1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256
|
function compareTimings(statusBefore, statusAfter)
% COMPARETIMINGS compare timing of matlab2tikz test suite runs
%
% This function plots some analysis plots of the timings of different test
% cases. When the test suite is run repeatedly, the median statistics are
% reported as well as the individual runs.
%
% Usage:
% COMPARETIMINGS(statusBefore, statusAfter)
%
% Parameters:
% - statusBefore and statusAfter are expected to be
% N x R cell arrays, each cell contains a status of a test case
% where there are N test cases, repeated R times each.
%
% You can build such cells, e.g. with the following snippet.
%
% suite = @ACID
% N = numel(suite(0)); % number of test cases
% R = 10; % number of repetitions of each test case
%
% statusBefore = cell(N, R);
% for r = 1:R
% statusBefore(:, r) = testHeadless;
% end
%
% % now check out the after commit
%
% statusAfter = cell(N, R);
% for r = 1:R
% statusAfter(:, r) = testHeadless;
% end
%
% compareTimings(statusBefore, statusAfter)
%
% See also: testHeadless
%% Extract timing information
time_cf = extract(statusBefore, statusAfter, @(s) s.tikzStage.cleanfigure_time);
time_m2t = extract(statusBefore, statusAfter, @(s) s.tikzStage.m2t_time);
%% Construct plots
hax(1) = subplot(3,2,1);
histograms(time_cf, 'cleanfigure');
legend('show')
hax(2) = subplot(3,2,3);
histograms(time_m2t, 'matlab2tikz');
legend('show')
linkaxes(hax([1 2]),'x');
hax(3) = subplot(3,2,5);
histogramSpeedup('cleanfigure', time_cf, 'matlab2tikz', time_m2t);
legend('show');
hax(4) = subplot(3,2,2);
plotByTestCase(time_cf, 'cleanfigure');
legend('show')
hax(5) = subplot(3,2,4);
plotByTestCase(time_m2t, 'matlab2tikz');
legend('show')
hax(6) = subplot(3,2,6);
plotSpeedup('cleanfigure', time_cf, 'matlab2tikz', time_m2t);
legend('show');
linkaxes(hax([4 5 6]), 'x');
% ------------------------------------------------------------------------------
end
%% Data processing
function timing = extract(statusBefore, statusAfter, func)
otherwiseNaN = {'ErrorHandler', @(varargin) NaN};
timing.before = cellfun(func, statusBefore, otherwiseNaN{:});
timing.after = cellfun(func, statusAfter, otherwiseNaN{:});
end
function [names,timings] = splitNameTiming(vararginAsCell)
names = vararginAsCell(1:2:end-1);
timings = vararginAsCell(2:2:end);
end
%% Plot subfunctions
function [h] = histograms(timing, name)
% plot histogram of time measurements
colors = colorscheme;
histostyle = {'DisplayStyle', 'bar',...
'Normalization','pdf',...
'EdgeColor','none',...
'BinWidth',0.025};
hold on;
h{1} = myHistogram(timing.before, histostyle{:}, ...
'FaceColor', colors.before, ...
'DisplayName', 'Before');
h{2} = myHistogram(timing.after , histostyle{:}, ...
'FaceColor', colors.after,...
'DisplayName', 'After');
xlabel(sprintf('%s runtime [s]',name))
ylabel('Empirical PDF');
end
function [h] = histogramSpeedup(varargin)
% plot histogram of observed speedup
histostyle = {'DisplayStyle', 'bar',...
'Normalization','pdf',...
'EdgeColor','none'};
[names,timings] = splitNameTiming(varargin);
nData = numel(timings);
h = cell(nData, 1);
minTime = NaN; maxTime = NaN;
for iData = 1:nData
name = names{iData};
timing = timings{iData};
hold on;
speedup = computeSpeedup(timing);
color = colorOptionsOfName(name, 'FaceColor');
h{iData} = myHistogram(speedup, histostyle{:}, color{:},...
'DisplayName', name);
[minTime, maxTime] = minAndMax(speedup, minTime, maxTime);
end
xlabel('Speedup')
ylabel('Empirical PDF');
set(gca,'XScale','log', 'XLim', [minTime, maxTime].*[0.9 1.1]);
end
function [h] = plotByTestCase(timing, name)
% plot all time measurements per test case
colors = colorscheme;
hold on;
if size(timing.before, 2) > 1
h{3} = plot(timing.before, '.',...
'Color', colors.before, 'HandleVisibility', 'off');
h{4} = plot(timing.after, '.',...
'Color', colors.after, 'HandleVisibility', 'off');
end
h{1} = plot(median(timing.before, 2), '-',...
'LineWidth', 2, ...
'Color', colors.before, ...
'DisplayName', 'Before');
h{2} = plot(median(timing.after, 2), '-',...
'LineWidth', 2, ...
'Color', colors.after,...
'DisplayName', 'After');
ylabel(sprintf('%s runtime [s]', name));
set(gca,'YScale','log')
end
function [h] = plotSpeedup(varargin)
% plot speed up per test case
[names, timings] = splitNameTiming(varargin);
nDatasets = numel(names);
minTime = NaN;
maxTime = NaN;
h = cell(nDatasets, 1);
for iData = 1:nDatasets
name = names{iData};
timing = timings{iData};
color = colorOptionsOfName(name);
hold on
[speedup, medSpeedup] = computeSpeedup(timing);
if size(speedup, 2) > 1
plot(speedup, '.', color{:}, 'HandleVisibility','off');
end
h{iData} = plot(medSpeedup, color{:}, 'DisplayName', name, ...
'LineWidth', 2);
[minTime, maxTime] = minAndMax(speedup, minTime, maxTime);
end
nTests = size(speedup, 1);
plot([-nTests nTests*2], ones(2,1), 'k','HandleVisibility','off');
legend('show', 'Location','NorthWest')
set(gca,'YScale','log','YLim', [minTime, maxTime].*[0.9 1.1], ...
'XLim', [0 nTests+1])
xlabel('Test case');
ylabel('Speed-up (t_{before}/t_{after})');
end
%% Histogram wrapper
function [h] = myHistogram(data, varargin)
% this is a very crude wrapper that mimics Histogram in R2014a and older
if ~isempty(which('histogram'))
h = histogram(data, varargin{:});
else % no "histogram" available
options = struct(varargin{:});
minData = min(data(:));
maxData = max(data(:));
if isfield(options, 'BinWidth')
numBins = ceil((maxData-minData)/options.BinWidth);
elseif isfield(options, 'NumBins')
numBins = options.NumBins;
else
numBins = 10;
end
[counts, bins] = hist(data(:), numBins);
if isfield(options,'Normalization') && strcmp(options.Normalization,'pdf')
binWidth = mean(diff(bins));
counts = counts./sum(counts)/binWidth;
end
h = bar(bins, counts, 1);
% transfer properties as well
names = fieldnames(options);
for iName = 1:numel(names)
option = names{iName};
if isprop(h, option)
set(h, option, options.(option));
end
end
set(allchild(h),'FaceAlpha', 0.75); % only supported with OpenGL renderer
% but this should look a bit similar with matlab2tikz then...
end
end
%% Calculations
function [speedup, medSpeedup] = computeSpeedup(timing)
% computes the timing speedup (and median speedup)
dRep = 2; % dimension containing the repeated tests
speedup = timing.before ./ timing.after;
medSpeedup = median(timing.before, dRep) ./ median(timing.after, dRep);
end
function [minTime, maxTime] = minAndMax(speedup, minTime, maxTime)
% calculates the minimum/maximum time in an array and peviously
% computed min/max times
minTime = min([minTime; speedup(:)]);
maxTime = min([maxTime; speedup(:)]);
end
%% Color scheme
function colors = colorscheme()
% defines the color scheme
colors.matlab2tikz = [161 19 46]/255;
colors.cleanfigure = [ 0 113 188]/255;
colors.before = [236 176 31]/255;
colors.after = [118 171 47]/255;
end
function color = colorOptionsOfName(name, keyword)
% returns a cell array with a keyword (default: 'Color') and a named color
% if it exists in the colorscheme
if ~exist('keyword','var') || isempty(keyword)
keyword = 'Color';
end
colors = colorscheme;
if isfield(colors,name)
color = {keyword, colors.(name)};
else
color = {};
end
end
|