+
−clear all;
+
−compile
+
−
+
−Nsamples   = 10;
+
−Nstates    = 10;
+
−Nstatesout = 1;
+
−Ninputs    = 2;
+
−Noutputs   = 2;
+
−
+
−
+
−Nruns = 100;
+
−
+
−
+
−%%
+
−[yx,xx,lxx,y,x,lx] = validate_mex(Nsamples, Nstates, Nstatesout, Ninputs, Noutputs);
+
−
+
−sum(sum(yx-y))
+
−% sum(sum(xx-x))
+
−sum(sum(lxx-lx))
+
−
+
−return
+
−
+
−%%
+
−
+
−[tmex,yx,xx] = do_a_run(Nruns, Nsamples, Nstates, Nstatesout, Ninputs, Noutputs);
+
−[tmat,y,x] = do_a_run_mat(Nruns, Nsamples, Nstates, Nstatesout, Ninputs, Noutputs);
+
−
+
−tmex
+
−tmat
+
−tmat/tmex
+
−
+
−
+
−return
+
−
+
−%% Run-time Vs Nstates
+
−
+
−Nsamples   = 1000;
+
−Nstatesout = 1;
+
−Ninputs    = 2;
+
−Noutputs   = 10;
+
−Nruns      = 10;
+
−Nstates = 5:20:250;
+
−tmex = zeros(length(Nstates),1);
+
−tmat = zeros(length(Nstates),1);
+
−for jj=1:length(Nstates)
+
−  
+
−  disp('-----------------')
+
−  Ns = Nstates(jj) 
+
−  [tmex(jj),y,x] = do_a_run(Nruns, Nsamples, Ns, Nstatesout, Ninputs, Noutputs);
+
−  [tmat(jj),y,x] = do_a_run_mat(Nruns, Nsamples, Ns, Nstatesout, Ninputs, Noutputs);
+
−  tmex(jj)
+
−  tmat(jj)
+
−  drawnow
+
−end
+
−
+
−%%
+
−figure
+
−plot(Nstates, tmex, 'r-', Nstates, tmat, 'b-');
+
−legend('mex', 'matlab');
+
−ylabel('Run-time [s]');
+
−xlabel('Nstates')
+
−s = sprintf('Nsamples=%d, Ninputs=%d, Noutputs=%d', Nsamples, Ninputs, Noutputs);
+
−title(s)
+
−
+
−
+
−%% Run-time Vs Inputs
+
−
+
−Nsamples   = 10000;
+
−Nstatesout = 1;
+
−Ninputs    = 1:2:100;
+
−Noutputs   = 10;
+
−Nruns      = 10;
+
−Nstates    = 10;
+
−tmex = zeros(length(Ninputs),1);
+
−tmat = zeros(length(Ninputs),1);
+
−for jj=1:length(Ninputs)
+
−  
+
−  disp('-----------------')
+
−  Ni = Ninputs(jj) 
+
−  [tmex(jj),y,x] = do_a_run(Nruns, Nsamples, Nstates, Nstatesout, Ni, Noutputs);
+
−  [tmat(jj),y,x] = do_a_run_mat(Nruns, Nsamples, Nstates, Nstatesout, Ni, Noutputs);
+
−  tmex(jj)
+
−  tmat(jj)
+
−  drawnow
+
−end
+
−
+
−%%
+
−figure
+
−plot(Ninputs, tmex, 'r-', Ninputs, tmat, 'b-');
+
−legend('mex', 'matlab');
+
−ylabel('Run-time [s]');
+
−xlabel('Ninputs')
+
−s = sprintf('Nsamples=%d, Nstates=%d, Noutputs=%d', Nsamples, Nstates, Noutputs);
+
−title(s)
+
−
+
−
+
−%% Run-time Vs Outputs
+
−
+
−Nsamples   = 1000;
+
−Nstatesout = 1;
+
−Ninputs    = 2;
+
−Noutputs   = 1:2:100;
+
−Nruns      = 10;
+
−Nstates    = 10;
+
−tmex = zeros(length(Noutputs),1);
+
−tmat = zeros(length(Noutputs),1);
+
−for jj=1:length(Noutputs)
+
−  
+
−  disp('-----------------')
+
−  No = Noutputs(jj) 
+
−  [tmex(jj),y,x] = do_a_run(Nruns, Nsamples, Nstates, Nstatesout, Ninputs, No);
+
−  [tmat(jj),y,x] = do_a_run_mat(Nruns, Nsamples, Nstates, Nstatesout, Ninputs, No);
+
−  tmex(jj)
+
−  tmat(jj)
+
−  drawnow
+
−end
+
−
+
−%%
+
−figure
+
−plot(Noutputs, tmex, 'r-', Noutputs, tmat, 'b-');
+
−legend('mex', 'matlab');
+
−ylabel('Run-time [s]');
+
−xlabel('Noutputs')
+
−s = sprintf('Nsamples=%d, Nstates=%d, Ninputs=%d', Nsamples, Nstates, Ninputs);
+
−title(s)