Issue with MetaTrader 5 Strategy Tester Agent

 

I am not sure where this should go and this is my first time here. The https://www.mql5.com/en/contact portion of the website does not give the option to address this issue.

I am having an issue with the MetaTrader 5 Strategy Tester (https://cloud.mql5.com/en) is the thing that users can install to make small sums of money from selling their CPU time and resources.

The issue I am having is it not reading the 12 threads of my i7-8700K,like it used to. I recently migrated my Windows 10 installation to my NVME M.2 SSD and,prior to this,MetaTester 5 was installed to that drive,which was otherwise used for games.

Now,it no longer detects the 12 threads of my 6C/12T CPU and only reads six cores,while everything else does. The PR of each thread is significantly lower than before too and as such,it is making me next to nothing.

I would really like this fixed somehow (if anyone knows how), as it is now pointless and redundant to use. If this is the wrong place to put this question, please forgive and a mod may put it elsewhere, so long as I can get some help. Otherwise, I think there is a bug in the software. Only 6 threads used

 

Not a bug, but a feature.

Forum on trading, automated trading systems and testing trading strategies

Metatester 5 agent manager strategy tester does not see processor threads.

Renat Fatkhullin , 2019.11.27 06:12

Due to the apparent lack of memory with an excessive number of agents and a decrease in the speed of calculations on hyper-trading kernels, we decided to limit ourselves to only physical kernels when working in cloud.

Now in Cloud they drive very expensive tasks, and the vast majority of Cloud participants have a banal lack of memory with many agents. For example, 2 GB per agent and 16 agents already require at least 32 GB of memory, although usually 8-16 GB are available.

As a result, many passes fail, wasting the network and agents. But most importantly, they are slowing down the calculation due to expectations of results and restarting tasks.

We have long been evaluating the approximate resource sufficiency of agents before issuing tasks to them, and one of the most effective is to work only on the physical cores in cloud.

Locally, you can use all the kernels as you can easily control their shutdown.


 
Alain Verleyen:

Not a bug, but a feature.


I do have 32GB of memory. I have seen the MetaTester process use all of it as well as expand the page file a ton but it was working just fine. You would think my 32GB would be enough.

 
This new "feature" may be nice for the network but it makes this software virtually pointless to anyone and everyone wanting to sell their resources. Unless it is addressed differently, nobody will ever have incentive to use it. The amounts I was getting from it has been reduced to pretty much nothing. Selling my resources was the only reason I used it,to make some pocket change. It is now redundant and makes basically nothing anymore.
 
Alain Verleyen:

Not a bug, but a feature.


Hi Alain,

I am afraid we can not really sell this as a feature, since it is still rather half-backed.

I have invested some money in my local network farm - first of all not for selling it but to use it on my own local network farm.
(and it exceeds the 2 GB/Processor you mentioned). Now I see that your decision is cutting down the benefits of my investment significantly.
While the agents available are cut by half, it takes now even 3 times as much time to do my calculations as it did before.
You can check not only for the kernels / processors installed but also for the memory available to assure performance and install agents appropriately.
Please rethink your decision and allow at least for the local network farms to run the machines at 100% of their capacity. At the moment the machines do not exceed 50% CPU and 25-30% Memory even when all agents are busy.

Many thanks for your understanding and best regards
Peter

 
2420044:

Hi Alain,

I am afraid we can not really sell this as a feature, since it is still rather half-backed.

I have invested some money in my local network farm - first of all not for selling it but to use it on my own local network farm.
(and it exceeds the 2 GB/Processor you mentioned). Now I see that your decision is cutting down the benefits of my investment significantly.
While the agents available are cut by half, it takes now even 3 times as much time to do my calculations as it did before.
You can check not only for the kernels / processors installed but also for the memory available to assure performance and install agents appropriately.
Please rethink your decision and allow at least for the local network farms to run the machines at 100% of their capacity. At the moment the machines do not exceed 50% CPU and 25-30% Memory even when all agents are busy.

Many thanks for your understanding and best regards
Peter

Hey Guy, is this problem solve?  I have two PCs in my local network now and face the same problem...

 
Xie Jiansi:

Hey Guy, is this problem solve?  I have two PCs in my local network now and face the same problem...

read post
can be optimized on local agents only, no Cloud, no Farm
can be optimized on local agents only, no Cloud, no Farm
  • 2021.07.09
  • www.mql5.com
Hello there, I have this message in my system. 2021.07.08 21:53:14.943 Tester Cloud servers switched off 2021.07.09 03:08:10...
Reason: