MT5 terminal updated today and the "Optimisation" window does not show up during the test - page 14

 
Yury Kirillov:
Come on, a standard muwling-expert from the delivery, one year timeframe, enumeration of all parameters - it's easy to get at least a google of options! :-)

What's the practical use of that?

 
Petros Shatakhtsyan:

Every developer, after seeing the results of the optimisation, determines from the first lines whether it makes sense to continue or not. That's why you should allow to see the results before the end of optimization, without interruption.

The tab "Optimization results" should be returned, and the "Refresh table" button could be added at the top of the table.

That would make more sense.

+100500

Now optimization is like buying a "Cat in a Bag"!

It's all sad.

 

In my opinion, it is sufficient to fix the optimisation interruption. So that you can press the Stop button, look at the table, and continue if necessary. And then - forming table of results in real-time is not so critical. Although, I find rather strange statement that table sorting takes resources - after all, pass cache must have possibility to be sorted by different keys anyway, and at run time it is sorted anyway...

But, as I've already said - optimization interruption is much more critical. It's not quite clear why optimization starts all over again now as everything was... Apparently, the order of work with pass cache is being revised, otherwise I can't explain disappearance of ability to continue optimization.

Well... let's see what's being pumped out instead.

 
Georgiy Merts:

In my opinion, it is sufficient to fix the optimisation interruption. So that you can press the Stop button, look at the table, and continue if necessary. And then - forming table of results in real-time is not so critical. Although, I find rather strange statement that table sorting takes resources - after all, pass cache must have possibility to be sorted by different keys anyway, and at run time it is sorted anyway...

But, as I've already said - optimization interruption is much more critical. It's not quite clear why optimization starts all over again now as everything was... Apparently, the order of work with pass cache is being revised, otherwise I can't explain disappearance of ability to continue optimization.

Well... let's see what's being pumped out instead.

Maybe dynamic table handling is free during testing on one computer. But when there is optimization on third-party agents, there is parallel running of test agents (especially when there are dozens or hundreds of them (farms)). That's when the problems come.

 
Georgiy Merts:

In my opinion, it is sufficient to fix the optimisation interruption. So that you can press the Stop button, look at the table, and continue if necessary. And then - forming table of results in real-time is not so critical. Although, I find rather strange statement that table sorting takes resources - after all, pass cache must have possibility to be sorted by different keys anyway, and at run time it is sorted anyway...

But, as I've already said - optimization interruption is much more critical. It's not quite clear why optimization starts all over again now as everything was... The order of pass cache handling must be reworked. I cannot explain disappearance of the ability to continue optimization in any other way.

Well... let's see what is pumped out instead.

Interrupting optimization will most likely always result in recalculation of current passes counted at XX%.

For suspended passes should not block the use of cores for selling to the cloud or calculations from another tester.

Consequently, on average we will lose about half of the work of the cores on the current pass.

 
Yury Kirillov:

Interrupting the optimisation will most likely always result in a recalculation of the current passes counted at XX%.

For the suspended passes should not block the use of cores for selling to the cloud or calculations from another tester.

Consequently, on average we will lose about half of the work of the cores on the current pass.

This is correct. All calculations which were not completed at the moment of interruption will be re-generated by randomizer and recalculated already on new data sets, no issues here.

But only all calculations which by that moment have ALREADY been fully calculated and put into the optimization cache must "stay in place" and optimization must be continued. And now - at interruption it looks very much like optimization starts all over again. This is not a problem if the optimization lasts 15 minutes. But if optimization lasts 15 hours - I wouldn't want to lose the cache.

 
good evening, why can't I access my platform?
 
Сергей Таболин:

It's a total cakewalk... Optimisation in 25 passes, OHLC on M1 went over two hours...

Cool!...

Use builds 1755, 1795. they're fine.
 

We have adopted a compromise solution: we show the tab of optimization results, but we update it only when we go to this tab. That is, while you look at the results in the tab, new results do not appear, for example, if you go to the testing log, return to the optimization tab, when you return to the optimization tab, the data on the passes already made will be updated.

A build 1834 dated 30 May has been published on MetaQuotes-Demo.

Now you can select results of previous optimizations in the optimization tab, if there are actual optimization caches. You can see the results of past optimizations, you can change the optimization criterion (in the optimization tab) and see new figures and graphs, you can continue optimization that was not completed in the past.

 
Slava:

We have adopted a compromise solution: we show the tab of optimization results, but we update it only when we go to this tab. That is, while you look at the results in the tab, new results do not appear, for example, if you go to the test log, return to the optimization tab, when you return to the optimization tab, the data on the passes already made will be updated.

A build 1834 dated 30 May has been published on MetaQuotes-Demo.

Now you can select results of previous optimizations in the optimization tab, if there are actual optimization caches. You can view the results of past optimizations, change the optimization criterion (in the optimization tab) and see new figures and graphs. You can also continue optimizations that were not completed in the past.


Thank you very much!