[Gems-users] sequencial benchmark


Date: Wed, 5 Jan 2011 23:25:41 +0800
From: 李 建樟 <piliman9176@xxxxxxxxxxx>
Subject: [Gems-users] sequencial benchmark
Hi all,
 
          I use GEMS + simics to run hello world test program.  like the following
 
          int main()
         {
                  printf("hello world\n");
               
                  return 0;
         }
 
          the ruby configure is
 
          g_NUM_PROCESSORS: 2
          g_NUM_L2_BANKS: 1
          g_NUM_MEMORIES: 1
          g_PROCS_PER_CHIP: 2
          g_MEMORY_SIZE_BYTES: 4294967296
          g_DATA_BLOCK_BYTES: 64
 
          when simulation is finished, I dumped the result.
          There still had two processor  running.
          I passed NodeID  with the argument    m_chip_ptr->getID()*RubyConfig::numberOfProcsPerChip()+m_version
          The NodeID is 0 or 1.
          But I run sequencial code,  why the NodeID had two value?  
          I  use the following command
 
          instruction-fetch-mode instruction-fetch-trace
          istc-disable
          dstc-disable
          cpu-switch-time 1
 
          Another question:    Is  simics  and  GEMS  the number of processor configuration  must the same ?
          Because  if the number is different,  the simics would output error  and  end up simulation.
 
          Above two questions, is there something wrong?
 
          Thanks for your help!
         
Allen
         
[← Prev in Thread] Current Thread [Next in Thread→]