Chris Pollach's Blog Site for up-to-date news and information on various Sybase products, technologies, and methodologies.
 
The Source of Quality PowerBuilder™ Information Since 1994
   
 
Chris Pollach - Great White North Evangelist

Search

 

PB 11 blows away VS2005's C#

posted 2006.05.01 Monday
All;
 
    Last night I took the PowerBuilder benchmark application (similar to what the database vendors do in their TPC benchmarks) and compiled a native PB machine executable (C++) and a VS2005 C#.net executable. Since they are functionally identical, the same coding (C# and PB scripting is so similar it's scary) logic was executed. The outcome though is quite interesting .. here is what I found (see test run also ):
 
Action                 Outcome
=====                ===========
 
Looping               - 300% better in PB (3 : 1)
TAK                    - same
Sort                    - 500% better in PB (5 : 1)
Global Functions  - same
Object Functions  - 200% better in C# (2 : 1)
Event handling      - 200% better in C# (2 : 1)
Instantiation O/H  - 133% better in PB (1.3 : 1)
String Parsing      - almost the same
Global Variable    - 420% better in PB (4.2 : 1)
Global Constant   - 493% better in PB (4.9 : 1)
Data Handling      - 424% better in PB (4.2 : 1)
 
    I was quit interested in the sorting and manipulating of data as these two pieces of functionality are heavily used in most business applications. What I would like to do as well is compare PB's native DB driver I/O to C#'s ADO.net in Retrieving and Updating ASE and SQLServer2005 data (OK, I will save that for another icy day <lol>)! Another interesting thing would be to do the same comparison between VB.Net and C#.net and see what the execution speed difference would be even in the pure MS world (VB is P-Code, not machine code like PB and C#).
 
FYI: The TAK benchmark is (and I had to look this one up ... learned something new! *g* ):
 
    The TAK benchmark is a variant of a benchmark originated by Japanese researcher Ikuo Takeuchi and is widely used to compare programming  tools that make extensive use of function calls.
The commonly used TAK benchmark algorithm is a variant originated when John McCarthy of Stanford University recalled the original Takeuchi function incorrectly at the start of a benchmarking effort. TAK uses a recursive function that when called with the initial values of 18, 12, and 6 will put the processor through 63,609 function calls and 47,706 subtractions by 1.
 
   Also, check out the better PB vs C# database performance as well - even against MS's own SQLServer: http://www.dotnet247.com/247reference/msgs/29/148119.aspx !!!
 
Regards ... Chris
Use the "Force" (PB) Luke!