Errors, bugs, questions - page 1489

[Mikhail Tkachev]

Artyom Trishkin:
But the Kovalev textbook does work. What prevents you from copying it correctly and adjusting it to the new requirements?

Artem, what is Kovalev's textbook?

[Artyom Trishkin]

Mike:
Artem, what is the Kovalev textbook ?

The place where the code was taken from in the post I was responding to. On mql4.com the tutorial.

[Mikhail Tkachev]

Artyom Trishkin:
The place where the code was taken from in the post I was responding to. On mql4.com tutorial.

I have it locally ... turns out :)

[Forester]

Renat Fatkhullin:

Yes, in fact we have disabled FileFlush cache resetting, as almost no one knows how to use it, it kills disk and our system performance.

We have implemented very efficient buffering and decide when to reset the caches ourselves. The goal is performance and realisation of the platform's own objectives.

If you really want to dump data to disk, then close the file.

You should put it in the documentation so people don't waste their time... because there's not a word about it there.

[Maxim Khrolenko]

I'm thinking of making a RAM disk to install MT4/MT5 for optimisation. If anyone has done this, can you share your experience - does it increase the speed of optimisation?

[Renat Akhtyamov]

Maxim Khrolenko:
I'm thinking of making a RAM disk to install MT4/MT5 for optimisation. If anyone has done it, can you share your experience - does it increase optimization speed?

The fastest way is to install metac and windup on SSD disk

[Vladislav Andruschenko]

new-rena:
the fastest way is to install metac and winD on an SSD

i have a friend who argued for a long time that sd is ineffective and a waste of money - i gave it a try 3-4 years ago - now i don't know how to keep the wind without sd?

i have an sd itself and winnd and terminals on it - i need to compare the speed of the ndd and the sd. hmmm, interesting

[Maxim Khrolenko]

During optimization/testing, MT5 uses historical quote data stored on disk. So I wonder if this data (1) is constantly read from disk and it is a bottleneck for optimization/testing or (2) it is all uploaded to RAM and speed depends on CPU. If 2 is true, then there is no point in optimising/testing on RAM disk.

[Renat Akhtyamov]

Vladislav Andruschenko:

i have a friend who argued for a long time that sd is ineffective and a waste of money - i tried it 3-4 years ago - now i don't know how to keep the windup without sd?

i myself have an sd and winD and terminals on it - i need to compare the speed of the ndd and the sd. hmm, interesting

of course it's faster.

when choosing a SSD, pay attention to the number indicating write speed, the higher it is, the faster it will work.

The speed gain can be estimated by the ratio of SSD and HDD write speeds

when using an SSD, it is advisable to disable the virtual memory located on the HDD in the Windows settings

[Vladislav Andruschenko]

in general the results are ambiguous, the test is rough - a quick one..:

the script simply displays the time and price on the tester:

//+------------------------------------------------------------------+
//|                                                      ProjectName |
//|                                      Copyright 2012, CompanyName |
//|                                       http://www.companyname.net |
//+------------------------------------------------------------------+
int timestart=0; 
int ticknumber=0; // номер тика 
double PriceBID=0;// Цена тика БИД 
double PriceASK=0; // Цена тика АСК 
int TimemsTick=0; // Время тика в мс 
//+------------------------------------------------------------------+
//|                                                                  |
//+------------------------------------------------------------------+
int OnInit()
  {
   datetime Time[];
   ArraySetAsSeries(Time,true);
   CopyTime(_Symbol,_Period,0,1,Time);
   timestart=(int)Time[0]; // начнем отсчет с нового бара, чтобы все советники на всех терминалах стартовали одинаково.!  
   return(INIT_SUCCEEDED);
  }
void OnTick()
  {
   datetime Time[];
   ArraySetAsSeries(Time,true);
   CopyTime(_Symbol,_Period,0,1,Time);
   if(timestart!=(int)Time[0])
     {
      if(TimemsTick==0)TimemsTick=(int)GetTickCount();
      ticknumber++;
      MqlTick last_tick;
      SymbolInfoTick(_Symbol,last_tick);
      double Bid=last_tick.bid;

      SymbolInfoTick(_Symbol,last_tick);
      double Ask=last_tick.ask;
      PriceBID=Bid;
      PriceASK=Ask;
      Comment("\n\n\n\n Номер тика="+ticknumber+"\nPriceBID="+PriceBID+"\nPriceASK="+PriceASK+"\nTimeTickinms="+(GetTickCount()-TimemsTick));
      Print("Номер тика="+ticknumber+" PriceBID="+PriceBID+" PriceASK="+PriceASK+" TimeTickinms="+(GetTickCount()-TimemsTick));
     }
   return;
  } 
//+------------------------------------------------------------------+

two terminals :

so we run it VERY ONE TIME:

and immediately the test time counts down

After a couple of minutes, the SSD takes the lead:

After another minute the score is almost even, the HDD takes the lead:

let's move on to optimisation:

I start the optimisation on the teak, for example:

HDD is faster

but the results are somehow different, sometimes SSD is faster on a new run on a different parameter, sometimes a regular VIT.

But here is the fastest : AGENTS :-) so a couple of quid for a quick optimizer run is worth overpaying for: