OVP Forums - A community of assistance, help, questions, and answers.
|
| View previous topic :: View next topic |
| Author |
Message |
KyleMcgrath
Joined: 31 Mar 2017
Posts: 2
|
 Posted: Wed Jun 14, 2017 6:19 am Post subject: iGen generator code for 'manycore24' SystemC TlM2.0 Model |
 |
|
Hello,
I was looking through the package OVPsim_demo_tlm2.0 provided under OSCI SystemC TLM2.0 Models, and the example "manycore24" had a system very similar to what I need for my project. I would like to dynamically change the amount of cores, and potentially processors used with this example. This example was generated with iGen, yet I was unable to find the .tcl file to recreate it.
I was wondering if I could be provided with this code?
Secondly, I when generating my own code that the sc_main function was never filled out. Is there a way that I can configure iGen so it loads the application on each processor in the code like manycore24 does? Or is it necessary for me to specify this myself?
Thanks. |
|
| Back to top |
|
 |
DuncGrah
OVP Technologist
Joined: 27 Feb 2008
Posts: 1654
Location: United Kingdom
|
| Posted: Wed Jun 14, 2017 7:13 am Post subject: |
|
|
I would create a module, in the normal way, that instances the sub-systems required. For example module.op.tcl would be
| Code: | ihwnew -name BareMetal -vendor test.ovpworld.org -library module -version 1.0 -stoponctrlc -releasestatus ovp -visibility visible -purpose module
for {set instance 0} {$instance < 24} {incr instance} {
#
# main bus
#
ihwaddbus -instancename bus${instance} -addresswidth 32
#
# main processor
#
ihwaddprocessor -instancename cpu${instance} \
-type arm -library processor -vendor arm.ovpworld.org \
-semihostvendor arm.ovpworld.org -semihostname armNewlib \
-mips 100.0
#
# processor (default) attributes
#
ihwsetparameter -handle cpu${instance} -name variant -value Cortex-A9MPx1 -type enum
#
# Connect processor to bus
#
ihwconnect -bus bus${instance} -instancename cpu${instance} -busmasterport INSTRUCTION
ihwconnect -bus bus${instance} -instancename cpu${instance} -busmasterport DATA
#
# Memory 0x00000000 to 0xffffffff
# (this may be un-realistic and memory should be provided in correct locations)
#
ihwaddmemory -instancename mem${instance}_0 -type ram
ihwconnect -instancename mem${instance}_0 -bus bus${instance} -busslaveport sp1 -loaddress 0x00000000 -hiaddress 0x1fffffff
ihwaddmemory -instancename mem${instance}_1 -type ram
ihwconnect -instancename mem${instance}_1 -bus bus${instance} -busslaveport sp1 -loaddress 0x20000000 -hiaddress 0xffffffff
}
|
I would then create another file platform.tml.tcl file that loads this and adds the Command Line Parser
| Code: |
# load design defined in module
puts "# Loading module.op.tcl"
source module.op.tcl
# add command line parser
ihwaddclp -allargs
|
This platform.tlm.tcl can be built with a combination of a common Makefile and some additional targets
| Code: | # Default to local build
NOVLNV?=1
USERHEADER?=$(IMPERAS_HOME)/ImperasLib/fileheaders/refESLA.txt
include $(IMPERAS_HOME)/ImperasLib/buildutils/Makefile.common
#
# generate all iGen source
#
IGEN?= igen.exe
DEPENDS?=
IGENFLAGS?=--quiet --nobanner --excludem GPT_NH --excludem GPT_UFNR
all: platform.$(IMPERAS_ARCH).exe
platform-igen-ext = platform.cpp \
platform.sc_clp.igen.h \
platform.sc_constructor.igen.h \
platform.sc_options.igen.h
platform.tlm.tcl : module.op.tcl
@ touch $@
$(platform-igen-ext) : platform.tlm.tcl
@ echo "# iGen Generate Platform files"
$(V) $(IGEN) --batch $^ --writetlm $@ \
$(WRITEHEADER) $(IGENFLAGS) \
--overwrite --op --output igen.log
platform.$(IMPERAS_ARCH).exe : $(platform-igen-ext)
@ echo "# Build $@"
$(V) $(MAKE) -f $(IMPERAS_HOME)/ImperasLib/buildutils/Makefile.TLM.platform \
NOVLNV=1 SRC=platform.cpp OPTLM=1
clean:
$(V) $(MAKE) -f $(IMPERAS_HOME)/ImperasLib/buildutils/Makefile.TLM.platform \
NOVLNV=1 SRC=platform.cpp OPTLM=1 clean
$(V) rm -f *.igen.*
realclean: clean
$(V) $(MAKE) -f $(IMPERAS_HOME)/ImperasLib/buildutils/Makefile.TLM.platform \
NOVLNV=1 SRC=platform.cpp OPTLM=1 realclean
$(V) rm -f platform.cpp *.log
|
that will build a SystemC executable that you can then run with
| Code: | platform.${IMPERAS_ARCH}.exe \
--program application/app.ARM_CORTEX_A9.elf \
--output platformTLM.log \
--verbose
|
Note that I am NOT adding the loading of the application into the hardware definition. This should only be done when there is some resident software that is always loaded otherwise it make using the virtual platform inflexible.
By using --program and specifying the application this program is loaded onto ALL the processors, if we want to load different applications onto processors we can specify the processor to load, e.g. --program <platform>/<processor>=app.elf |
|
| Back to top |
|
|
DuncGrah
OVP Technologist
Joined: 27 Feb 2008
Posts: 1654
Location: United Kingdom
|
| Posted: Wed Jun 14, 2017 7:15 am Post subject: |
|
|
You could also just build the module and run this stand-alone without using SystemC
to build the module
| Code: | make -f $(IMPERAS_HOME)/ImperasLib/buildutils/Makefile.module NOVLNV=1
|
and to run with the module
| Code: | harness.exe \
--modulefile module/model.${IMPERAS_SHRSUF} \
--program application/app.ARM_CORTEX_A9.elf \
--output module.log \
--verbose
|
|
|
| Back to top |
|
|
|
|
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum
|
Information regarding OVP © 2008-2022 Imperas Software
|
FAQ
Search
Register
Profile
Log in to check your private messages
Log in
