RE:[babintsev@mx.ihep.su: [ROOT] time consu

From: Masaharu Goto (MXJ02154@nifty.ne.jp)
Date: Mon Feb 05 2001 - 13:20:56 MET


Hello Fons and Volodja,

Yes, this is a normal case.

As a reference, I benchmarked cint, eic and g++ -O2.
 case 1. 
    cint 10.01sec    eic 16.49sec   g++ -O2 0.04sec
 case 2.
    cint 16.04sec    eic 21.86sec   g++ -O2 0.17sec

EiC is a bytecode based C interpreter whoose performance
is not bad. Cint out-performs in both cases.  Of course
both are much slower than native code.

Script interpreter is aimed at more interactivity, at the cost
of execution speed.  I do not have an intention to compete with
native code speed.  It is meaningless for interpreter. Rather,
interpreter should set its' value in interactivity. 

Thank you
Masaharu Goto


>     babintsev@mx.ihep.su,  4-FEB-2001
>
>Hello rooters.
>I have a simple program below which has different time
>consuming in the following 3 cases:
>(ROOT Version   2.25/03 )
>
> 1) program has lines:
>    .....
>       // case 1:
>       one_ev[1] = i;
>    .....
>
>    [] .L toy1.C
>    []  exam()
>       .......
>      CpuTime=6.860000 seconds
>
> 2) program has lines:
>    .....
>         // case 2:
>         one_ev[j*10+1] = i;
>    .....
>
>    [] .L toy1.C
>    []  exam()
>       .......
>      CpuTime=11.290000 seconds
>
> 3) case 2, but:
>    [] .L toy1.C++
>    []  exam()
>       ........
>      CpuTime=0.150000 seconds
>
>
> Is it a normal situation ?
>



This archive was generated by hypermail 2b29 : Fri Jun 08 2001 - 11:51:01 MEST