Welcome to NexusFi: the best trading community on the planet, with over 150,000 members Sign Up Now for Free
Genuine reviews from real traders, not fake reviews from stealth vendors
Quality education from leading professional traders
We are a friendly, helpful, and positive community
We do not tolerate rude behavior, trolling, or vendors advertising in posts
We are here to help, just let us know what you need
You'll need to register in order to view the content of the threads and start contributing to our community. It's free for basic access, or support us by becoming an Elite Member -- see if you qualify for a discount below.
-- Big Mike, Site Administrator
(If you already have an account, login at the top of the page)
Since NT is only single core and can not process optimization data quickly I am wondering if any other members have suggestions into what other platforms / resources may be used to say, run 20 NT optimizations at once.
Thanks,
Resirca
Can you help answer these questions from other members on NexusFi?
Thanks for the reply; however I was asking the question based on a response from NT support. See copied tesxt below :
Hello,
Thank you for contacting us.
It is true that NinjaTrader 7 would use a single core for nearly all processes with the exception of backtesting and system optimization. Because of that, additional CPU cores would not be of additional benifit to NinjaTrader 7.
Sincerely,
Kyle
NinjaTrader Customer Service
I was looking to speed processing time for an optimization from multiple hours to minutes. Below are my PC specs :
Thanks for the clarification and just wanted to point out what Kyle from customer service stated:
"NinjaTrader 7 would use a single core for nearly all processes with the exception of backtesting and system optimization"
His point is that it is multi-core (the exception) for backtesting/optimization. I will make sure that my team explicitly states multi-core so that there is no confusion. Thanks for pointing that out.
Thanks for the clarification; in lieu of that would it be possible then to utilize a more powerful PC / server with say 16 or 32 cores to process multiple optimizations rapidly. Our goal here is to be able to run multiple optimizations in a matter of minutes and not days.
We are currently running an optimization with our strategy now, which has two optimizable parameters ( one is the data series 1- 60 in 1 increment steps ) and the other is a custom filter. We are using NT's default optimizer for a two week period, a custom optimizer type and the optimizer completion guage shows approx another 8 - 15 hrs before completion. It has currently been running 7hrs 40min.
As I stated in my inital enquiry to NT support, we do not want to purchase or lease any additional server processing resources if it will make no noticeable improvement. The information we obatined was that it made no difference how much processing power we had, the optimization process was dependant upon NT's processing power...which could not handle these sorts of demands. That being said, I can provide the strategy, the optimization type and the relevant parameters if it would help provide a solid answer here.
Is anyone running NinjaTrader with more than 16 cores?
For example, a dual or quad Opteron 62xx (Interlagos) box?
Mike
Short version, NT7 won't benefit from a "bigger" computer once you get past say 4 to 8 real cores. You'll need to wait for NT8. That was my experience at least. I've got a Dual Xeon E5-2600 class with 128GB, there isn't much benefit in running NT7 on that machine over a desktop i7. But hopefully with NT8 that will change.
However, unfortunately most people overlook the fact that object-oriented code architecture in general is a prime L1, L2, L3 memory cache destroyer so, regardless of the number of cores applied, performance may not scale in the ways expected, under NT7 or NT8.
In addition C# was primarily developed to benefit correctness of code before performance, I prefer it that way, but, having written a lot of high performance code in a lot of language, scheduler and machine environments, I'm under no illusions about what to expect.
Staying in the .NET Domain, I've found F Sharp to be the answer. A true OOPL, it is optimized for performance, processing and handling of large datasets and algorithmic calculations and I'm rewriting my custom application in F Sharp and am considering doing my AI in F Sharp also.