Re: [Gems-users] STAMP in GEMS

[Jiří Šimša <jiri.simsa@xxxxxxxxx>]

Hello everyone,

after successfully resolving my Simics 3.0.31 problems (see below) I am now looking into running the STAMP 0.9.10 benchmark using Log-TM implementation that is part of GEMS 2.2.1.

Has someone tried / succeeded in running the STAMP benchmark on a similar setup? If so, would you mind sharing with me advice on how to do this? I am quite new to using system like Simics and GEMS and would appreciate any help in setting up my experiment.    

As for my intentions, I am trying to investigate the use of static analysis for reducing the
number of conflict checks a hardware transactional memory system has to
carry out (by passing down hints to HTM as to which cache lines need to
be checked).

Best,

--Jiri Simsa

On Mon, Nov 2, 2009 at 4:57 PM, Jiří Šimša <jiri.simsa@xxxxxxxxx> wrote:

Ah, it seems like the problem is in the Simics. Thanks Byn for pointing me out in the correct direction.

 I was able to recreate the same error by simply creating a new Simics workspace, and running "make" and "./simics targets/ebony/ebony-linux-firststeps.simics" in it. It still produces:

Checking out a license... Segmentation fault (SIGSEGV) in main thread

The simulation state has been corrupted. Simulation cannot continue.
Please restart Simics.
There is no frontend to return control to. Simics will exit.

I am currently trying to find out if there is a problem with the license.

Best,

--Jiri

On Mon, Nov 2, 2009 at 4:10 PM, Byn Choi <bynchoi1@xxxxxxxxxxxx> wrote:

This is kind of bizarre. I have to admit I've never encountered this type of problem before. Just so I know, were you able to successfully run vanilla simics without any problem? i.e. run $GEMS/home/$DESTINATION/simics ?

Byn

On Nov 2, 2009, at 10:19 AM, Jiří Šimša wrote:

Hello Byn,

I have actually obtained 64-bit version of Simics and undid my
previous hacks. I am still seeing the same error:

=== Building module "ruby" ===
make[2]: Entering directory
`/media/External_2/SVN/Courses/740/project/gems-2.1.1/simics_3_workspace/amd64-linux/obj/modules/ruby'
Using the Simics 3.0 API for ruby module
CCLD    ruby.so
Segmentation fault (SIGSEGV) in main thread
The simulation state has been corrupted. Simulation cannot continue.
Please restart Simics.
There is no frontend to return control to. Simics will exit.
make[2]: *** [/media/External_2/SVN/Courses/740/project/gems-2.1.1/simics_3_workspace/amd64-linux/lib/ruby.so]
Error 1
make[2]: *** Deleting file
`/media/External_2/SVN/Courses/740/project/gems-2.1.1/simics_3_workspace/amd64-linux/lib/ruby.so'
make[2]: Leaving directory
`/media/External_2/SVN/Courses/740/project/gems-2.1.1/simics_3_workspace/amd64-linux/obj/modules/ruby'
make[1]: *** [ruby] Error 2
make[1]: Leaving directory
`/media/External_2/SVN/Courses/740/project/gems-2.1.1/simics_3_workspace'
make: *** [all] Error 2

In your previous email you said that the call to Simics is failing. Is
that the SIGSEGV? Any ideas how to get it to work? Thanks!

--Jiri

On Sun, Nov 1, 2009 at 9:25 PM, Byn Choi <bynchoi1@xxxxxxxxxxxx> wrote:

Hi,

You should try with SIMICS_HOST environment variable set to x86-linux
instead of hacking the calc_host.sh.

Byn

On Nov 1, 2009, at 8:17 PM, Jiří Šimša wrote:

Hello,

I am trying to build GEMS following the "Quickstart" at the Wiki. I
have obtained and installed x86-linux distribution of Simics 3.0.22.
Although the machine I am trying to build GEMS is x86_64-linux, I have
change the script calc_host.sh to identify my host as "x86-linux" and
added "-m32" flag to both the CFLAGS and LDFLAGS. That way I am able
to successfully built 32-bit Ruby closure on my machine. However, when
the process of making progress pass that point I get the following
error:

Template closure done
Made module
make[1]: Leaving directory
`/media/External_2/SVN/Courses/740/project/gems-2.1.1/ruby'
Ruby: Linking simics module
cd ../simics; make ruby
make[1]: Entering directory
`/media/External_2/SVN/Courses/740/project/gems-2.1.1/simics_3_workspace'
=== Environment Check ===
Checking Python version... Segmentation fault (SIGSEGV) in main thread
The simulation state has been corrupted. Simulation cannot continue.
Please restart Simics.
There is no frontend to return control to. Simics will exit.
Segmentation fault (SIGSEGV) in main thread
The simulation state has been corrupted. Simulation cannot continue.
Please restart Simics.
There is no frontend to return control to. Simics will exit.
Checking compiler... gcc (GCC) 4.1.2 20070925 (Red Hat 4.1.2-27)
Checking GNU make... 3.81
=== Building module "ruby" ===
make[2]: Entering directory

`/media/External_2/SVN/Courses/740/project/gems-2.1.1/simics_3_workspace/x86-linux/obj/modules/ruby'
Using the Simics 3.0 API for ruby module
make[2]: Leaving directory

`/media/External_2/SVN/Courses/740/project/gems-2.1.1/simics_3_workspace/x86-linux/obj/modules/ruby'
make[1]: *** [ruby] Error 2
make[1]: Leaving directory
`/media/External_2/SVN/Courses/740/project/gems-2.1.1/simics_3_workspace'
make: *** [all] Error 2

I have little idea as to what went wrong and why. Can someone please
clarify the error and hopefully offer a solution? Thanks!

Best,

--
Jiri Simsa
Ph.D. Student
Carnegie Mellon University
_______________________________________________
Gems-users mailing list
Gems-users@xxxxxxxxxxx
https://lists.cs.wisc.edu/mailman/listinfo/gems-users
Use Google to search the GEMS Users mailing list by adding
"site:https://lists.cs.wisc.edu/archive/gems-users/" to your search.

--
Jiri Simsa
Ph.D. Student
Carnegie Mellon University

--

Jiri Simsa
Ph.D. Student
Carnegie Mellon University