performance issue with the Strategy Tester on build 5809

 
Hi everyone,

I'm experiencing a major performance issue with the Strategy Tester on build 5809 and I'm not sure if this is a recent regression or a known limitation.

The issue: backtesting is extremely slow. To rule out any issue with my own EA, I tested with the native built-in Moving Average example EA (Moving Average.ex5) — the slowdown is identical. This confirms the problem is not related to my code but to the Strategy Tester engine itself.

My setup:
- Symbol: XAUUSD
- Timeframe: M3
- Modelling: OHLC on M1
- Period: 1 month
- OS: Windows 10
- CPU: Intel Core i7-7500U

Expected: 1 month of M3 data should process in a few seconds.
Actual: same period takes several hours.
 
elgorab:

next time do a search of the forum please.

This issue is common to happen after an update, both windows updates and Mt4/5 LiveUpdate.

It is recommended to do several things. 1 or more may be a fix. In my experience 1 of them is always a fix.

1. clean the tester folder. ie completely delete the Tester folder.

2. defragment the history and tic folders by copying these folders to another drive, and delete the original folders. And then copy the copied folders, to their original location. This is not the same as doing a full 100% defrag, however, it is at least 80% and in my experience fixes mt5 bugs most often than the other fixes in this list.

3. your version of mt5 is a beta version. Always install a "release version"; and only get it from a broker. not from this website as it is almost always a beta. Release versions tend to be much less bugs; while betas are almost guaranteed to be bugged. You can search this site for the latest release files to download via a link to a git page.

I am sure there are other possible ways to fix your issue, but you will have to do a search of this site to get more.

 
elgorab:
7500U

why do you say it is a performance issue? that cpu is 10 years old. Have you done the same backtests on that cpu+computer before?

If i remember back then, those cpus were slow, but good in laptops or handhelds because they ran with very little heat expelled. This alone means that it would be very slow and not expected to run such a load as strategy tester. Even a m1 OHLC can be intensive on modern cpus.

 
Michael Charles Schefe #:

why do you say it is a performance issue? that cpu is 10 years old. Have you done the same backtests on that cpu+computer before?

If i remember back then, those cpus were slow, but good in laptops or handhelds because they ran with very little heat expelled. This alone means that it would be very slow and not expected to run such a load as strategy tester. Even a m1 OHLC can be intensive on modern cpus.

I have an "old-ish" i7 gaming tower pc that runs M1 OHLC backtests lightning fast. Even lengthy real tick backtests only take a few hours. It has 3 intake fans, 1 internal fan, and 1 exhaust fan with multiple serviceable vents. It also supports virtual cores which I don't even use.

Therefore, yeah, it's gotta be the 🔥.

 
Ryan L Johnson #:

I have an "old-ish" i7 gaming tower pc that runs M1 OHLC backtests lightning fast. Even lengthy real tick backtests only take a few hours. It has 3 intake fans, 1 internal fan, and 1 exhaust fan with multiple serviceable vents. It also supports virtual cores which I don't even use.

Therefore, yeah, it's gotta be the 🔥.

i have 6 in my FX-8320
 
Michael Charles Schefe #:
i have 6 in my FX-8320

That's a good double confirmation.

Truth be told, I manually clean out my main Tester and my MQL5==>Profiles==>Tester folders after every backtest. All valuable backtest Reports are stored on another computer.