Hi Bill and everyone, thank you for your reply. actually at the beginning we want to run a multithreaded protocol stack on it, but it truns that it would not be available in time. so we decided to turn to other multithreaded software. the software(or application program) should be able to produce huge computation load, as the protocol stack we wanted to use can produce the load about 4G cycle per second.that is also why we decided to use multiprocessor to handle this.so the software should be able to demostrate this. if such high load can not be fullfilled , then the higher the better. maybe I should give the load in terms of MIPS, but to be honest I don't know it exactly. I think 4G cycles per second, can give us a gerneal idea how hugh it is. of course the software should be multithreaded, so that the OS can distribute threads to different cores. it would be better if the software has something to do with the communication, such as some TCP IP protocol stack, some server etc. if there is any bench mark that fullfills the requirement , it is also ok. I hope I have made it clear. if not just let me know please. I appreciate your help very much. Best regards, PS: if you know any multithreaded application program that uses Pthreads API, I would also like to know that. because I have a ukernel , which supports only pthreads APi, I also want to run some applicatio program on it. the requirement is the same as before, could be simpler. 2007/8/27, Bill Davidsen <davidsen@xxxxxxx>: > Xu Yang wrote: > > Hi everyone, > > > > I want to run some multithreaded software on my realview_eb_mpcore > > system to evaluate the mpcore(with 4 arm11 within it). > > > > > > does anyone know any software or benchmark for this? it would be > > better if the software is involving communication stuff. > > > > Do you require a formal benchmark or just a useful approximation? ie. > for some formal study, or to estimate how much hardware X is than > hardware Y? > > Oh, and doing what? Integer, float, tree chasing through memory? > Frequently these don't scale linearly with the number of threads, in the > CPU or in the test case. Multicore CPUs run out of memory bandwidth, > etc, etc. > > Tell us what you want to compare (in terms of load) and maybe we can > offer better help. > > -- > bill davidsen <davidsen@xxxxxxx> > CTO TMR Associates, Inc > Doing interesting things with small computers since 1979 > > - To unsubscribe from this list: send the line "unsubscribe linux-smp" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html