It can be caused by several things: a bug in the memory system (unlikely in this case), a bug in the load/store queue (possible, but not completely likely), a bug in the dependence tracking in the CPU (probably the most likely), or ???? (maybe Kevin could create an exhaustive list). The way to find the problem is to compare an execution trace of the O3 cpu with the AtomicSimpleCPU. They should commit the same instructions with the same values and read/write the same data from memory. The execution trace is generated at commit, so if that is the same you would have to start looking the current state in the CPU at the time of the problem.
If you find a bug please let us know, Ali On Jun 8, 2008, at 9:35 PM, nathaniel jones wrote: > hi! > > i modified the 03CPU and and i was able to compile it. it runs hello > world on the regression tests too.. > When i try running eon fron SPEC2000 on it however I get > > M5 compiled Jun 9 2008 09:05:50 > M5 started Mon Jun 9 09:09:15 2008 > M5 executing on system > command line: ./build/ALPHA_SE/m5.opt tests/run.py long/30.eon/alpha/ > tru-64/o3-timing > Global frequency set at 1000000000000 ticks per second > 0: system.remote_gdb.listener: listening for remote gdb #0 on port > 7001 > warn: Entering event queue @ 0. Starting simulation... > warn: Increasing stack size by one page. > warn: ignoring syscall sigprocmask(1, 0, ...) > Eon, Version 1.1 > getting pixel output filename pixels_out.cook > opening control file chair.control.cook > opening camera file chair.camera > opening surfaces file chair.surfaces > reading data > warn: Increasing stack size by one page. > panic: Tried to access unmapped address 0x94. > @ cycle 163786500 > [invoke:build/ALPHA_SE/arch/alpha/faults.cc, line 200] > Program aborted at cycle 163786500 > Aborted > [EMAIL PROTECTED]:~/data/m5$ > > i was able to run eon to completion with the default 03CPU. what > could be causing this? i also encountered unmapped addresses when > trying to run some other benchmarks(like crafty) to completion using > the default 03CPU. is there any way to work around this? > > thanks a bunch! > _______________________________________________ > m5-users mailing list > [email protected] > http://m5sim.org/cgi-bin/mailman/listinfo/m5-users _______________________________________________ m5-users mailing list [email protected] http://m5sim.org/cgi-bin/mailman/listinfo/m5-users
