If you issue the command
list-failed-modules
in Simics, Simics should tell you why it did not load the module. Odds
are it will claim there is a version mismatch. If this is the case:
1) In the flexus root directory, issue the command:
make all-clobber
2) In <simics install>/x86-linux/lib, open modcap.buf. Remove any lines
that refer to flexus simulators.
3) Rebuild flexus (this will be a complete rebuild from scratch).
4) Try loading again in Simics. If it was a version number mismatch, it
should now work.
Regards,
-Tom Wenisch
On Sat, 12 Nov 2005, shan wrote:
> Hi Tom,
> Are you still available through e-mail now?
> I tried 2.2.19, but I can not find the flexus model when I typed
> list-modules in simics window. At first I thought this is some simics
> version problem, now based on what you said, I guess there is something else
> wrong. However, I have double checked that, I have changed the SIMCS path in
> flexus make file and I have run the make install, copy the libflexus...so to
> simics-2.2.19/x86-linux/lib. I don't know what is wrong.
> Do you have any clue for this?
> Thanks so much
> Shan
>
> -----Original Message-----
> From: Thomas Wenisch [mailto:[email protected]]
> Sent: Saturday, November 12, 2005 3:42 AM
> To: shan
> Subject: Re: [Simflex] [help~~~~]does anyone has simics 2.2.12?
>
> Shan,
>
> Although it has not been tested, it is very likely that Flexus will work
> fine with a newer release in the 2.2.x series. It is relatively unusual
> (but has happened) for a Simics release to break us.
>
> Regards,
> -Tom Wenisch
>
> On Sat, 12 Nov 2005, shan wrote:
>
> > Hi,
> > Does any one has the simics 2.2.12? I just found that simics 2.0.18
> seems
> > to be much slower than 2.2.12. And I cannot find simics 2.2.12 from the
> > simics website :(.
> > Can anyone help me?
> > Thanks very much
> > Shan
> >
> > _______________________________________________
> > SimFlex mailing list
> > [email protected]
> > https://sos.ece.cmu.edu/mailman/listinfo/simflex
> > SimFlex web page: http://www.ece.cmu.edu/~simflex
> >
>
>
From penglu01 at hotmail.com Mon Nov 14 14:51:00 2005
From: penglu01 at hotmail.com (lu peng)
List-Post: [email protected]
Date: Mon Nov 14 14:53:30 2005
Subject: [Simflex] About Piranha Cache
In-Reply-To: <[email protected]>
Message-ID: <[email protected]>
An HTML attachment was scrubbed...
URL:
http://sos.ece.cmu.edu/pipermail/simflex/attachments/20051114/a99c4a46/attachment.html
From jsmolens+ at ece.cmu.edu Mon Nov 14 21:55:29 2005
From: jsmolens+ at ece.cmu.edu (Jared C. Smolens)
List-Post: [email protected]
Date: Mon Nov 14 21:55:34 2005
Subject: [Simflex] About Piranha Cache
Message-ID: <[email protected]>
Hi Lu,
See inline...
Excerpts From "lu peng" <[email protected]>:
[Simflex] About Piranha Cache: "lu peng" <[email protected]>
><html><div style='background-color:'><P>Hi, Tom and Jared,</P>
><P>I read Piranha cache code. Here is a question. Seems that all cache
are
>created by BaseCacheControllerImpl::construct with a parameter
>eCacheType type. If this type is kPiranhaCache, then it will call
new
>PiranhaCacheControllerImpl(). However, I can't find any caller with
>kPiranhaCache as its input. My question is where was the Piranha Cache
>created?
The basic flow of instantiation starts from CmpCacheImpl.cpp in the
CmpCache component. A CacheController is instantiated from here with the
cacheTypeName set to "piranha". This string is translated into a
kPiranhaCache, which is used in the BaseCacheControllerImpl::construct()
function you mentioned above. This structure allows us to have different
wiring interfaces for the CmpCache vs Cache components, but reuse most of
the core cache code.
>In addition, Jared, could you
> please take a look at my last message about building a D-NUCA?</P>
CMP cache models a single delay for visiting the tag and data arrays at
the shared L2.
The directory is a logical structure in the Piranha cache which maintains
cache line locations and state. You shouldn't have any problem adding
more fields to a directory entry. However, this logical directory should
not be confused with a physical hardware structure in an actual
processor. For instance, Compaq's Piranha uses duplicate L1 tags at the
L2 to track the same information, but that is not the only possible
implementation. You will have to think carefully about where and how
your D-NUCA cache stores the information you need and when it needs to be
accessed.
Cheers,
Jared
Jared Smolens ----------- Electrical and Computer Engineering
www.rabidpenguin.org ------------- Carnegie Mellon University
jsmolens AT ece.cmu.edu ------ HH A-313 ------ Pittsburgh, PA
