wine performance question on different machines
Hi All , I am executing a windows program, that calls function in the vendor supplied dll from one of its functions. I executed this program in Linux using wine, in two PC's. PC 1 ( with Enterprize Linux 4.0 ): Processor : Intel Pentium 4 , 2.80 GHz Cache Size : 512 KB RAM : 1 GB Result : When I execute this program in the above machine using wine , I am getting the same performance as in windows for the execution of the function that is in the vendor supplied dll. PC 2 ( with Enterprize Linux 4.0 ): Processor : Intel Xeon , with CPU - 2.80 GHz Cache Size : 1024 KB RAM : 2 GB Result : When I execute this program in the above machine using wine , the execution time of the function that is in the vendor supplied dll is nearly 10 times morethan as in case of first machine. Can any one has same type of observation earlier ? Is there any issue like this w.r.t wine ? Thank you ajay .
Hi, On Mon, Jan 23, 2006 at 06:55:16PM +0530, Ananth M wrote:
Hi All ,
I am executing a windows program, that calls function in the vendor supplied dll from one of its functions.
I executed this program in Linux using wine, in two PC's.
PC 1 ( with Enterprize Linux 4.0 ):
Processor : Intel Pentium 4 , 2.80 GHz Cache Size : 512 KB RAM : 1 GB
Result : When I execute this program in the above machine using wine , I am getting the same performance as in windows for the execution of the function that is in the vendor supplied dll.
PC 2 ( with Enterprize Linux 4.0 ):
Processor : Intel Xeon , with CPU - 2.80 GHz Cache Size : 1024 KB RAM : 2 GB Result : When I execute this program in the above machine using wine , the execution time of the function that is in the vendor supplied dll is nearly 10 times morethan as in case of first machine.
Can any one has same type of observation earlier ? Is there any issue like this w.r.t wine ?
This is not too much data to do some speculations with. E.g. you don't even mention at all what your vendor function does (is it graphics-bound, CPU-bound, network-bound, ...?). I assume both machines have (almost) exact RHEL4 installation state. Are both using an SMP/SMT-aware kernel? Or one does and the other one doesn't? Also, there may be a severe performance hit if P4 is low-memory only whereas Xeon 2GB might be using 1GB low and 1GB high (high is more expensive since it needs remapping!!). Try changing Linux memory mapping (e.g. try booting with "mem=800M" first). Andreas Mohr
Hi, On Mon, Jan 23, 2006 at 06:55:16PM +0530, Ananth M wrote:
When I execute this program in the above machine using wine , the execution time of the function that is in the vendor supplied dll is nearly 10 times morethan as in case of first machine.
Can any one has same type of observation earlier ? Is there any issue like this w.r.t wine ?
Oh, and probably also supply output of vmstat and bonnie and possibly iostat. Andreas Mohr
participants (2)
-
Ananth M -
Andreas Mohr