From: Vincent LAFAGE <lafage@ipno.in2p3.fr>
Subject: Re: Parallel_Simulation
Date: Tue, 26 Feb 2013 08:20:46 +0100
Date: 2013-02-26T08:20:46+01:00 [thread overview]
Message-ID: <kghnob$3rp$1@ccpntc8.in2p3.fr> (raw)
In-Reply-To: <973a091e-4650-41d8-bddf-f229abcb5adc@googlegroups.com>
Le 25/02/2013 19:18, Shark8 a �crit :
> On Monday, February 25, 2013 11:17:23 AM UTC-6, Vincent LAFAGE wrote:
>>
>> Would you think of another way?
>
> Well, you could forgo the independent RNG requirement and make a protected object that is say a stack filled with random values and is replenished as-needed.
This was indeed my first attempt. But it turned out that the workers for
the computation were not so slow (quite optimized) compared to the
generator, and would dry the RNG stack quickly. So I try to explore the
more parallel way.
> OR you could have your different RNGs nested into the task itself (say the declaration area) -- then make the seed value take values from the protected object above.
Could be another interesting try.
Thanks!
next prev parent reply other threads:[~2013-02-26 7:20 UTC|newest]
Thread overview: 11+ messages / expand[flat|nested] mbox.gz Atom feed top
2013-02-25 10:20 Parallel_Simulation Vincent LAFAGE
2013-02-25 12:40 ` Parallel_Simulation John B. Matthews
2013-02-25 17:17 ` Parallel_Simulation Vincent LAFAGE
2013-02-25 18:18 ` Parallel_Simulation Shark8
2013-02-26 7:20 ` Vincent LAFAGE [this message]
2013-02-25 18:30 ` Parallel_Simulation John B. Matthews
2013-02-26 7:13 ` Parallel_Simulation Vincent LAFAGE
2013-02-25 21:04 ` Parallel_Simulation Simon Wright
2013-02-25 21:40 ` Parallel_Simulation gautier_niouzes
2013-02-26 7:09 ` Parallel_Simulation Vincent LAFAGE
2013-02-28 11:47 ` Parallel_Simulation John B. Matthews
replies disabled
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox