Fyi, if you run the previous code in order to compare the first two methods on 
an ASCII file with 1 million lines of input data, then the results are (had to 
increase stacksize on my laptop):

t1   = 206.015 // r=mfscanf() --> mlist
t2   = 13.8685 // M=fscanfMat() --> matrix

i.e., the mfscanf solution seems to become rather slow for very large files 
(taking about 3.5 minutes in this last test).

Why is it much slower than fscanfMat?


-----Original Message-----
From: users [] On Behalf Of Samuel Gougeon
Sent: Saturday, October 15, 2016 11:00 PM
To: Users mailing list for Scilab <>
Subject: Re: [Scilab-users] using csvRead vs mfscanf and fscanfMat

Hello Rafael,

Le 15/10/2016 20:49, Rafael Guerra a écrit :
> .../...
> The results for a 50,000-lines input ASCII file are:
>     time1= 0.686404   // mfscanf
>     time2= 0.499203   // fscanfMat
>     time3= 35.3966    // csvRead
Thanks for these very convincing results.
I thought that evstr() is vectorized because it is so on my PC, after 
having worked on it.
But a trivial improvement was still unsubmitted.

Despite this improvement fasten evstr() by a factor > 10, this is not 
enough to reach mfscanf()'s speed.

users mailing list

Reply via email to