Re: [Gems-users] SMT in GEMS2.1

[Luke Yen <lyen@xxxxxxxxxxx>]

   No, numProcs should be exactly equal to g_NUM_PROCESSORS, and is 
equal to the number of processors in the checkpoint used by Simics.
  In your case, if g_NUM_PROCESSORS = 4, g_NUM_SMT_THREADS = 2, and 
CONFIG_LOGICAL_PER_PHY_PROC = 2, numProcs should be equal to 4 and 
numSMTProcs should be equal to 2.
  Luke

On Tue, 25 Mar 2008, Xuan Qi wrote:

> Hi Luke,
>
> Thanks for your quick reply. I think I didn't explain my confusion well.
>
> If I have 4 physical procs initiated by simics, and 2 threads/processor,
> as Logical = total number of HW thread contexts, numProcs should be 4 * 2
> = 8.
>
> However, when setting the parameter g_NUM_PROCESSORS, I should set it to
> 4 not 8 (right?). And 4 is passed into and numProcs will be assigned 4. So
> there is a mismatch here as numProcs is supposed to be 8 if I understand
> correctly. Any idea how should I configure to make numProcs assigned to
> the right value? Thank you very much for your patience...
>
>
> Best regards,
> Xuan Qi
>
>
> On Tue, 25 Mar 2008, Luke Yen wrote:
>
> >    You need checkpoints that support the total number of g_NUM_PROCESSORS
> > (in your case, 4).  It might be that you are trying to simulate a CMP with
> > 4 HW contexts with only 2 physical processors, which is not allowed.
> >
> >   Luke
> >
> > On Tue, 25 Mar 2008, Xuan Qi wrote:
> >
> > > Thanks Luke, could you help me with one more question:
> > >
> > > I am simulating a system with 2 processor, each processor supports 2
> > > threads. This is the configuration in my script, which basically follows
> > > mfacet.py:
> > >
> > > ruby0.setparam g_NUM_PROCESSORS 2
> > > ruby0.setparam g_PROCS_PER_CHIP 2
> > > ruby0.setparam g_NUM_SMT_THREADS 2
> > > ruby0.setparam g_NUM_MEMORIES 1
> > > ruby0.setparam g_MEMORY_SIZE_BYTES 536870912
> > > ruby0.setparam NUMBER_OF_VIRTUAL_NETWORKS 6
> > > ruby0.init
> > > load-module opal
> > > opal0.setparam CONFIG_LOGICAL_PER_PHY_PROC 2
> > > opal0.setparam CONFIG_IREG_PHYSICAL 384
> > > opal0.setparam CONFIG_FPREG_PHYSICAL 256
> > > opal0.setparam CONFIG_CCREG_PHYSICAL 74
> > > opal0.init
> > >
> > > And I add this line in system.C just to check if the value of numProcs and
> > > m_numSMTProcs is corret:
> > > printf("numProcs[%d] numSMTProcs[%d]\n", numProcs, m_numSMTProcs);
> > >
> > > The print out is:
> > > numProcs[2] numSMTProcs[1]
> > >
> > > If I understand your explanation correctly, numProcs should be 4 and
> > > numSMTProcs should be 2 in this case. And my configuration apparently
> > > didnt pass in the correct values. How should I set the parameters to make
> > > the value correct? I tried to do: ruby0.setparam g_NUM_PROCESSORS 4, but
> > > simics won't take it... Thank you.
> > >
> > >
> > > Best regards,
> > > Xuan Qi
> > >
> > >
> > > On Tue, 25 Mar 2008, Luke Yen wrote:
> > >
> > > > > If this is the case, the first assert should be:
> > > > >
> > > > > ASSERT( (cpuNumber >= 0) && (cpuNumber < system_t::inst->m_numSMTProcs) );
> > > > >
> > > > > instead of:
> > > > >
> > > > > ASSERT( (cpuNumber >= 0) && (cpuNumber < system_t::inst->m_numProcs) );
> > > > >
> > > > > Is it?
> > > >     You can change it if you want, but in any case the error condition you
> > > > care about will be caught by the assertion on seq_index, since the
> > > > assignment makes seq_index get the value of cpuNumber.
> > > >
> > > > > (2) in system.C
> > > > >
> > > > > first, the numProcs is defined as:
> > > > >
> > > > > // establish how many processors there are in the system
> > > > >  int numProcs = SIM_number_processors();   //for SMT, this is the total
> > > > > number of LOGICAL processors in the system
> > > > >
> > > > > then,
> > > > >
> > > > > m_numSMTProcs = numProcs / CONFIG_LOGICAL_PER_PHY_PROC;
> > > > >
> > > > > I am confused here because numProcs is first defined as the total number
> > > > > of LOGICAL processors in the system. Is this the number of SMT processors?
> > > > > If so, m_numSMTProcs = numProcs / CONFIG_LOGICAL_PER_PHY_PROC doesn't make
> > > > > sense. It looks to me that numProcs should be set to the number
> > > > > of PHYSICAL processors initiated by simics, is it? Maybe I misunderstand
> > > > > the definition of LOGICAL processor?
> > > >     My definitions:
> > > >       Logical = total number of HW thread contexts
> > > >             (e.g. physical procs*(SMT contexts per proc), available to Simics for
> > > >                   scheduling)
> > > >       Physical = total number of physical procs (as seen by Ruby)
> > > >
> > > >     Luke
> > > > _______________________________________________
> > > > Gems-users mailing list
> > > > Gems-users@xxxxxxxxxxx
> > > > https://lists.cs.wisc.edu/mailman/listinfo/gems-users
> > > > Use Google to search the GEMS Users mailing list by adding "site:https://lists.cs.wisc.edu/archive/gems-users/"; to your search.
> > > _______________________________________________
> > > Gems-users mailing list
> > > Gems-users@xxxxxxxxxxx
> > > https://lists.cs.wisc.edu/mailman/listinfo/gems-users
> > > Use Google to search the GEMS Users mailing list by adding "site:https://lists.cs.wisc.edu/archive/gems-users/"; to your search.
> > _______________________________________________
> > Gems-users mailing list
> > Gems-users@xxxxxxxxxxx
> > https://lists.cs.wisc.edu/mailman/listinfo/gems-users
> > Use Google to search the GEMS Users mailing list by adding "site:https://lists.cs.wisc.edu/archive/gems-users/"; to your search.
> _______________________________________________
> Gems-users mailing list
> Gems-users@xxxxxxxxxxx
> https://lists.cs.wisc.edu/mailman/listinfo/gems-users
> Use Google to search the GEMS Users mailing list by adding "site:https://lists.cs.wisc.edu/archive/gems-users/"; to your search.