Why is MATLAB UI so slow?

What may cause MATLAB UI to be so slow?

Executing

bench

ranks this machine second, returning

    0.0908    0.0889    0.1109    0.2294    0.3621    0.1953

However, if I execute a comment in the command prompt

% hello world

takes a couple seconds before coming back to the command prompt. After executing the command several times, the delay goes away. This is just an example. I'm showing graphs, using the editor to execute blocks, debug, etc. and in general behaves slowly. If I time commands using tic toc, there is a delay that does not contribute to the timing. If I execute

ver

some of the output is:

MATLAB Version: 8.2.0.701 (R2013b)
Operating System: Microsoft Windows 8 Version 6.2 (Build 9200)
Java Version: Java 1.7.0_11-b21 with Oracle Corporation Java HotSpot(TM) 64-Bit Server VM mixed mode

MATLAB                                                Version 8.2        (R2013b)
Image Processing Toolbox                              Version 8.3        (R2013b)
Parallel Computing Toolbox                            Version 6.3        (R2013b)
Signal Processing Toolbox                             Version 6.20       (R2013b)
Statistics Toolbox                                    Version 8.3        (R2013b)

[Ashish Uthama]

" If I time commands using tic toc, there is a delay that does not contribute to the timing"

A thought - Are you using the default path? If the path has network folders, the network speed might potentially impact this.

[Vicente Parot]

System load is a possibility; but I have no other programs running and bench is not affected. I have 6GB free RAM. The disk may be as well, this computer has a solid state drive which I have no experience using with MATLAB before. When reading lots of data from the disk, some portion that was read earlier seems to load faster the second time. I didn't notice this when using HDD disks in other computers before.

[Royi Avital]

Is build 8.2.0.701 the is the release version of MATLAB 8.2 (R2013b)?

Answer 1

[Walter Roberson]

Try

rehash toolboxreset
rehash toolboxcache

and see if things speed up after that.

When you use your own code, the first time (in each MATLAB session, or since the last "clear all") any routine is executed it must be parsed and converted into internal code. Using pcode can reduce the parsing time.