Backtest consuming too much memory until crashes

New comment
 

Hello,

I want to know if anyone else is having this problem. Since last update on mt5, my bot is running into problem when backtest in real ticks. 

When I start the backtest it goes normal, the cpu goes to 100% while the memory is loading each agent, normally it would be 100% CPU and about 40% of memory ram in use, now the memory goes straight to 100% until bottlenecks the CPU, then the CPU starts to go down to around 20% to 50%. 
Before the update 1 test took about 30 to 40 minuites to be completed now it shows almost 3h and the computer start to freeze and crashed once.

I have a Lenovo P620 with a AMD Threadripper 3995wx pro (64cores) with 256gb of memory. I read in some really old posts people having similar problem with AMD processors, I wonder if this problem is back.

Someone else with a AMD processor having the same issue?

Thank you

 

Images with the computer spec and with a example of the problem

 
Hello
Nice pictures <indeed>
I work with 4 GB of memory and a 2 core processor
Everything else I see is the same

Optimization is always insatiable
But I have no crashes
I want to make special test code now <then will see by Will of God>


 
I tested with several bots and Im get the same results, too much memory per core. Its like 3 to 4 times the Normal amount. I hope that will be fixed fast. I wrote to service desk too.
 
NathanYsp:

Hello,

I want to know if anyone else is having this problem. Since last update on mt5, my bot is running into problem when backtest in real ticks. 

When I start the backtest it goes normal, the cpu goes to 100% while the memory is loading each agent, normally it would be 100% CPU and about 40% of memory ram in use, now the memory goes straight to 100% until bottlenecks the CPU, then the CPU starts to go down to around 20% to 50%. 
Before the update 1 test took about 30 to 40 minuites to be completed now it shows almost 3h and the computer start to freeze and crashed once.

I have a Lenovo P620 with a AMD Threadripper 3995wx pro (64cores) with 256gb of memory. I read in some really old posts people having similar problem with AMD processors, I wonder if this problem is back.

Someone else with a AMD processor having the same issue?

Thank you

Update: with the latest version 4260, the problem still happening. I just noticed that its happening on pair like EURCAD, USDJPY but not in other like EURUSD.  


If someone is having the same problem can you test that to confirm?

 
NathanYsp #:

Update: with the latest version 4260, the problem still happening. I just noticed that its happening on pair like EURCAD, USDJPY but not in other like EURUSD.  


If someone is having the same problem can you test that to confirm?

Hi there,

I am also having same problem, 40Core with 768GB, still not sufficient to run 10yrs of backtest data. Memory shoot up to max and Pagefile starts. then SSD 100%. Everything slowing down. Optimizer total 291 Billion combination.

I dont think the optimization count is an issues, but rather is the backtest history data of 10yrs. running 3yrs of data on 30M, no issues (1day to complete). Test on EUR/USD pairs.

Have you tried running shorter history like 3yrs?

Thanks

 
Norman Tan #:

I am also having same problem, 40Core with 768GB, still not sufficient to run 10yrs of backtest data.

Do a single run and include these lines here. 
250 total ticks for all symbols
AUDUSD: generate 50 ticks in 0:00:00.000, passed to tester 59 ticks
EURAUD: generate 50 ticks in 0:00:00.000, passed to tester 95 ticks
EURUSD: generate 50 ticks in 0:00:00.000, passed to tester 96 ticks
41 Mb memory used including 0.47 Mb of history data, 64 Mb of tick data
New comment