft2(I.*M)); res = gather(res);

  17. Linear system solution benchmark

  18. Convolution benchmark

  19. Profiling • Before optimizing (trying to use GPU) locate promising parts of code like • custom code consuming the majority of time • build-in functions that support GPUArray (consuming the majority of time) • large input/output data, simple data types • Test the speed afterwards • GPU code cannot be profiled

  20. Profiling

More Related