MT5 and speed in action - page 84

 
Am I right in thinking that having EAs for 20 pairs would be the most profitable to run 20 terminals, 1 in each?
 
traveller00:
Do I understand correctly that having EAs for 20 pairs would be the most profitable to run 20 terminals with 1 in each?

Yes.

 
traveller00:
Do I understand correctly, that having EAs for 20 pairs will be the most profitable to run 20 terminals, 1 in each?

Queues are different, it's faster to get through actions from 20 terminals that are not heavily loaded than it is to queue in a single loaded terminal.

 

A comparison with the Server logs showed that OrderSend from Terminal reaches the Server without any delay. That is, if you need to grab liquidity quickly, you probably will.

But you can only find out in the Terminal whether you have grabbed it or not with a huge delay. In other words, there is a lag in the trading environment in the Terminal.

 
fxsaber:

On one account, I split all trading into five terminals. On the other, I left everything in one.


The logs show that splitting the account into terminals decreases the operation time of OnTradeTransaction.

I can also say for sure that the time OnTradeTransaction is influenced by the number of orders sent: the larger the number, the longer on average.


Generally speaking, everything seems to stay quiet.

We will check it out. But it's better to attach sample code right away. Judging by your log, you have delays when trading at around 1 ms, i.e. you either send an order at every tick or timer every 1 ms.

 
Anton:

We'll check it out. But it is better to attach a code example. Judging by your log, your delays start when you trade with a frequency of about 1 ms, i.e. either on every tick an order is sent, or on a timer every 1 ms.

Modifications can go frequently. I cannot attach a fighting Expert Advisor. Probably, we should write an autopooler.

 
fxsaber:

Modifications can go frequently. I can't attach a battle advisor. Probably need to write an autopuller.

Modify imho need as rarely as possible, I think that there is some sort of ranking on the number of requests to the server, and if a lot of "noise" on the server can be at the tail end.

 
Mikhail Mishanin:

modify as little as possible, I think there is some sort of ranking of the number of requests to the server, and if you make a lot of "noise" on the server, you may end up at the tail end.

There is no such thing.

 
fxsaber:

you need to write an auto-bullet.

In the trailer. Running it on a machine with near-zero pings causes GUI terminal to hang down, not responding to anything (via RDP), but rendering.

Apparently, unnecessary hundreds of FPS are affecting it. It's a bug.


However, running three copies simultaneously causes this to happen all at once.

DH      0       14:52:25.817    Trades  '88756': order #2229679 sell 0.01 / 0.01 EURUSD at market done in 31.402 ms (OrderSendTest_Position)
This is a demo with a ping of ~0.5 ms.
Files:
 

On a machine with zero ping I've never seen (with my eyes running through the log) the result of a trade order executed by Terminal (2/20 EAs running) faster than 1.5 ms. Very much ~2-3 ms.

What does this time consist of? Is it a peculiarity of OS?


The MQ-Demo server is now b2789. Is there anything changed in it in terms of speed compared to Server b2755?

Reason: