Hi folks:
I ran a test over Ikon and EP10 comparing the performance between these two units with the same application I made. The results were the same in execution time. That result surprised to me because I was expecting a shocking improved performance. But not.
Should I have to do something different in my coding to get a better performance, it might be to use another SDK, dll. or similar thing and take advantage of this new architecture? My application is developped on Visual Studio 2008 C# language.
Thank you,
Carlos
If I understand good the conlusion of above discussion, then Sitara does not bring us ANY performance increase in real applications now (even in interpreted enviromnets like NET or Java). And it will remain until C++ compiler, WHE, NET (or JVM) will be optimized for Sitara processor. Which can take long time, I thing :(
Am I right?
Carlos, I am a group leader in the Software Department at Psion. I would like to have someone in our department take a look at your application and see if there is anything else we can do to help. Would you be willing to send us the source code, build procedures and executables for your test application?
pointx If I understand good the conlusion of above discussion, then Sitara does not bring us ANY performance increase in real applications now (even in interpreted enviromnets like NET or Java).
If I understand good the conlusion of above discussion, then Sitara does not bring us ANY performance increase in real applications now (even in interpreted enviromnets like NET or Java).
ANY is a tad strong in this case... Naturally, clock speed and certain other architectural aspects of these new processors will execute machine instructions faster.
However, we're not going to get a percentage point-per-point increase based on the new specs (i.e. if clock speed doubles, then application execution time should double).
Now, with that said, we should be seeing better results than what we are, and we're addressing that as we speak.
As Dave mentioned above, it would be highly advantageous to get a hold of the application for us to run (source code would be nice too, but the binaries would do just as well).
It seems like we are facing a wall with ARM architecture and .NET engine what will slown down any further development, so your recommendation to boost up applications making a better use of hardware resources would be : 1) to use another development platform instead .NET(C# by example), if so, what would be that one? 2) Wait for an improvement of .net engine 3) Use Java 4) Other
Carlos It seems like we are facing a wall with ARM architecture and .NET engine what will slown down any further development, so your recommendation to boost up applications making a better use of hardware resources would be : 1) to use another development platform instead .NET(C# by example), if so, what would be that one? 2) Wait for an improvement of .net engine 3) Use Java 4) Other
Without seeing the code, or even running the application, it's going to be difficult to point you in the best direction.
I really recommend that you send us the binaries to run from our location, along with source code if possible.
From there, we could analyze the performance as it runs under special debug builds. We may be able to figure out whether we can increase performance on our end, or recommend things that you can do on yours.
I wouldn't bet that Java would get better performance. Its been my experience that you get worse with Java.
You could go native (C/C++), but I couldn't guarantee how much performance you could increase with this route. It might not end up being worth the development effort.
Please consider sending us the binaries to run from here.