Hello All, I was running the “456.hmmer” from the SPEC benchmark Suite, which was compiled by the “linux-gcc-sparc” compiler available from the Gem5 website. I did successfully run the “bzip” and “mcf” benchmarks, but while running “hmmer” I ran into error below (I used the script described in Mark Grottscho’s blog to easily run SPEC in GEM5):
Selected SPEC_CPU2006 benchmark -->hmmer Process stdout file: /home/gem5/gem5-stable/Performance /SPARC_default/hmmer/hmmer.out Process stderr file: /home/gem5/gem5-stable/Performance /SPARC_default/hmmer/hmmer.err ['hmmer_base.gcc43-64bit', '--fixed', '0', '--mean', '325', '--num', '45000', '--sd', '200', '--seed', '0', 'bombesin.hmm'] Global frequency set at 1000000000000 ticks per second 8192 Mbytes) does not match the address range assigned (512 Mbytes) 0: system.remote_gdb.listener: listening for remote gdb on port 7000 **** REAL SIMULATION **** info: Entering event queue @ 0. Starting simulation... panic: attempt to execute unimplemented instruction 'for' (inst 0x7c99b30f88) @ tick 979633823 [execute:build/SPARC/arch/sparc/generated/exec-ns.cc.inc, line 31] Memory Usage: 800844 KBytes Program aborted at cycle 979633823 real 0m5.048s user 0m3.420s sys 0m0.051s I have found some similar posts with different ISA’s, but haven’t found a concrete solution. Based on the discussion in previous posts with similar issue, following are the possible solutions: · Change the ISA to more robustly supported ISA by GEM5 community · Add the “for” instruction as a Warning instead of Error for Unimplemented Instruction · Find this instruction which is implemented in another ISA in GEM5 and call that Syscall from the ISA in use (SPARC in this case) · Actually implement the instruction in the ISA. Have anyone added instructions to ISA and have some guidance as to what and where to look at? The other option for calling the Syscall implemented in another ISA, does that maintain the correct functionality if implemented?
_______________________________________________ gem5-users mailing list gem5-users@gem5.org http://m5sim.org/cgi-bin/mailman/listinfo/gem5-users