Re: Question regarding the Wine Vs WineLib performance
{\rtf1\ansi\ansicpg1252\fromtext \deff0{\fonttbl {\f0\fswiss Arial;} {\f1\fmodern Courier New;} {\f2\fnil\fcharset2 Symbol;} {\f3\fmodern\fcharset0 Courier New;}} {\colortbl\red0\green0\blue0;\red0\green0\blue255;} \uc1\pard\plain\deftab360 \f0\fs20 On 1/3/06, Ananth M <mekaananth(a)gmail.com> wrote:\par
Then I converted the dll into .so and compiled and linked this .so\par with win32 program using WineLib and Wineg++\par The timing that was taken to execute the function exported by the\par dll ( through .so ) is almost two times, compread with executing using wine.\par \par Does any one has any reference for further investigation on this\par problem or any information\par
\par I'm not sure if this is the reason or not, but we discussed this a bit\par at WineConf last year. The concensus was that MSVC produces much\par faster code compared to GCC. Maybe try compiling with MinGW on\par Windows and see what happens with the DLL in Windows?\par \par -Brian\par \par }
On 1/5/06, wine-devel-bounces(a)winehq.org <wine-devel-bounces(a)winehq.org> wrote:
Thank you !!
I will try to compile the windows application using MinGW and check the performance using the same DLL. Here is the some more information that I would like to share with you is : I am using the Binary only DLL. I am not having the source for the DLL. SO I am not compiling the DLL in windows or in Linux. I am linking this binary only dll with my windows application, (since my application uses the function exported by this binary only DLL) Then I moved the executable to Linux and executed using the wine <windows executable>. I checked the execution time of the function exported by the Binary Only DLL In the second Case , I created the .so for the binary only dll, such that the functions that are exported by the binary only dll will be called from the stub functions exported by the .so Then I called the stub function (exported by .so) from my winelib application. But, In this case also, I checked the time duration for the execution of the dll function only.( by adding a code that gives the time duration, in the stub function that calls the dll function). As a third Case, I loaded the binary dll directly (without creating the .so), using LoadLibrary ("Binaryonlydll"), and compiled and linked the application using Wine supplied DLL's and header files. Then I called the function exported by the binary only dll directly from the application and checked the time duration for the execution of the dll function . In this case also we are getting the same timing as in Second Case, but we are not getting the better performance as in first case.
What are the timing for the three classes ? Ananth M wrote:
On 1/5/06, *wine-devel-bounces(a)winehq.org <mailto:wine-devel-bounces(a)winehq.org>* <wine-devel-bounces(a)winehq.org <mailto:wine-devel-bounces(a)winehq.org>> wrote:
Thank you !!
I will try to compile the windows application using MinGW and check the performance using the same DLL.
Here is the some more information that I would like to share with you is : I am using the Binary only DLL. I am not having the source for the DLL. SO I am not compiling the DLL in windows or in Linux. I am linking this binary only dll with my windows application, (since my application uses the function exported by this binary only DLL) Then I moved the executable to Linux and executed using the wine <windows executable>. I checked the execution time of the function exported by the Binary Only DLL
In the second Case , I created the .so for the binary only dll, such that the functions that are exported by the binary only dll will be called from the stub functions exported by the .so Then I called the stub function (exported by .so) from my winelib application. But, In this case also, I checked the time duration for the execution of the dll function only.( by adding a code that gives the time duration, in the stub function that calls the dll function).
As a third Case, I loaded the binary dll directly (without creating the .so), using LoadLibrary ("Binaryonlydll"), and compiled and linked the application using Wine supplied DLL's and header files. Then I called the function exported by the binary only dll directly from the application and checked the time duration for the execution of the dll function .
In this case also we are getting the same timing as in Second Case, but we are not getting the better performance as in first case.
------------------------------------------------------------------------
A number of people have suggested that GCC is the problem. If so, perhaps we should be testing with Intel's C compiler - <http://www.intel.com/cd/software/products/asmo-na/eng/compilers/clin/index.htm> - to see if this offers an improvement. There is an evaluation available for commercial use, and a free download for non-commercial (which by Intel's definitions means only developers who are not being paid) use. -- Troy Rollo - wine(a)troy.rollo.name
participants (4)
-
Ananth M -
Boaz Harrosh -
Troy Rollo -
wine-devel-bounces＠winehq.org