Alireza, I was having exactly the same problem last week. But my issue was, as Seyed pointed out, that I was not giving in the configuration file a coherent cache for those caches that were private. I see that here you are doing so. So clearly, this is not your issues. One thought that comes to mind is if you compiled Marss with the config=configfilename option and if in the util.cfg file you specified the correct machine name. It could very well be that you are not providing the correct machine name and thus it is not using your coherent caches as you specified them... Just a thought. Let me know how it goes. Good luck.
Tamara. On Mon, Mar 31, 2014 at 6:39 PM, Seyed Majid Zahedi < [email protected]> wrote: > I was mistaken. Since L3 is shared you don't need cache coherence > protocol!!! > > > On Mon, Mar 31, 2014 at 6:09 PM, Seyed Majid Zahedi < > [email protected]> wrote: > >> L3 cache is shared and needs to implement a coherence protocol. >> Use mesi_cache for L3. >> >> Seyed >> >> >> On Mon, Mar 31, 2014 at 5:54 PM, alireza nazari <[email protected]>wrote: >> >>> Hello, >>> >>> This error is killing me! Could anybody help me please? I have run a >>> benchmark(Graph500) on 1,2 and 4 cores and now I am trying to run it on 8 >>> and 16 cores but it errors after 3 days and jumps out! I don't know if >>> that's my config file problem or sth else. >>> The error is >>> Completed 10285993000 cycles, 19495859693 commits: 87446 >>> Hz, 74849 insns/sec: rip ffffffff8105745f ffffffff81013091 >>> ffffffff81013091 ffffffff81013091 0000000000405cb2 ffffffff81013091 >>> ffffffff81013091 ffffffff81013091 ffffffff81013091 ffffffff81013091 >>> ffffffff81013091 ffffffff81013091 ffffffff81013091 ffffffff81013091 >>> ffffffff81013091 ffffffff81013091 >>> [vcpu 0] thread 0: WARNING: At cycle 10286004263, 19495870713 user >>> commits: no instructions have committed for 1048577 cycles; the pipeline >>> could be deadlocked >>> dump_state for core[0]: SMT common structures: >>> >>> the unusual lines in logfile are due to the code that I added into the >>> source to collect some specific data but I dont think my code is having >>> anything to do with this error. >>> Any help is highly appreciated. >>> >>> Thank you >>> Alireza >>> >>> >>> >>> _______________________________________________ >>> http://www.marss86.org >>> Marss86-Devel mailing list >>> [email protected] >>> https://www.cs.binghamton.edu/mailman/listinfo/marss86-devel >>> >>> >> > > _______________________________________________ > http://www.marss86.org > Marss86-Devel mailing list > [email protected] > https://www.cs.binghamton.edu/mailman/listinfo/marss86-devel > >
_______________________________________________ http://www.marss86.org Marss86-Devel mailing list [email protected] https://www.cs.binghamton.edu/mailman/listinfo/marss86-devel
