Re: [Quickfix-developers] Performance issue: FIX::message_sorter

[Djalma R. d. S. F. <drs...@gm...>]

Hi Yuriy,

Very interesting!

I think that I got your idea.

I guess that if you use extern instead of static you can initialize
all shared pointers in a single CPP file.

I would appreciate to take a look in your code.

Regards,
Djalma

On 7/9/07, Yuriy Baranovskyy <Yur...@tr...> wrote:
> QuickFIX Documentation: http://www.quickfixengine.org/quickfix/doc/html/i=
ndex.html
> QuickFIX Support: http://www.quickfixengine.org/services.html
>
> Hi,
>
> I'm having a performance problem with my application, which uses QuickFix=
 1.12.4. After profiling I discovered that FIX::message_order constructor a=
nd copy operator both are very slow due to excessive heap allocation.
>
> FIX::message_order class has rather na=EFve implementation, which uses ar=
ray of ints for reverse lookup of the field ranks. This array is created/co=
pied each time the field is created/copied, which happens a lot during the =
message construction. The array size is typically few K each (max field num=
ber * sizeof(int)), so all these copies cost you dear.
>
> It's easy to implement light-weight copying just by using shared pointer =
to that array. (In my implementation I use either vector<int> or hash_map<i=
nt,int>, which are chosen in run-time depending on the predicted size).
> The copy constructor and assignment operator then only copy shared pointe=
r, not the array.
>
> To make construction light-weight is more tricky. Currently, FIX::message=
_order is created on the stack and passed to FIX::Group constructor, as def=
ined in message header files (src/C++/FIX44/*.h).
>
> Ideally, you'd need one instance of FIX::message_order class per field ty=
pe. The simplest way to do it is to initialise each FIX::Group with static =
instances of FIX::message_order. But this would require making .cpp file pe=
r each message header file, and adding them to the project/makefile.
>
> I can post solution for FIX::message_order here, but it is not clear how =
to solve the problem with cpp files. I played a bit with ruby code which ge=
nerated CPP sources, but adding 159 cpp files to the project seems to be qu=
ite a big change.
>
> Regards,
>
> Yuriy Baranovskyy
>
> -------------------------------------------------------------------------
> This SF.net email is sponsored by DB2 Express
> Download DB2 Express C - the FREE version of DB2 express and take
> control of your XML. No limits. Just data. Click to get it now.
> http://sourceforge.net/powerbar/db2/
> _______________________________________________
> Quickfix-developers mailing list
> Qui...@li...
> https://lists.sourceforge.net/lists/listinfo/quickfix-developers
>