Multi-core testing network for all comers - 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
But the question of discrimination is interesting, faster cores should also cost more, again you need to judge by pings and communication stability.
Although I think the price will be so small that neither the landlord nor the tenant will know the difference.
I've already written somewhere that I tried agents on a dual core Atom. so the PR rating of one core there =40. The PR rating of my Q6600 core =80 and the actual difference in time of one run turned out to be about 6 times, not 2 times as it would seem from the rating.
So if I buy cores based on PR rating and see such results, I will quickly ask for my money back!
I already wrote somewhere that I tried agents on a dual core Atom. so the PR rating of one core there =40. The PR rating of my Q6600 core =80 and the actual difference in time per run was about 6 times, not 2 times as it would seem from the rating.
so if i buy cores based on PR rating and see such results i will quickly ask for my money back!
You don't forget that things will change on remote agents (although it depends on the frequency of queries).
You may have a fast agent but a slow ping.
We already have more than 10 metatester agents in our network.
Join in!!!
1. Install Comodo EasyVPN http://easy-vpn.comodo.com/download.html2. Join a network of agents: Networks, Join a network, Network name: metatester_agents, Password:1234567890
Don't forget that on remote agents it will change (although it depends on the frequency of requests).
You may have a fast agent but a slow ping.
It all depends on how long one run is counted. If it counts for at least 10 seconds, ping is almost irrelevant.
History is downloaded once by the agent. then only Expert Advisor parameters are transferred for each run.
It all depends on how long one run is counted. If it counts for at least 10 seconds, the ping is almost irrelevant.
History is downloaded once by the agent. then only Expert Advisor parameters are transferred for each run.
Don't forget that after some time machine thinks that there will be no answer and then, for example, browser pops up a message that time to answer is over, so ping also matters. Your agent's responses may simply not reach the recipient (not all the time, but regularly).
I decided to play around with CamStudio. The first time, it turned out to be a small promotional video ;)
Oops. The avi file didn't load. Apparently, the How to Publish Video in MQL4.community article is outdated and avi files are not uploaded anymore. I will try to convert it to flv. It did not work either. I have a different way then.
Performance boost when adding metatester agents.
I uploaded a good quality avi file, but the output quality is very poor.
I decided to play around with CamStudio. The first time, it turned out to be a little commercial ;)
Oops. The avi file is not uploaded. Apparently, the How to Publish Video in MQL4.community article is outdated and avi files are not uploaded anymore. I will try to convert it to flv. It did not work either. The other way then.
No more than 5 times in a row.
But, after a successful start (a successful start is one that returns a result) the failed start counter is reset to 0.
Are there plans to write a manager for the cluster ,
The counter is reset to 0 after a successful run (the one that returns a result).
We already have more than 10 metatester agents in our network.
Join in!!!
1. Install Comodo EasyVPN http://easy-vpn.comodo.com/download.html2. Login to agent network: Networks, Join a network, Network name: Metatester_agents, Password:1234567890
There is a limit of 16 machines in this software ?
I have 28 cores now... on 9 machines so far
I can probably find hundreds or more cores if I solve the problem with the grey IPs
--
Who has a job in an organisation with a lot of servers and has the ability to administer
network at work, can cluster as many cores as the organization has.
Are there plans to write a manager for the cluster that distributes resources among the participants .
The MQL5.community Pool is already being written, which will allow some to easily sell resources to their testers and others to buy them.
With one button and some money it will be possible to buy 500-1000 agents. Testing of strategies will be done in a few minutes.