Automated Trading Championship 2011 - Rules - page 12
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
http://ark.intel.com/Product.aspx?id=37100
Yedelkin
I don't quite understand your question. The number of cores has no direct effect on the speed of a single EA test (executed in one thread).
Or your interest was not related to speed of testing?
Yedelkin
I don't quite understand your question. The number of cores has no direct impact on speed of test of one Expert Advisor (executed in one thread).
Or your interest was not related to the speed of testing?
It so happens that my Expert Advisor runs with several "program execution threads" (I don't know how to call it correctly, but it is when n indicators on different symbols plus one Expert Advisor on the main symbol). I noticed that the optimization is twice as slow (very roughly speaking) as a single test. I've looked at my Atom's dual core load. It turned out that during single testing both cores were loaded "to the max". Concluded that the number of cores directly affects the testing speed of my EA.
Once upon a time it was already discussed that the number of cores and number of threads shouldn't be confused - I didn't bother myself with it (to delve into details), but set a task to design a program that would induce as many "execution threads" as possible. As a result, the task was successful :)
It so happens that my Expert Advisor runs with several "program execution threads" (I don't know how to call it correctly, but it is when n indicators on different symbols plus one Expert Advisor on the main symbol). I noticed that the optimization is twice as slow (very roughly speaking) as a single test. I've looked at my Atom's dual core load. It turned out that during single testing both cores were loaded "to the max". Concluded that the number of cores directly affects the testing speed of my EA.
Once upon a time it was already discussed that the number of cores and number of threads shouldn't be confused - I didn't bother myself with it (to delve into details), I just set a task to design a program that would induce as many "execution threads" as possible. As a result it turned out that I didn't try for nothing :)
It's hard to believe...
And what does the "Agents" tab show during the test? I have a feeling that your Expert Advisor has just loaded one core "to the max" and the second core has taken care of all the other OS tasks, otherwise you would have hardly seen anything but the testing window.
And you are wrong about the kernels and threads - you should have at least a general idea
It's hard to believe...
And what does the "Agents" tab show during testing? I have a feeling that you have just loaded one core "to the brim" with your EA and the second core has taken over all other OS tasks, otherwise you would hardly see anything but the testing window.
Of course, one local agent is running. Your version is quite possible, ...only it is not clear which tasks can load the second core in such a case.
I tried to explain the reason for my question. It is quite possible that my question was formulated on the basis of wrong conclusions.
And you should not think of kernels and threads since you should have at least a general idea of what they are.
Yes, I understand :( But I physically have no time to cover everything. I have a lot of plans.
Of course, one local agent is running. Your version is quite possible, ...but it is not clear which tasks are loading the second core.
I tried to explain the reason for my question. It is quite possible that the question was formulated on the basis of wrong conclusions.
Yes, I understand :( But I have not time to cover everything. There are a lot of plans.
There are some technologies in some operating systems and processors that allow you to redirect a part of load to free cores when there is a large load on one core.
Maybe this is the case.
But as I understood from the discussion of some topics and articles, there may indeed be several threads in which the inductors are counted when using the inductors on different pairs.
Yedelkin
In this article you will find information on threads.
Among other things, it will make it clear that the different pair indices are divided into different threads. But. During testing, the entire "collective" (EA+indulators) is run on one thread (agent).
Therefore if you try to pass the 15-minute testing requirement, switch off all visualization + perhaps some of the indulators. One CPU/core will normally be enough for the real-time calculation.
IMHO.
I cannot understand the championship rules.
With their rules, the Metacquotes are misleading all traders.
It's clear that the traders don't care. Metacquotes want to demonstrate their new work.
Who are the Metacquotes targeting?
Why all these rules?
The beginning (gist) - who has more money is right. Or are the Metakwots against it? I.e. no contrived rules.
It turns out that the Metakwots are demonstrating their inferiority. Going outside the rules does not guarantee profits for the sponsors?
I would like to revisit some points in the rules, namely:
IV. Trading Conditions
4.The maximum number of pending orders on all symbols at any one time is 12.
...
As long as the total volume does not exceed 60 lots (5.00 * 12 = 60.00), there is no problem. These points are executed and do not limit the work of the Expert Advisor. But when the total volume of exposed lots exceeds 60.00 lots, this is when the mentioned clauses of the Rules begin to change the strategy. On one hand, I can place only one pending order per symbol (point 4), and on the other hand, I cannot place an order with the volume exceeding 5.00 lots (point 3). This is a problem to solve. Of course, this situation has some solutions.
It does not follow from point 4 that we are only allowed to have 1 active order per symbol. If you follow only the rules, then at any given moment 12 orders may be used, and the way they are dealt with is an individual matter (depending on the strategy).
2. If we are only following the rules, a monovaluator can also set 12 orders (this can greatly simplify our life during the implementation of certain strategies and at the same time greatly complicate the algorithm).
I am asking about this point for the second year in a row, and at the same time, I suggest increasing the number of orders to 24-36. I have not yet received any official comments explaining everything and putting everything in its place.
I assume that the organizers will not increase the number of orders (sorry, and I hope I am wrong) referring to the load on the equipment on which the championship is held.
However, I cannot understand how increasing the number of orders (even up to 36) can overload resources (unless you set/delete orders on every tick of course).
Is it better to constantly perform market operations (frequency may be even higher than for setting orders) and risk getting unwanted results in case we lose connection with the server?
This is a question for the organisers and developers...
papaklass:
I see three ways out of this situation:
1) To leave item 4 unchanged and remove the 5.00 lot limit in item 3 of the Rules;
2) leave the 5.00 lot limit in item 3 unchanged, but increase the number of maximum orders to 36 (3 orders per symbol. 5.00 * 3 = 15.00) in item 4;
3. For multicurrency Expert Advisors modify point 5 by decreasing total volume from 15.00 lots to 5.00 lots. However these changes could cause server configuration problems.
4. I see the only possible option, would be to increase the number of orders to 24-36 (or alternatively 50. Since 50 orders/positions is the limit existing at the moment in many brokerage companies).
But since only 12 currency pairs are traded during the Championship, it would be reasonable to increase the number of orders from 12 to 24-36 (if not during this Championship, then during the next one).
5. Point 3 I find unacceptable, in any form or in any sauce.
3. For multi-currency EAs, amend item 5 by reducing the aggregate volume from 15.00 lots to 5.00 lots.