Quantcast

Scilab 6.0.0 vs 5.2.2 - performance decrease

classic Classic list List threaded Threaded
4 messages Options
kjubo kjubo
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Scilab 6.0.0 vs 5.2.2 - performance decrease

Dear all,

I observed a performance decrease with relatively simple code that use a for cycles.
bench_1.sci

Almost everybody is reporting massive speed up in for cycles, but in my case it is opposite.
In 5.2.2 run time is 10 seconds, while in 6.0.0 is 18 seconds.

Both Scilab versions are 64 bit, Win 7 64 bit.

Can somebody confirm my results?

Thanks in advance.

BR
J.K.
Amanda Osvaldo Amanda Osvaldo
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Scilab 6.0.0 vs 5.2.2 - performance decrease

Hi, I  confirm that Scilab 6.0.0 it's a bit slower than 5.5.2. It's the time spent in the benchmark.

5.5.2
    7.151  
    7.127  
    7.123  
    7.158  
    7.169  
    7.117 
    7.154  
    7.209  
    7.117  
    7.107

6.0.0
   8.97989
   8.965782
   8.998687
   9.013485
   9.241977
   9.100208
   9.057786
   9.00625
   8.963514
   9.205469
  
I have attached some information about my CPU in this message.

-- Amanda Osvaldo

On Fri, 2017-04-07 at 03:32 -0700, kjubo wrote:
Dear all,

I observed a performance decrease with relatively simple code that use a for
cycles.
bench_1.sci <http://mailinglists.scilab.org/file/n4036150/bench_1.sci>  

Almost everybody is reporting massive speed up in for cycles, but in my case
it is opposite.
In 5.2.2 run time is 10 seconds, while in 6.0.0 is 18 seconds.

Both Scilab versions are 64 bit, Win 7 64 bit.

Can somebody confirm my results?

Thanks in advance.

BR
J.K.



--
View this message in context: http://mailinglists.scilab.org/Scilab-6-0-0-vs-5-2-2-performance-decrease-tp4036150.html
Sent from the Scilab users - Mailing Lists Archives mailing list archive at Nabble.com.
_______________________________________________
users mailing list
[hidden email]
http://lists.scilab.org/mailman/listinfo/users

_______________________________________________
users mailing list
[hidden email]
http://lists.scilab.org/mailman/listinfo/users

test-6.0.0.csv (9K) Download Attachment
test-5.5.2.csv (8K) Download Attachment
cpuinfo (5K) Download Attachment
Heinz Nabielek Heinz Nabielek
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Scilab 6.0.0 vs 5.2.2 - performance decrease

SciLab 6 is mightily slower....
Mac OS X 10.12.4 Intel(R) Core(TM) i7-4770S CPU @ 3.10GHz   8 GB 1600 MHz DDR3
SciLab 5.5.2 -->disp(toc()) 4.211
SciLab 6.0.0 -->disp(toc()) 10.421212

SciLab 5.5.2 Mac OS X 10.7.5        2.4 GHz Intel Core 2 Duo 4 GB 667 MHz DDR2 SDRAM
                        -->disp(toc()) 11.384  


This was a mighty shock to me, but thanks for pointing it out.
Heinz
Samuel GOUGEON Samuel GOUGEON
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: SciLab euivalent of Perl's map() function?

In reply to this post by kjubo
Hello,

Le 08/04/2017 à 04:47, [hidden email] a écrit :
Hi,

I want to do the equivalent of 

    plot( xdata, xdata .^ 2, xdata .^ 3, xdata '^ 4, ... );

and benefit from the plot() color rotation, without having to repeat the equation.

In Perl I might do (something like):

    plot( xdata, map xdata .^ $_, 2 .. 5 );

Something like that in SciLab?

x = -2:0.1:2;
pw =
[1 2 3 4];
[X,P] = ndgrid(x, pw);
clf
plot(x, X.^P)


xtitle("Power laws","x")
legend(msprintf("$x^%d$\n",pw'),4)

HTH
Samuel


_______________________________________________
users mailing list
[hidden email]
http://lists.scilab.org/mailman/listinfo/users
Loading...