On Mon, May 14, 2012 at 8:46 PM, mihai pricopi <[email protected]>wrote:

> I commented the assert lines for $ and * in the config_gen.py and it
> seems it built now. I checked the generated file and looks OK.
> I am testing the machine now.
>
> But the issue is that in auto-generated code, there will be two caches
with name L1_I_0
and L1_D_0.  This will mess up the stats and other things even if
simulation may run
without any issue.

- Avadh

>
> On Tue, May 15, 2012 at 11:05 AM, mihai pricopi <[email protected]>
> wrote:
> > Thanks Avadh,
> >
> >  Is exactly as I tried before, my cache config looks like this:
> >
> >    caches:
> >      - type: l1_32K_mesi
> >        name_prefix: L1_I_
> >        insts: 6 # Per core L1-I cache
> >        option:
> >            private: true
> >            last_private: true
> >      - type: l1_32K_mesi
> >        name_prefix: L1_D_
> >        insts: 6 # Per core L1-D cache
> >        option:
> >            private: true
> >            last_private: true
> >      - type: l1_64K_mesi
> >        name_prefix: L1_I_
> >        insts: 1 # the coalition cache
> >        option:
> >            private: true
> >            last_private: true
> >      - type: l1_64K_mesi
> >        name_prefix: L1_D_
> >        insts: 1 # the coalition cache
> >        option:
> >            private: true
> >            last_private: true
> >      - type: l2_256K
> >        name_prefix: L2_
> >        insts: 1 # Shared L2 config
> >    memory:
> >      - type: dram_cont
> >        name_prefix: MEM_
> >        insts: 1 # Single DRAM controller
> >        option:
> >            latency: 112
> >    interconnects:
> >      - type: p2p
> >        # '$' sign is used to map matching instances like:
> >        # cpu_0, L1_I_0
> >        connections:
> >            - core_$: I
> >              L1_I_$: UPPER
> >            - core_$: D
> >              L1_D_$: UPPER
> >            - L2_0: LOWER
> >              MEM_0: UPPER
> >      - type: mesi_bus
> >        connections:
> >            - L1_I_*: LOWER
> >              L1_D_*: LOWER
> >              L2_0: UPPER
> >
> > But it doesn't build. This is the build error I get:
> >
> > Compiling :: ptlsim/build/sim/ptlsim.cpp ==> ptlsim/build/sim/ptlsim.o
> > Traceback (most recent call last):
> >  File "/home/mihai/work/marss/ptlsim/tools/config_gen.py", line 517,
> > in <module>
> >    gen_output_file(config, options)
> >  File "/home/mihai/work/marss/ptlsim/tools/config_gen.py", line 502,
> > in gen_output_file
> >    generate_machine(config, options)
> >  File "/home/mihai/work/marss/ptlsim/tools/config_gen.py", line 426,
> > in generate_machine
> >    write_interconn_logic(config, m_conf, of)
> >  File "/home/mihai/work/marss/ptlsim/tools/config_gen.py", line 337,
> > in write_interconn_logic
> >    assert c_cfg["insts"] == "$NUMCORES"
> > AssertionError
> >
> > Is there any constraint that the $ insts must be equal to NUMCORES ?
> >
> > Thanks
> >
> > On Tue, May 15, 2012 at 1:20 AM, avadh patel <[email protected]>
> wrote:
> >>
> >> On Mon, May 14, 2012 at 4:27 AM, mihai pricopi <[email protected]
> >
> >> wrote:
> >>>
> >>> Hi,
> >>>
> >>>  I want to build a 7-core machine where 6 cores use l1_32K_ caches and
> >>> the 7'th uses a l1_64K cache. I am having difficulties adapting the
> >>> config file for this.
> >>>  Can this be done using the cache section and interconnection ?
> >>>
> >> It can be done, but its a little tricky configuration.
> >>
> >> First create machine with min_context and max_context set to 7.
> >> Create cores with insts: $NUMCORES
> >>
> >> In caches, create 6 instances of L1_32K_cache and 1 instance of
> >> L1_64K_cache and prefix them with same name 'L1_'.
> >>
> >> Now when you connect those caches to cores use standard '$' sign
> >> to map core-0 to L1_0 and so no.  As your L1_64K_cache will be
> >> named as L1_6 (it starts with 0), your core-6 will be connected to
> >> L1_6.
> >>
> >> Try this configuration, and to make sure everything is created
> >> correctly, check the generated cpp file :
> >> 'build/sim/gen_YOUR_MACHINE_NAME.cpp'.
> >>
> >>
> >> - Avadh
> >>>
> >>>
> >>> Thanks
> >>>
> >>> _______________________________________________
> >>> http://www.marss86.org
> >>> Marss86-Devel mailing list
> >>> [email protected]
> >>> https://www.cs.binghamton.edu/mailman/listinfo/marss86-devel
> >>
> >>
>
_______________________________________________
http://www.marss86.org
Marss86-Devel mailing list
[email protected]
https://www.cs.binghamton.edu/mailman/listinfo/marss86-devel

Reply via email to