If i have 8 cores , i need a single Test to be processed by the 8 cores .. is it possiple ???

New comment
 
there is some thing i don't know if it can be possible or not in metatrader.. but i need it badly

i know that each single run can be processed by single core .. i need this single run to be processed by the 8 processors (or more that i have) in both cases simple test and optimization process. and also if possible in cloud issues .. i think this will be faster in my case ..

the problem is because I've been working on an expert for along time .. and the final version is now too heavy .. takes long time in single test from 2000 to end of 2012

this is not actually my problem .. the problem appears in optimization process (from 2008 to end of 2012) if i have say 4000 different test case .. only 100 (for example) that appear and then looks like it can't handle any more test cases .. still running but with zero progress

Note: i have made my code very simple .. the expert doesn't use any indicator, no arrays are global .. the local arrays in functions are of size 4 max

so what do you think?
Documentation on MQL5: Standard Constants, Enumerations and Structures / Environment State / Running MQL5 Program Properties
Documentation on MQL5: Standard Constants, Enumerations and Structures / Environment State / Running MQL5 Program Properties
  • www.mql5.com
Standard Constants, Enumerations and Structures / Environment State / Running MQL5 Program Properties - Documentation on MQL5
 

To reply to your topic question : no you can't.

Anyway, you have to search others ways, if you have such issue, there is probably at least 1 problem somewhere. You have to find and resolve this problem. How much time take 1 pass ? Is this problem arise when running optimization on you computer or in the Cloud, or both ? Is there any output in the logs (please post it).

There is a contradiction between "the final version is now too heavy" and "i have made my code very simple".

 

Many thanks "angevoyageur" for your reply ..

How much time take 1 pass ?

from 5 to 7 minutes from 2000 to end of 2012

Is this problem arise when running optimization on you computer or in the Cloud, or both ? Is there any output in the logs (please post it).

in using cloud yes,  2013.10.02 14:19:09 MQL5 Cloud Europe 2 genetic pass (0, 337) tested with error "no memory in OnTick function (cannot get 5088 Kb, used 0 Mb)" in 16 ms (PR 161)

in any other case no , there is no error messages

There is a contradiction between "the final version is now too heavy" and "i have made my code very simple". 

what i mean by Final Version is after all the requirements of the problem are all added(all the needed features are present)

 
eman_samy_alex:

This error "no memory in OnTick function" has already been reported, but unfortunately not the solution (as far as I know). Can you post your code, eventually by pm, so I can investigate ?

 
You can also see if this topic helps you.
 

Since you seem to have coded everything yourself with a very heavy result and no indicators, you might want to go for OPENCL

 How to Install and Use OpenCL for Calculations

It's always a nuisance to recode, but (apart from the reported issue) I think it might be a more permanent solution 

 
angevoyageur:
You can also see if this topic helps you.

it seems like i need to recheck again my code with you .. but, as i told you, i'm waiting permission to share it

 
jlwarrior:

Since you seem to have coded everything yourself with a very heavy result and no indicators, you might want to go for OPENCL

 How to Install and Use OpenCL for Calculations

It's always a nuisance to recode, but (apart from the reported issue) I think it might be a more permanent solution 

it seems very interesting to me as i didn't work with OPENCL before .. i will go on and hope it helps .. many thanks jlwarrior
 
eman_samy_alex:
there is some thing i don't know if it can be possible or not in metatrader.. but i need it badly

i know that each single run can be processed by single core .. i need this single run to be processed by the 8 processors (or more that i have) in both cases simple test and optimization process. and also if possible in cloud issues .. i think this will be faster in my case ..

the problem is because I've been working on an expert for along time .. and the final version is now too heavy .. takes long time in single test from 2000 to end of 2012

this is not actually my problem .. the problem appears in optimization process (from 2008 to end of 2012) if i have say 4000 different test case .. only 100 (for example) that appear and then looks like it can't handle any more test cases .. still running but with zero progress

Note: i have made my code very simple .. the expert doesn't use any indicator, no arrays are global .. the local arrays in functions are of size 4 max

so what do you think?
i did not think it is possible
New comment