Question for developers - using all computational cores during optimisation - page 4
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
You will go far with such a message... Slava, by the way, is one of the main developers of MT, he knows how the algorithm works.
I'm telling you, just give me the optimisation set of the standard EA, which leads to the same results. You had some parameters optimised before, now others. Perhaps the automatic switch to genetics with all that it implies has triggered.
Add some constructiveness and the problem will be solved much faster.
If you really want to change and not just grumble (like me), why not give the developers a reproducible example where kernels are disabled/standby?
Maybe based on the standard EA (if it's also playable), but with as much detail as possible, so they can replicate the behaviour for themselves.
is that clearer?
only local agents are used, 6 out of 8 are enabled, 3 fall off immediately after the first batch of jobsis that clearer?
only local agents are used, 6 out of 8 are enabled, 3 fall off immediately after the first batch of jobsIt's much more constructive that way.
Attach the tester's log and the log of one of the agents who finished early:
This is much more constructive.
Attach the tester log and the log of one of the agents who finished early:
Log of the tester, the working agent and the one that failed:
Log of tester, working agent and failed agent:
Now we wait for @Slava's response
Looks like genetics stopped engaging some of the cores after generation 3:
Figured there was no point?
>Slava, by the way, one of the main developers of MT
Well then Slava - all hope is on you, we pray and we raise our voices .... help us from non-working network agents :-)
I would also like to thank Andrey Khatimlianskii for the logs
>Slava, by the way, one of the main developers of MT
Well then Slava - all hope is on you, we pray and we raise our voices .... help us from non-working network agents :-)
I would also like to thank Andrey Khatimlianskii for the logs
Now we wait for @Slava's response
Looks like genetics stopped engaging some of the cores after generation 3:
Figured there was no point?
No.
There's more in the log.
Confirmed by agent logs
I would like to point out that there are actually two problems with downtime.
With genetics there is a waiting period for the end of the generation calculation. It is not clear whether rebalancing of the job package is possible in this case.
With slow optimization, downtime of previously freed agents can be avoided by dynamically reallocating jobs. Developers did not do this, and now jobs are being distributed at the beginning of optimization. They did not do this because the same distribution algorithm is applied when using cloud agents, and taking jobs away from them is "inappropriate". It is worth separating the methodology for on-premises and cloud agents.
In the meantime, the developers have, relatively recently, improved the methodology slightly, leaving a small reserve for agents who have finished work early. Unfortunately, this doesn't always save the day. In addition, this reserve is the remainder of the division of tasks by the number of agents, so it can be equal to zero.
No.
There is another entry in the log
Confirmed by the agent's logs.
So that's afterwards, at the end. The agents dropped out earlier, at 01:00:50, and you can see it in the log and the video.