Matrix Multiplication Using Parallel Computing in Matlab, with multithreadingMarch 5, 2021 \/ Lorand Gabriel Parajdi<\/span><\/a><\/p>\n\n\n\n \u25ba<\/mark><\/strong> Save the following lines of code in a file called matrix_multiplication.m, then run it using Matlab. \u25ba<\/mark><\/strong> Instead of using Matlab matrix multiplication (C = A * B), we have defined matrix multiplication using multiple <\/span>for<\/em> loops. Matlab linear algebra library, used in A*B, is already parallelized, and we want to implement parallelization by ourselves.<\/span><\/a><\/p>\n\n\n\n <\/p>\n\n\n\n Matrix Multiplication on the CPU v.s. GPU in MatlabMarch 3, 2021 \/ Lorand Gabriel Parajdi<\/mark><\/a><\/p>\n\n\n\n \u25ba<\/mark><\/strong><\/mark> Save the following lines of code in a file called matrix_multip_cpu_vs_gpu.m, then run it using Matlab. <\/p>\n\n\n\n Matrix Multiplication Using Parallel Computing in Matlab, with multithreadingMarch 5, 2021 \/ Lorand Gabriel Parajdi \u25ba Save the following lines of code in a file called matrix_multiplication.m, then run it using Matlab. \u25ba Instead of using Matlab matrix multiplication (C = A * B), we have defined matrix multiplication using multiple for loops. Matlab linear … Continue reading Code Samples<\/span><\/a><\/p>\n","protected":false},"author":1,"featured_media":0,"parent":0,"menu_order":0,"comment_status":"closed","ping_status":"closed","template":"","meta":[],"_links":{"self":[{"href":"https:\/\/www.cs.ubbcluj.ro\/~lorand\/wp-json\/wp\/v2\/pages\/3297"}],"collection":[{"href":"https:\/\/www.cs.ubbcluj.ro\/~lorand\/wp-json\/wp\/v2\/pages"}],"about":[{"href":"https:\/\/www.cs.ubbcluj.ro\/~lorand\/wp-json\/wp\/v2\/types\/page"}],"author":[{"embeddable":true,"href":"https:\/\/www.cs.ubbcluj.ro\/~lorand\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/www.cs.ubbcluj.ro\/~lorand\/wp-json\/wp\/v2\/comments?post=3297"}],"version-history":[{"count":176,"href":"https:\/\/www.cs.ubbcluj.ro\/~lorand\/wp-json\/wp\/v2\/pages\/3297\/revisions"}],"predecessor-version":[{"id":5315,"href":"https:\/\/www.cs.ubbcluj.ro\/~lorand\/wp-json\/wp\/v2\/pages\/3297\/revisions\/5315"}],"wp:attachment":[{"href":"https:\/\/www.cs.ubbcluj.ro\/~lorand\/wp-json\/wp\/v2\/media?parent=3297"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}} clear all; clc;\n \n nthreads = [ 4 ];\n %nthreads = [ 1, 2, 3, 4, 8, 12 ];\n \n % set the size\n sizes = [ 1000 ];\n %sizes = [ 100 500 1000 ];\n \n for s=1:length(sizes)\n n = sizes(s); % set the matrix size\n A = rand(n); % create random matrix\n B = rand(n); % create another random matrix\n for l=1:length(nthreads)\t % vector implementation (may trigger multithreading)\n nt=nthreads(l);\n lastnt=maxNumCompThreads(nt); % set the thread count\n parpool('threads')\n C = zeros(n);\n tic % starts timer\n parfor i = 1:n % matrix multiplication algorithm\n for j = 1:n\n sum = 0;\n for k = 1:n\n sum = sum + A(i,k) * B(k,j);\n end\n C(i,j) = sum;\n end\n end\n \n walltime(l) = toc; \t\t % wall clock time\n Speedup = walltime(1)\/walltime(l);\n Efficiency = 100*Speedup\/nt;\n printstring=sprintf(['size=%d time=%8.4f threads=%2d speedup=%4.1f efficiency=%5.1f%%'],\n n, walltime(l), nt, Speedup, Efficiency);\n disp(printstring);\n \n par = gcp('nocreate'); % release pool instance\n delete(par);\n \n end\n disp(' '); % to display matrix C --> disp(C);\n end\n \n plot(nthreads,walltime) % plot of running time and no. of threads\n hold on\n plot(nthreads,walltime,'.','MarkerSize',25)\n xlabel('No. of threads');\n ylabel('Running time')\n grid on\n hold off<\/pre><\/div>\n\n\n\n
\n\n\n\n clear all; clc;\n\n v1 = 10:10:90; % set the matrix size\n v2 = 100:100:900;\n v3 = 1000:1000:9000;\n v = [v1 v2 v3];\n %v = [v1 v2];\n\n % CPU computations\n for i = 1:length(v) \n A = rand(v(i)); % create random matrix\n B = rand(v(i)); % create another random matrix\n f_cpu = @() bsxfun(@mtimes, A, B);\n t(i) = timeit(f_cpu);\n disp(strcat( 'CPU matrix_dimension: ', num2str(v(i)), ' time: ', num2str(t(i))))\n end\n\n % GPU computations\n for i = 1:length(v)\n AA = rand(v(i),'gpuArray'); % create random matrix\n BB = rand(v(i),'gpuArray'); % create another random matrix\n f_gpu = @() bsxfun(@mtimes, AA, BB);\n tt(i) = gputimeit(f_gpu);\n disp(strcat( 'GPU matrix_dimension: ', num2str(v(i)), ' time: ', num2str(tt(i))))\n end\n\n save cpu_gpu.mat v t tt % save the data values of v, t and tt\n % in the file named: cpu_gpu.mat\n\n loglog(v, t, 'b') \n hold on\n loglog(v, tt, 'r')\n loglog(v, t, 'b.', 'MarkerSize', 20)\n loglog(v, tt, 'r.', 'MarkerSize', 20)\n grid on\n xlabel('Log(Matrix size)')\n ylabel('Execution time')\n legend('CPU', 'GPU')\n hold off<\/pre><\/div>\n\n\n\n
\n","protected":false},"excerpt":{"rendered":"