Anyone have a sample program w the nxcore data feed... Will pay $$$

[tradelink]

def the biggest obstacle in terms of best pricing and fast execution is in the network.

nyc co-location would eliminate this for equities and other nyc-concetrated markets.

 

[jaguar1637]

Quote from NetTecture:

I can offer you C++/CLI code reading the NxCore feed and caling back into .NET interfaces. I went with C++/CLI because the standard wrapper approach is slow compared to that - no way to optimize for example string generation.

Needs some bawe classes for my own trading system but nothing too deep (interfaces, message classes etc.).

Hi
Is there a place to fetch such code for reading the NwCode Feed ?
rgds

 

[vicirek]

Quote from NetTecture:

Just as anote - a connector like this is the reason I went with C++/CLI. TradeLink is quite inefficient in this area. C# can not handle this interaction with good performance. it is fine for a life ffeed, but quit a difference loading historical data.

I have question regarding C# vs C++/CLI; both are using the same virtual machine so there should not be any difference in performance.

 

[Occam]

Quote from jaguar1637:

Hi
Is there a place to fetch such code for reading the NwCode Feed ?
rgds

Is this what you mean (assuming you want NxCore rather than "NwCode")? They've posted some examples here, although I'm not sure it's what you're looking for:

http://nxcoreapi.com/doc/

 

[sprstpd]

Quote from vicirek:

I have question regarding C# vs C++/CLI; both are using the same virtual machine so there should not be any difference in performance.

A few years ago I tried out the Nxcore trial and found that reading the quote stream from C# (using the Nxcore supplied code) was almost an order of magnitude slower than doing the same in C/C++.

 

[NetTecture]

Quote from vicirek:

I have question regarding C# vs C++/CLI; both are using the same virtual machine so there should not be any difference in performance.

Except the C++/CLI code gets compiled into native assembler code in methods that can do that if you do not block it. You can pretty much decide on a method by method basis whether or not a method is managed (bytecode) or native (assembler).

It also is a lot easier / faster to work with native elements like strings as you can avoid a lot of "stupid" marshalling to .NET - I can basically pregenerate .NET strings (like symbols) and reuse them - something not possible with a standard "method call per method call" wrapper.

There is a TON of low level optimization that C++/CLI can do, given that all data structures in NxCore are C++.

SOME of that COULD be done by working on a (VERY COMPLEX) C# stack, but that is more work than just doing it in C++

 

[vicirek]

Thank you NetTecture, I will check it out.