On Wed, Nov 17, 2010 at 7:52 PM, Gabe Black <[email protected]> wrote:
> You might be accidentally assigning an individual SPMemory to the
> parameter instead of a list of SPMemory objects.
I don't know if that is the case.
"System.spm" is the list of SPMemory objects. After the following loop, for
a 3-cpu configuration, the list has the expected 3 SPMemory objects. An
example of a output from the print:
"[<m5.objects.SPM.SPMemory object at 0x149b250>, <m5.objects.SPM.SPMemory
object at 0x149b850>, <m5.objects.SPM.SPMemory object at 0x149bdd0>]"
Here is the code that sets up the list of objects:
for i, cpu in enumerate(cpus):
cpu.addPrivateSplitL1Caches(L1(size = options.l1size, assoc = 1),
L1(size = options.l1size, assoc = 4))
cpu.addScratchpadMem(SPMemory(size = options.spmsize,
latency = options.spmlatency))
system.spm.append(cpu.spm)
cpu.connectMemPorts(system.toL2bus)
system.spm[i]._parent = system
I don't know it that's the right thing to do, but I had to forcibly change
the parent of each SPMemory to "system" because m5 was throwing the
following error:
"RuntimeError: SimObjectVector elements have inconsistent parent value."
The only line I added to the System.py file was:
=> spm = VectorParam.SPMemory([], "scratchpad memory")
I forgot to mention in the first message that I'm using m5-dev.
Gabe
>
> Gustavo Henrique Nihei wrote:
> > Hi,
> >
> > I created a custom section in the object file, and I need this section
> > to be loaded in several memory modules besides the PhysicalMemory.
> > At first, each of these memory modules are connected to a CPU core,
> > behaving like a scratchpad memory.
> >
> > Is there an easy way to achieve that?
> >
> > I've already modified the ObjectFile, ElfObject, Process and
> > SparcLiveProcess classes.
> > I also added a VectorParam of those memory modules (SPMemory) in the
> > System.py script, which in turn generated a new std::vector in the
> > SystemParams file.
> >
> > * spm = VectorParam.SPMemory([], "scratchpad memory")
> >
> > This vector is populated in the simulation script.
> > Now, when trying to start the simulation, I'm stuck with the following
> > error:
> >
> > * TypeError: object of type 'SwigPyObject' has no len()
> > * swig/python detected a memory leak of type 'std::vector< SPMemory *
> > > *', no destructor found.
> >
> > I am sure that the vector is not empty, since len() returns the
> > expected number of elements.
> >
> > Thanks in advance.
> >
> > --
> > Gustavo Henrique Nihei
> >
> > ------------------------------------------------------------------------
> >
> > _______________________________________________
> > 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
>
--
Gustavo Henrique Nihei
LAPS - Laboratório de Automação do Projeto de Sistemas
NIME - Núcleo Interdepartamental de Microeletrônica
Universidade Federal de Santa Catarina
Florianópolis - Santa Catarina - Brasil
_______________________________________________
m5-users mailing list
[email protected]
http://m5sim.org/cgi-bin/mailman/listinfo/m5-users
