What should be added for additional support of universal mathematical calculations in MQL5 and MQL5 Cloud Network? - page 4

 
By the way, tester agents get through proxies and nats normally. When working in MQL5 Cloud Network, they do not need to have open ports.
Распределенные вычисления в сети MQL5 Cloud Network
Распределенные вычисления в сети MQL5 Cloud Network
  • cloud.mql5.com
Заработать деньги, продавая мощности своего компьютера для сети распределенных вычислений MQL5 Cloud Network
 
Renat:
By the way, tester agents pass through proxies and nats normally. They do not need open ports when working in MQL5 Cloud Network.

I opened ports in NAT, I'm waiting for traffic and no clients (I'm laughing!!!)


By the way, do you want to know how much traffic is wasted or is it also packaged both ways to send and to receive?

 
ruslanchik:

I opened ports in NAT, I am waiting for traffic, no clients (I am joking!!!)


by the way ask what is a lot of traffic or is it also packed in both directions on sending and on receiving

Traffic is shown for each agent in the manager, as well as the number of performed tasks.

All traffic is necessarily compressed.

 
MORE Question... how does the bit depth of the operating system affect computing performance?
 
ruslanchik:
ANOTHER question... how does the bit depth of the operating system affect the computing performance

Straight.

In general, 32-bit is no longer relevant in today's environment.

Case in point -- I need at least 12 gigs of RAM to work comfortably on a project (at work).

 
TheXpert:

Straight.

In general, 32-bit is no longer relevant in today's environment.

Case in point - I need at least 12 gigs of RAM to work comfortably on a project (at work).

I cut all the services I didn't need and used the performance counter to check the CPU load ... and some of the RAM. 4 gigabytes is good enough for me to work with
 
ruslanchik:
I did so, cut off unnecessary services and unloaded the CPU ... and some of the RAM

OK :) 10 studios. There are 700(!) and a penny more projects in the solus. Studio alone eats up a gig or more.

Rebuild on linking will eat up a total of 11.5 gigs. If RAM is less, broom starts eating and linking time increases several times. The less RAM, the more.

Services will not help here.

 
TheXpert:

OK :) 10 studios. There are 700(!) and a penny more projects in the solus. Studio alone eats up a gig or more.

Rebuild on linking will eat up a total of 11.5 gigs. If RAM is less, broom starts eating and linking time increases several times. The less RAM, the more.

Services will not help here.

Wow, I didn't realize how serious this was...
 
Renat:

Hence the question - what other functions need to be included to improve the capacity of the calculation network?

I have now read the strategy tester documentation and realised that the mathematical calculations mode does not allow loading data from outside. Say, some array of data such as prices of some instrument or DNA sequence. Without this feature makes no sense in mathematical calculations, because there is no actual calculation subject. I am not even talking about the possibility of loading such a feature, as a custom tool.
 
C-4:
Now I've read the strategy tester documentation, and realized that the math mode doesn't allow to load data from outside. Say, a certain data set like prices of some instrument or DNA sequence. Without this feature makes no sense in mathematical calculations, because there is no actual calculation subject. I am not even talking about the possibility of loading such features as a custom tool.

The following options are available for data transfer (described at the beginning of the topic):

  • Transferring source data via the #property tester_file "my_file.dat" attachment, which allows source data of any size to be processed and results to be returned.
  • Transmission of source data via #resource attachments.
  • Passing source data via named pipes only when working with local and remote agents

This is exactly what we are thinking of extending the data exchange functions:

For our part, we are thinking about independent communication between agents without a terminal. For example, one of the agents can generate initial data and send it to the others by fordcast.

Reason: