A little surprised :) Thought I'd share and ask a NOT rhetorical question. - page 11

 
Academic:
Well, let's make fun of MT5 - here's the C++ code and the result :)

Waiting for at least 400 million cycles per second. :) But here you have 1,282 MILLION cycles per second. :))

It turns out that the calculator was written not in MQL5 but in CPP. Is it so or not?
 
Urain:
So, it turns out that the calculator was written not in MQL5, but in CPP. Is it or isn't it?
Yeah. :) In C++. What's the problem? We've known that from the beginning. :)
 

The work on the MQL5 optimizer is in progress, the speed of the MQL5 code will still increase.

Here is a comparison of the performance of this code on i7

С++

0 = 0.000000, it = 1.87e-009 (sec), 534.759358 mio per sec


MQL5

2011.03.31 16:31:55    111 (EURUSD,H1)    0 = 0.000000, it = 4.36e-009 (sec), 229.357798 mio per sec


I.e. MQL5 is 2-3 times slower than C++.
 
Academic:
Yeah. :) In C++ What's the problem? It's been known all along. :)
Then run your DLL compiled in C++ on a 32-bit system on a 64-bit machine and then you'll understand the advantage of an ex5 file which doesn't care where it was compiled.
 
mql5:

i.e. MQL5 is 2-3 times slower than C++ today.

What a great result! Just a few tester agents, and the advantage will be on the side of MQL5 for the case of UNIVERSAL optimizer (not tester).
 
Urain:
Then run your DLL compiled in C++ on a 32bit system on a 64bit machine and then you'll see the advantage of an ex5 file which doesn't care where it was compiled.
:) What for?
 
mql5:

The work on the MQL5 optimizer is in progress, the speed of the MQL5 code will still increase.

Here is a comparison of the performance of this code on i7

С++

0 = 0.000000, it = 1.87e-009 (sec), 534.759358 mio per sec


MQL5

2011.03.31 16:31:55    111 (EURUSD,H1)    0 = 0.000000, it = 4.36e-009 (sec), 229.357798 mio per sec


I.e. MQL5 is 2-3 times slower than C++.

You forgot to enable OPTIMIZATION in C++. :) And you compiled it on x86 instead of x64 for some reason. Not to mention the fact that it's 1600 mio on an intellect.

And why?

 
hrenfx:


All you have to do is compare the speed of the UNIVERSAL MT4 and MT5 testers using the methodology described above.

I'm sorry, why ? Or who needs it ?
 
hrenfx:

Once again I ask the participants of this thread not to reduce it to crap. You would also say that JForex API is better because it is written in cross-platform Java. What the fuck is all this crap about!

All you have to do is compare the speed of the UNIVERSAL MT4 and MT5 testers using the above methodology.

What kind of shit is this? Ideology of universality of compiled files provided an opportunity to distribute safe closed code regardless of system bitness. Which is very urgent now because there is still a large fleet of 32-bit machines but 64-bit ones are being implemented at a full speed.

Well naturally Net is slower than pure C++, but the optimization (which the developers are working on) reduces that difference by a lot.

A year ago, I was comparing the performance and MQL5 was 6 times slower, while now it is 2-3 times slower. The progress is evident.

 
Mischek:
I'm sorry, why? Or who needs it?

Because the MT5 optimiser cannot really be used, in my opinion. Then why does it draw 3D charts at all? Can't I just copy three columns of numbers into Matlab? And get the same 3D in one movement?


In a word, if a car is all that beautiful but does not drive, then it is furniture.

Reason: