Hi, I tried to run a 16 cores CMP configuration on SE O3 cpu mode. I use Spec2006 445.gobmk, and each core has one copy of this benchmark. It works for 1, 2, 4, 6, 8 cores' configuration, and the benchmarks can finish correctly. But when I try to increase the number of cores (>8), I begin to encounter the problems.
Why does this happen on larger number of cores? I tried the timing mode, it works fine on 16 cores' configuration. Also the largest number of cores is depend on the benchmarks. Thanks a lot, Meng-Ju M5 compiled May 31 2008 14:23:59 M5 started Wed Jun 4 12:10:13 2008 M5 executing on avw1450-cearch_21 command line: /import/home1/mjwu/m5-2.0b5/build/ALPHA_SE/m5.fast --outdir=16cpu_p /import/home1/mjwu/work_spec2006/configs/cmt4.py -d --caches --num_cpus 16 --l2cache --benchmark gobmk /import/home1/mjwu/work_spec2006/configs /import/home1/mjwu/work_spec2006/configs /import/home1/mjwu/work_spec2006 Global frequency set at 1000000000000 ticks per second Dumping Instruction List Instruction:0 PC:0x1200001e0 [tid:0] [sn:88] Issued:0 Squashed:0 Instruction:1 PC:0x1200001e4 [tid:0] [sn:89] Issued:0 Squashed:0 Instruction:2 PC:0x1200001e8 [tid:0] [sn:90] Issued:0 Squashed:0 Instruction:3 PC:0x1200001ec [tid:0] [sn:91] Issued:0 Squashed:0 Instruction:4 PC:0x1200001f0 [tid:0] [sn:92] Issued:0 Squashed:0 Instruction:5 PC:0x1200001f4 [tid:0] [sn:93] Issued:0 Squashed:0 _______________________________________________ m5-users mailing list [email protected] http://m5sim.org/cgi-bin/mailman/listinfo/m5-users
