You may have noticed some recent changes in the format of this blog. Here’s what to expect on a regular basis – two topics
per week.
On Tuesdays Doug will provide MATLAB tutorials.
On Fridays guest bloggers Jiro, Brett and Bob will highlight File Exchange submissions.
Bob's pick this week is timeit by Steve Eddins.
If you follow his weekly blog Steve on Image Processing you may have already seen this recent leap day post, Timing code in MATLAB. If you haven't read it yet and care about MATLAB performance, check it out. Steve's explanation is nice. He mentions other
MATLAB performance resources. And like all good M-file submissions it includes examples like this.
A = rand(12000, 400);
B = rand(400, 12000);
f = @() sum(A.' .* B, 1);
timeit(f)
ans =
0.14863
What I love about timeit is what it does automatically so I don't have to think about. It warms up the code, decides how many times to repeat the
operation, accounts for overhead and determines how long the operation took using statistics. Thank you Steve!
Do you have a favorite tip for measuring performance to share?
评论
要发表评论,请点击 此处 登录到您的 MathWorks 帐户或创建一个新帐户。