AMD or Intel as well as the memory brand - page 50
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
I'm afraid to disappoint you, but the performance results of this script are even less relevant than those of the first script. Most compilers calculate expressions like these at compile time (there's no point in evaluating what can be evaluated beforehand). I.e. code inside loops is equivalent to something like this: Int = 120 and Double = 120.0. That is, nothing useful is done at all. And if you replace the code inside loops with what I have written, the result will remain the same (check it for yourself). In fact, it measures the overhead of loop organization. The result in the second case is a bit smaller because entering Double (8 bytes) is a bit longer than entering Int (4 bytes).
And if in the first script array handling (the longest part) at least somehow reflected performance of memory subsystem (and bus with caches), this test in general measures "spherical horses in vacuum" and shows results not so much in parrots.
joo, I hope no offence. :)
Hmm, maybe, and most likely you're right. Well, in that case, either make intricate-cunning branches in the code, or feed the CPU with numbers it cannot "know" beforehand: quotes. Like that non-trading Expert Advisor https://forum.mql4.com/ru/25722/page41 :(
Changing the priority to real time in this test changed the rating by just over 1%.
The test is giving a steady 104.8311. It's probably raining. I think shutting down the internet with the antivirus is a bit odd. Easier to give real time priority to the terminal.
Normal antivirus runs as a service on the system, and not getting resources can hang the computer.
Hmm, maybe, and most likely you're right. Well, in that case, either make intricate-cunning branches in the code, or feed the CPU with numbers it cannot "know" beforehand: quotes. Like that non-trading Expert Advisor https://forum.mql4.com/ru/25722/page41 :(
Have you tried inserting Int = 120 and Double = 120? :)
I suggest the following solution.
Use the attached Expert Advisor to test the more or less real performance.
It is the same Moving Average that was used before, but it has excluded the function of lot calculation. It always opens positions at 0.1 lots.
Conduct the test using opening prices. The test is for the period from 15.09.08 to 15.09.09.
The balance is 1 000 000 dollars. For sure, there was no margin call at any leverage and margin requirements. For this the permanent lot is 0.1.
I increased the number of passes by decreasing the MovingShift step. This will allow to smooth occasional "throws" of load on processor, IMHO.
My result is 7:08 with insignificant results turned off, and 7:11 with them. Maybe it's not a big deal, but just the influence of the background load.
Let's check on this EA. The results of begemot61 and four2one are particularly interesting.
Docent, not everyone here understands the language. Please post the expert yourself, so there won't be any discrepancies.
OK, I'll post it myself. I just commented out the calculations within lot calculation function.
My result is 9:10 (useless results are displayed). Considering the frequencies ff comes out like this:
431*3 = 1293 (yours).
550*2.53 = 1391.5 (mine).
The difference is 7.6% in your favour.
I deleted the file. It's a little faster on yours - 9:05.
Docent, not everyone here understands the language. Please post the Expert Advisor itself, so that there is no confusion here either.
It's actually attached below along with the settings file in the archive...
Just to have 1 file, not 2.
Please check with mine - there might be some difference.
And wait for the results of the others.
>> Which applications exactly?
On a heavy computational task, originally multi-processor.
Data (old script) on Atom.
I suggest the following solution.
Use the attached Expert Advisor to check the more or less real performance.
This is the same Moving Average as before, but it has excluded the lot calculation function. It always opens positions at 0.1 lots.
The balance is $1,000,000. So that for sure with any leverage and margin requirements there is no margin call. For this the permanent lot is 0.1.
I increased the number of passes by reducing the MovingShift step. This will allow to smooth occasional "throws" of load on processor, IMHO.
I agree. The script was originally a "spherical horse". We need results that are closer to reality. Your changes in EA and initial settings seem reasonable.
My result is 7:08 with insignificant results turned off, and 7:11 with them. Maybe it's not that important, but just the effect of the background load.
I would leave full output on. The number will fluctuate depending on the conditions of the particular DC. All are not.
In general, you can completely eliminate the influence of DC - there is a program that allows you to set any spread. And we can exclude the influence of different histories, just include the history for the symbol in the archive with the test expert. Further it is clear: off-line terminal, spread setting, import of history from the archive and the test.
Regarding the article. Until the methodology is fine-tuned and statistics are collected, there is nothing to talk about. It would be possible to create something like a page on the net with a form to be filled out by the visitor. There is also an archive with methodology, advisor, history, etc. We'll have to think about that. It would be interesting to put it on the forum site itself. But this is beyond our capabilities.