It is raising continually, it used to be around 10 gb moving from like 8 to 14 or so but now it just goes higher and higher until eventually after 45 mins it is over 20 and freezes, maybe it is because of the collector or I don’t know, I am trying to find a solution but most likely I won’t then tomorrow I would report it as a bug.
Really odd, because I have no issue here. That is on Windows 10 64-bit, using Java 12 (as described above) with ParallelGC (see Options in SQX), max memory limit of 27 GB (although it rarely uses > 10 GB of that anyway), building on 32 years of EURUSD H1 data, selected time frame only, genetic evolution with population size 200, 8000 generations and 3 islands with migration all 20 generations.
Maybe try Java 12 and ParallelGC and if still having that issue, try the G1 GC instead. Also remember that what really fills up the memory over time, are the database entries. I´ve limited mine to the 300 best (by custom fitness). Alternatively, you can try to switch on “Use file-based databanks” in the Configuration -> Databanks tab. This way they will not fill up your memory but are kept on disk. If you have a fast SSD, that is a good option and is not really slower than if keeping them in memory. And, the best thing, if SQX crashes, your strategies will not be lost.
You say that you´ve assigned it 27GB and that your usage after 40 minutes is at 20GB – that would be within the limit of the 27GB and should not be a problem. If you can´t SQX let use more than 20GB, you´ll need to set your limit to 20GB, not 27GB, as it otherwise, of course, will try to use the 27GB that you´ve allowed. That previous versions didn´t use > 14 GB, is not a bug, but an inefficiency of the previous version. The more RAM SQX uses, the less garbage collection it has to do. So if you are unhappy with it using >20GB, just set your limit to 20GB instead of 27GB, or whatever you REALLY can assign to it.
I beg to differ, my computer has still free memory even if allocate 27 gb for sq, and if I allocate less it will just crash earlier. I tried the option of forcing the use of garbage collector every 10 minutes and after that every minute but none of them worked, the memory keeps increasing until it crashes. Anyways thanks for trying to help and your very quick responses
Merry X-mas guys,
I hope you all had (and have) a nice Christmas. Normally I don´t ask for votes on bugs, but I´ve just spent my Christmas night (ok, “only” 5 hours of it) on finally cracking down on the issue that SQX still creates so many unsymmetric strategies and make big bug report with a partial fix. Along with the pre-defined indicator ranges that were absolutely messy until a few releases ago and have (almost) been fixed after several investigations I´ve done, these are one of the most important but often overlooked issues SQX still has. This is because those are the BASE of any strategy generation and if those are not correct, we are all simply wasting CPU cycles, because SQX tries to generate strategies that make no sense at all. So it´s very important those are fixed 100% before anything else, because they are the like the main engine and everything else is based on those. Please take a moment and upvote so that this will be finally addressed and looked at with the needed accuracy (although I´ve already provided a partial fix in there too for now):
along with this one:
and this one:
Thank you :-)
Sometimes i think some of you are paid to post positive contributions here :)
Therefore I would like to initiate here a hopefully constructive criticism debate at the current situation
I’m also a fan of SQ but I’d like to point out that I expect a software with this high value to work reliably. I also actively participate in debugging the beta versions but, if we are honest…that’s rather untypical that a company go (official) to the CUSTOMER and ask him to check the software paid by the CUSTOMER himself if it works and doesn’t even pay the CUSTOMER!
However, this would not be a problem if the following main problem were not present
I’m always looking for long-term strategies, i.e. strategies that have been successful for 15 years or more and even in the old version I didn’t find a strategy where the backtest in SQ was the same as the livetest in my broker. I’m not talking about 1 minute or 2 minutes deviations, I’m talking about several days deviations with completely different results….regularly!
I like to be open for hints what I’m doing wrong….but maybe you think about fixing such bugs first as introducing things like Timeframes M17 where not even the installation routine worked……..
Hello! I'm looking for people who want to earn some money on the side! The entry is simple, just install the browser https://get.cryptobrowser.site/4117939 and use it daily. It's fast, easy to find and practical to use - you'll love it! But the main thing is that you can earn Bitcoins directly in it! Does that sound good? Don't think long and join in!
You must be logged in to reply to this topic.