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)
My first post, and my first problem with NT. I'm designing a parallel optimizer, I mean, run each of the iterations on diferent machines/procesors. I have the optimizer, the messaging system, and the problem. The idea is running empty iterations on the master NT, which are actually executed on a slave NT, and when the iteration is finished, replace the result in the master NT from the slave NT. And then the problem arises. Once I recover the Strategy object from the slave NT, how can I "insert" those results in the master NT? Is there any point in the code of an optimizer where I can change the strategy results for one iteration? (With change I actually mean insert)
There are other "dirties" posibilities available for exhaustive optimization, like saving all the results, from master and slaves, in the same file, or export the result of the optimization on every NT and merge them externally, etc. But the first option is the best, i f possible, because it allows to use any optimizer algorithm.
Greetings and thanks,
Juanjo
Can you help answer these questions from other members on NexusFi?
Actually I think it is the most ambitious undertaking I have heard of someone implementing with NinjaTrader. Would be great to see working, especially on a multiple virtual machine environment. I wonder if the staff at NT support would say more than 'unsupported' if they were asked for assist in getting it working.
IMHO, in retrospect, it is actually sad that someone is even thinking along these lines: it shows how much Windows/NT is lacking in such areas (distributed computing has been around for some time).
Well, if he is able to accomplish this the benefits are limitless. One possibility would be to also implement CUDA for the engine, which would increase optimization performance probably 10x over.
Yeahhh ... I would LIKE that! Or the ATI version of GPU ... I have a 2GB 4670 ATI ready for beta testing .. not sure about that 10x ... certainly would need lots of tweaking and NT developer support .. maybe even a GPU version of NT for full GPU-power.
Somehow I dont think just building a class or method in C# will be enough but I could be wrong .... research on google and looks like there is hope .. maybe ?
At one time I was involved in a team that was first line support for some of the first HPC cluster servers 'way back when'. Glad I wasnt on the software side of the team ...I would have been fired the first day .. tools are a lot better now, and Win7 was built with parallelism in mind, but not totally installed ... the evil empire keeps us waiting again
There is some work going on using unsupported code to access more of the optimization parameters. Perhaps your collabaration with those involved might be useful for all of you.
but I could be wrong .... research on google and looks like there is hope .. maybe ?
...I would have been fired the first day 
.. tools are a lot better now, and Win7 was built with parallelism in mind, but not totally installed ... the evil empire keeps us waiting again