not enough memory

I have got the message i:n retest

“Not enough Memory

It was not possible to allocate enough Memory for all the history data. Please reduce the number of symbols loaded or their precision.”

But there is enough Memory…I have got this message several times and had to install SQ new then.

Please help me, it is much work to install it new and Import all the data new all days.

Try this

https://strategyquant.com/forum/topic/2968-how-to-tune-sq-for-max-performance-and-reduce-memory-usage-by-4-to-5-times/

Try this

 

https://strategyquant.com/forum/topic/2968-how-to-tune-sq-for-max-performance-and-reduce-memory-usage-by-4-to-5-times/

Thank you, mabi, but thats not the Problem. SQ is refusing every retest after this message.

okey, I would save the files then and reinstall the program and then visit http://www.java.com for an update. It never happend to me not even when i ran the first trial on a 2 gig VPS server for 1 week without stopping it and then retested it.But I did have some weird problems on this putor but that disappeared when i updated at www. java.com.

okey, I would save the files then and reinstall the program and then visit http://www.java.com for an update. It never happend to me not even when i ran the first trial on a 2 gig VPS server for 1 week without stopping it and then retested it.But I did have some weird problems on this putor but that disappeared when i updated at www. java.com.

Thank you Super-Mabi, now it works!! 🙂

This happens if you have the “randomize history data” activated with a lot of Monte Carlo runs. SQ generates history data X times (as per MC amount settings) and you need to be able to hold that in memory. For example, when I use this kind of test on 30 years of H1 data with 500 MC runs, RAM usage goes up to ~62gb. It effectively is like running the test on 500 symbols at once RAM-wise. So if you have low amounts of RAM, you need to cut down the amount of MC runs when using the “randomize history data” option.

A java-reinstall cannot have fixed this, but maybe you did also reset your robustness test settings for some reason and did decheck “randomize history data” and that´s why it works now.

This happens if you have the “randomize history data” activated with a lot of Monte Carlo runs. SQ generates history data X times (as per MC amount settings) and you need to be able to hold that in memory. For example, when I use this kind of test on 30 years of H1 data with 500 MC runs, RAM usage goes up to ~62gb. It effectively is like running the test on 500 symbols at once RAM-wise. So if you have low amounts of RAM, you need to cut down the amount of MC runs when using the “randomize history data” option.

 

A java-reinstall cannot have fixed this, but maybe you did also reset your robustness test settings for some reason and did decheck “randomize history data” and that´s why it works now.

Thank you, dear geektrader 🙂

I guess SQ4 will have the option  to run monte carlo only on strategies that have been selected for data bank automatically. That would be sweet.

Yes, might be, but even then, using randomized history data for 500 runs (or whatever amount) simply means SQ has to create 500 different history data-sets by randomization. So RAM-usage won´t be different for that.