This page compares benchmark data for different Matlab versions on a single computer.

 

Procedure:

-1- copy "c:\Program Files\MATLAB\R2013a\toolbox\matlab\general\bench.m" to "mybench.m"

-2- run "mybench.m" subsequentlty in 2012a (32 bit), 2013a (64 bit), 2016a (64 bit). This runs 5 repetitions of the benchmark.

 

Date:

August 15, 2016

 

Results:

Matlab 2012a (32 bit)

   LU          FFT          ODE        Sparse    2-D          3-D

    0.0525    0.1022    0.0524    0.1203    0.2312    0.6745

    0.0602    0.1075    0.0520    0.1276    0.2285    0.6913

    0.0646    0.1032    0.0523    0.1119    0.2280    0.6711

    0.0546    0.1056    0.0518    0.1120    0.2288    0.6992

    0.0539    0.1025    0.0520    0.1165    0.2374    0.7007

Matlab 2013a (64 bit)

   LU          FFT          ODE        Sparse    2-D          3-D

    0.0473    0.0729    0.0545    0.0948    0.2308    0.6913

    0.0494    0.0778    0.0560    0.1083    0.2502    0.6734

    0.0585    0.0767    0.0567    0.0944    0.2184    0.6804

    0.0539    0.0754    0.0552    0.0951    0.2188    0.6858

    0.0520    0.0761    0.0557    0.0948    0.2257    0.6798

Matlab 2014b (64 bit)

   LU          FFT          ODE        Sparse    2-D          3-D

    0.0389    0.0670    0.0729    0.0886    0.7948    0.5959

    0.0472    0.0691    0.0739    0.0927    0.7582    0.6114

    0.0427    0.0675    0.0737    0.0996    0.7288    0.6271

    0.0391    0.0682    0.0740    0.0978    0.7316    0.6303

    0.0388    0.0717    0.0751    0.0955    0.7096    0.6469

Matlab 2016a (64 bit)

   LU          FFT          ODE        Sparse    2-D          3-D

    0.0712    0.1004    0.0542    0.0888    0.7549    0.6586

    0.0823    0.1003    0.0527    0.0892    0.7273    0.6345

    0.0816    0.0977    0.0523    0.0885    0.7547    0.6280

    0.0842    0.1008    0.0542    0.0954    0.6634    0.6355

    0.0837    0.1000    0.0514    0.0919    0.6641    0.6440

 

Computer data:

CPU: i7-4770K CPU @ 3.50GHz

Graphics: GeForce 8400 GS/PCIe/SSE2

RAM: 8.00 GB

OS: Win 10 Pro

 

2nd experiment: Run the Matlab 2016a version of "bench.m" compatible Matlab versions. Note that this version of bench.m contains a call to animatedline.m that does not exist in Matlab versions that predate 2014b.

Results:

Matlab 2014b (64 bit)

   LU          FFT          ODE        Sparse    2-D          3-D

    0.1173    0.1585    0.0794    0.0913    1.0334    1.3668

    0.1206    0.1584    0.0834    0.0921    1.0436    1.3828

    0.1219    0.1566    0.0824    0.0918    1.0451    1.3808

Matlab 2016a (64 bit)

   LU          FFT          ODE        Sparse    2-D          3-D

    0.3106    0.1645    0.0924    0.0905    1.0420    1.1638

    0.3219    0.1661    0.0604    0.0909    1.0129    1.1411

    0.3183    0.1648    0.0626    0.1067    1.0440    1.1668

Matlab 2016b (64 bit) (prerelease)

    0.3359    0.1725    0.0572    0.1058    1.0070    1.1344

    0.3140    0.1559    0.0491    0.0914    1.0058    1.1361

    0.3156    0.1558    0.0529    0.0908    1.0078    1.1164