cAlgo back testing performance

Created at 16 Jan 2017, 13:18
How’s your experience with the cTrader Platform?
Your feedback is crucial to cTrader's development. Please take a few seconds to share your opinion and help us improve your trading experience. Thanks!
lec0456's avatar

lec0456

Joined 14.11.2012

cAlgo back testing performance
16 Jan 2017, 13:18


I am running cAlgo on an i7 3.4ghz CPU with 16gb of memory but backtesting is still slow.

What can i do to speed up and improve performance. 

I tried running it in compatability mode for Windows 8 & 7, but no differences.


@lec0456
Replies

FMogyi
16 Jan 2017, 17:29

What do you mean it slow? Backtesting is a very slow process. But much faster then the forward tests.


@FMogyi

lec0456
16 Jan 2017, 17:55

I mean that it starts taking up over 1.5 gigs of memory and slows to a crawl.  The CPU's don't seem to be fully utilized.

So, I am asking the engineers at spotware if they have any performance tuning tips or advice on how to fix this.


@lec0456

lec0456
17 Jan 2017, 08:40

It looks like a memory bleed.  The program is taking 2Gigs at this point


@lec0456

ColossusFX
21 Jan 2017, 18:29

It depends on what you are backtesting...

If you are backtesting an algo with about 1000 indicators that calculate on tick, then you will suffer from the consequences.

With an i7, you should be flying through 4 years of tick data backtesting.

I am running full tick data backtests in a few minutes on i7.


@ColossusFX

lec0456
21 Jan 2017, 19:18

I specifically bought this PC to get better performance after running my Algo on a celeron 1.6ghz.  This PC is a Z170 chipset, I7, 3.4ghz, 16GB ddr4, with aan M.2 Solid State Drive, transfer rate is 32gbs.

I even selected cAlgo in the Task manager and set its priority to high.  

Yes, I am using tickdata.  But it is really slow.  And the memory usage is higher than any program I have ever used.  2.5gigs and the processors are  not maxing out, ever.  so, it doesn't seem like CAlgo is taking advantage of the resources on the system anyway.

 


@lec0456