Thomas Hörmann <tho...@hoermi.com> wrote on 05/07/2013 05:21:37 AM:
>
> im using native X10 2.3.1
> I have the following code and would ask if you can help me for some
> performance tuning tips
>
Sorry for the late response.
Biggest single thing to get right for Native X10 is to invoke x10c++ with
-O. We pass this along to g++ and compile at -O2 instead of -O0.
Avoiding dynamic constraint checks can be important for Array performance
in 2.3.1, I'd compile with -VERBOSE_CHECKS and make sure that you get rid
of any reported dynamic checks.
You'll get slightly better array indexing code if the compiler knows
statically that the Array is really a Rail (dense, rank-1, zero-based. Ie
an array with elements from 0..(size-1)). If your Arrays are really Rails,
then declare them as Rail[double] instead of Array[double]. (Rail is a
typedef for an Array{rank==1,rect,zeroBased,rail}).
Finally, when the code is correct, if you are comparing performance to C++
(unsafe), you may want to disable X10's bounds checking of arrays by
telling the compiler to compile with -NO_CHECKS.
--dave------------------------------------------------------------------------------
How ServiceNow helps IT people transform IT departments:
1. A cloud service to automate IT design, transition and operations
2. Dashboards that offer high-level views of enterprise services
3. A single system of record for all IT processes
http://p.sf.net/sfu/servicenow-d2d-j
_______________________________________________
X10-users mailing list
X10-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/x10-users
