Thanks for all your responses. I didn't seem like Xilinx has support
for checkpointing of the swift models.
To get around this issue, I have been letting the simulation run well
past where I think will be analyzing....So far, that works for me.
NN
Brian Drummond wrote:
> On Thu, 23 Feb 2006 17:50:15 +0000 (UTC), Brian Drummond
> <brian_drummond@btconnect.com> wrote:
>
> >On 22 Feb 2006 16:30:02 -0800, "Nju Njoroge" <njoroge@stanford.edu>
> >wrote:
> >
> >>Hello,
> >>
> >>I have set-up checkpointing in ModelSim 6.0b (also using EDK 7.1 SP2).
> >>ModelSim complains that "two foreign architectures are missing save and
> >>restore callbacks" when I checkpoint. These "foreign architectures"
> >>happen to be the PPC swift models. When I restore the checkpoints, the
> >>waveforms are fully restored. However, I cannot continue the simulation
> >>because it makes the following complaint:
> >>
> >># ** Fatal (SmartModel):
> >># SWIFT protocol:
> >># The first call to lsm_Model_DCEvaluate must occur at time 0.
> >># Time: 424875000 ps
> >>Instance:/system/ppc405_0/ppc405_0/ppc405_i/ippc405_swift/ppc405_swift_inst
> >># ** Fatal: Foreign module requested halt.
> >># Time: 424875 ns Iteration: 0 Foreign Process:
> >>/system/ppc405_0/ppc405_0/ppc405_i/ippc405_swift/ppc405_swift_inst/SmartModel
> >>File: Foreign
> >># Fatal error at an unknown location
> >>#
> >>
> >>Anyone encountered this issue?
> >
> >Not quite; but I am having problems AT time 0ps; with the message
> >
> ># ** Fatal (SmartModel):
> ># SWIFT protocol:
> ># Call to lsm_Model_DCEvaluate must be preceded by a call
> ># to lsm_Model_SetSimTimeUnit.
> >
> >May be related; I don't have any answers yet. (I have had other projects
> >simulate correctly with SwiftModels; but this is my first of any
> >complexity, as opposed to EDK projects with one toplevel ISE module)
> >
>
> Mine appears to be related to starting ModelSim from within another tool
> (HDL Designer). The SmartModels don't seem to play well with the other
> stuff involved (cross-probing with the block diagrams).
>
> I don't know if this sheds any light on the checkpointing problem;
> suspect it doesn't.
>
> - Brian
Reply by Brian Drummond●February 24, 20062006-02-24
On Thu, 23 Feb 2006 17:50:15 +0000 (UTC), Brian Drummond
<brian_drummond@btconnect.com> wrote:
>On 22 Feb 2006 16:30:02 -0800, "Nju Njoroge" <njoroge@stanford.edu>
>wrote:
>
>>Hello,
>>
>>I have set-up checkpointing in ModelSim 6.0b (also using EDK 7.1 SP2).
>>ModelSim complains that "two foreign architectures are missing save and
>>restore callbacks" when I checkpoint. These "foreign architectures"
>>happen to be the PPC swift models. When I restore the checkpoints, the
>>waveforms are fully restored. However, I cannot continue the simulation
>>because it makes the following complaint:
>>
>># ** Fatal (SmartModel):
>># SWIFT protocol:
>># The first call to lsm_Model_DCEvaluate must occur at time 0.
>># Time: 424875000 ps
>>Instance:/system/ppc405_0/ppc405_0/ppc405_i/ippc405_swift/ppc405_swift_inst
>># ** Fatal: Foreign module requested halt.
>># Time: 424875 ns Iteration: 0 Foreign Process:
>>/system/ppc405_0/ppc405_0/ppc405_i/ippc405_swift/ppc405_swift_inst/SmartModel
>>File: Foreign
>># Fatal error at an unknown location
>>#
>>
>>Anyone encountered this issue?
>
>Not quite; but I am having problems AT time 0ps; with the message
>
># ** Fatal (SmartModel):
># SWIFT protocol:
># Call to lsm_Model_DCEvaluate must be preceded by a call
># to lsm_Model_SetSimTimeUnit.
>
>May be related; I don't have any answers yet. (I have had other projects
>simulate correctly with SwiftModels; but this is my first of any
>complexity, as opposed to EDK projects with one toplevel ISE module)
>
Mine appears to be related to starting ModelSim from within another tool
(HDL Designer). The SmartModels don't seem to play well with the other
stuff involved (cross-probing with the block diagrams).
I don't know if this sheds any light on the checkpointing problem;
suspect it doesn't.
- Brian
Reply by Brian Drummond●February 23, 20062006-02-23
On 22 Feb 2006 16:30:02 -0800, "Nju Njoroge" <njoroge@stanford.edu>
wrote:
>Hello,
>
>I have set-up checkpointing in ModelSim 6.0b (also using EDK 7.1 SP2).
>ModelSim complains that "two foreign architectures are missing save and
>restore callbacks" when I checkpoint. These "foreign architectures"
>happen to be the PPC swift models. When I restore the checkpoints, the
>waveforms are fully restored. However, I cannot continue the simulation
>because it makes the following complaint:
>
># ** Fatal (SmartModel):
># SWIFT protocol:
># The first call to lsm_Model_DCEvaluate must occur at time 0.
># Time: 424875000 ps
>Instance:/system/ppc405_0/ppc405_0/ppc405_i/ippc405_swift/ppc405_swift_inst
># ** Fatal: Foreign module requested halt.
># Time: 424875 ns Iteration: 0 Foreign Process:
>/system/ppc405_0/ppc405_0/ppc405_i/ippc405_swift/ppc405_swift_inst/SmartModel
>File: Foreign
># Fatal error at an unknown location
>#
>
>Anyone encountered this issue?
Not quite; but I am having problems AT time 0ps; with the message
# ** Fatal (SmartModel):
# SWIFT protocol:
# Call to lsm_Model_DCEvaluate must be preceded by a call
# to lsm_Model_SetSimTimeUnit.
May be related; I don't have any answers yet. (I have had other projects
simulate correctly with SwiftModels; but this is my first of any
complexity, as opposed to EDK projects with one toplevel ISE module)
- Brian
Reply by Hans●February 23, 20062006-02-23
You need to ask Xilinx if checkpointing/restore is supported for the PowerPC
model since you need to hardcode it in. Have a look at the FLI (used for
swift) documentation to get an idea what is required.
Hans
www.ht-lab.com
"Nju Njoroge" <njoroge@stanford.edu> wrote in message
news:1140654602.292568.161850@g43g2000cwa.googlegroups.com...
> Hello,
>
> I have set-up checkpointing in ModelSim 6.0b (also using EDK 7.1 SP2).
> ModelSim complains that "two foreign architectures are missing save and
> restore callbacks" when I checkpoint. These "foreign architectures"
> happen to be the PPC swift models. When I restore the checkpoints, the
> waveforms are fully restored. However, I cannot continue the simulation
> because it makes the following complaint:
>
> # ** Fatal (SmartModel):
> # SWIFT protocol:
> # The first call to lsm_Model_DCEvaluate must occur at time 0.
> # Time: 424875000 ps
> Instance:/system/ppc405_0/ppc405_0/ppc405_i/ippc405_swift/ppc405_swift_inst
> # ** Fatal: Foreign module requested halt.
> # Time: 424875 ns Iteration: 0 Foreign Process:
> /system/ppc405_0/ppc405_0/ppc405_i/ippc405_swift/ppc405_swift_inst/SmartModel
> File: Foreign
> # Fatal error at an unknown location
> #
>
> Anyone encountered this issue?
>
> Thanks,
>
> NN
>
Reply by Nju Njoroge●February 22, 20062006-02-22
Hello,
I have set-up checkpointing in ModelSim 6.0b (also using EDK 7.1 SP2).
ModelSim complains that "two foreign architectures are missing save and
restore callbacks" when I checkpoint. These "foreign architectures"
happen to be the PPC swift models. When I restore the checkpoints, the
waveforms are fully restored. However, I cannot continue the simulation
because it makes the following complaint:
# ** Fatal (SmartModel):
# SWIFT protocol:
# The first call to lsm_Model_DCEvaluate must occur at time 0.
# Time: 424875000 ps
Instance:/system/ppc405_0/ppc405_0/ppc405_i/ippc405_swift/ppc405_swift_inst
# ** Fatal: Foreign module requested halt.
# Time: 424875 ns Iteration: 0 Foreign Process:
/system/ppc405_0/ppc405_0/ppc405_i/ippc405_swift/ppc405_swift_inst/SmartModel
File: Foreign
# Fatal error at an unknown location
#
Anyone encountered this issue?
Thanks,
NN