I had this issue on Centos 5.5 (a free derivative of Ret Hat Enterprise Linux) today.
It seems that the older gcc 4.1.2 compiler wasn't recognizing the mtune and march switches. What I did to solve the problem involved manually commenting out the -march and -mtune switches in ptlsim/SConstruct and qemu/SConstruct files. Hope this helps. Zhong Liang Ong ________________________________________ From: [email protected] [[email protected]] On Behalf Of [email protected] [[email protected]] Sent: Friday, May 13, 2011 12:00 AM To: [email protected] Subject: Marss86-Devel Digest, Vol 15, Issue 10 Send Marss86-Devel mailing list submissions to [email protected] To subscribe or unsubscribe via the World Wide Web, visit https://www.cs.binghamton.edu/mailman/listinfo/marss86-devel or, via email, send a message with subject or body 'help' to [email protected] You can reach the person managing the list at [email protected] When replying, please edit your Subject line so it is more specific than "Re: Contents of Marss86-Devel digest..." Today's Topics: 1. Re: Fw: RE: Re: marss on cygwin windows Vista ? (DRAM Ninjas) 2. Switch to native mode and ptlstat results (Hui Kang) 3. Re: Switch to native mode and ptlstat results (Hui Kang) 4. Re: Switch to native mode and ptlstat results (Erdem Aktas) 5. Re: Switch to native mode and ptlstat results (Hui Kang) 6. consumption cpu and associatitity caches (Carlos Jos? Martinez Gavil?n) 7. Re: Switch to native mode and ptlstat results (Hui Kang) 8. Re: Switch to native mode and ptlstat results (DRAM Ninjas) ---------------------------------------------------------------------- Message: 1 Date: Wed, 11 May 2011 18:56:23 -0400 From: DRAM Ninjas <[email protected]> To: paul cao <[email protected]> Cc: [email protected] Subject: Re: [marss86-devel] Fw: RE: Re: marss on cygwin windows Vista ? Message-ID: <[email protected]> Content-Type: text/plain; charset="iso-8859-1" can you post the output of g++ -v ? On Wed, May 11, 2011 at 10:45 AM, paul cao <[email protected]> wrote: > > In a separate effort, run scons -Q on Redhat Linux 5, and see a following > error: > > Any idea ? > > > [root@specclient1 avadhpatel-marss-0a29c42]# scons -Q > running with -j32 > g++ -g3 -O3 -march=native -mtune=native -DDISABLE_LOGGING > -fno-trapping-math -fno-stack-protector -fno-exceptions -fno-rtti > -funroll-loops -fno-strict-aliasing -DNUM_SIM_CORES=1 > -DSINGLE_CORE_MEM_CONFIG -DNEED_CPU_H -D__STDC_FORMAT_MACROS -DMARSS_QEMU > -D__x86_64__ -DMONGO_HAVE_STDINT > -I/marss/avadhpatel-marss-0a29c42/ptlsim/cache > -I/marss/avadhpatel-marss-0a29c42/ptlsim/core > -I/marss/avadhpatel-marss-0a29c42/ptlsim/lib > -I/marss/avadhpatel-marss-0a29c42/ptlsim/sim > -I/marss/avadhpatel-marss-0a29c42/ptlsim/stats > -I/marss/avadhpatel-marss-0a29c42/ptlsim/tools > -I/marss/avadhpatel-marss-0a29c42/ptlsim/x86 > -I/marss/avadhpatel-marss-0a29c42/qemu > -I/marss/avadhpatel-marss-0a29c42/qemu/target-i386 > -I/marss/avadhpatel-marss-0a29c42/qemu/fpu > -I/marss/avadhpatel-marss-0a29c42/qemu/x86_64-softmmu -E -C > ptlsim/stats/stats.h > ptlsim/build/stats.i > ptlsim/stats/stats.h:0: error: bad value (native) for -march= switch > ptlsim/stats/stats.h:0: error: bad value (native) for -mtune= switch > Compiling ==> ptlsim/build/cache/bus.cpp > ptlsim/build/cache/bus.cpp:1: error: bad value (native) for -march= switch > ptlsim/build/cache/bus.cpp:1: error: bad value (native) for -mtune= switch > Compiling ==> ptlsim/build/cache/cacheController.cpp > ptlsim/build/cache/cacheController.cpp:1: error: bad value (native) for > -march= switch > ptlsim/build/cache/cacheController.cpp:1: error: bad value (native) for > -mtune= switch > Compiling ==> ptlsim/build/cache/cpuController.cpp > ptlsim/build/cache/cpuController.cpp:1: error: bad value (native) for > -march= switch > ptlsim/build/cache/cpuController.cpp:1: error: bad value (native) for > -mtune= switch > Compiling ==> ptlsim/build/cache/dcache.cpp > ptlsim/build/cache/dcache.cpp:1: error: bad value (native) for -march= > switch > ptlsim/build/cache/dcache.cpp:1: error: bad value (native) for -mtune= > switch > Compiling ==> ptlsim/build/cache/memoryController.cpp > ptlsim/build/cache/memoryController.cpp:1: error: bad value (native) for > -march= switch > ptlsim/build/cache/memoryController.cpp:1: error: bad value (native) for > -mtune= switch > Compiling ==> ptlsim/build/cache/memoryHierarchy.cpp > ptlsim/build/cache/memoryHierarchy.cpp:1: error: bad value (native) for > -march= switch > ptlsim/build/cache/memoryHierarchy.cpp:1: error: bad value (native) for > -mtune= switch > Compiling ==> ptlsim/build/cache/memoryRequest.cpp > ptlsim/build/cache/memoryRequest.cpp:1: error: bad value (native) for > -march= switch > ptlsim/build/cache/memoryRequest.cpp:1: error: bad value (native) for > -mtune= switch > Compiling ==> ptlsim/build/cache/mesiBus.cpp > ptlsim/build/cache/mesiBus.cpp:1: error: bad value (native) for -march= > switch > ptlsim/build/cache/mesiBus.cpp:1: error: bad value (native) for -mtune= > switch > Compiling ==> ptlsim/build/cache/mesiCache.cpp > ptlsim/build/cache/mesiCache.cpp:1: error: bad value (native) for -march= > switch > ptlsim/build/cache/mesiCache.cpp:1: error: bad value (native) for -mtune= > switch > Compiling ==> ptlsim/build/cache/p2p.cpp > ptlsim/build/cache/p2p.cpp:1: error: bad value (native) for -march= switch > ptlsim/build/cache/p2p.cpp:1: error: bad value (native) for -mtune= switch > Compiling ==> ptlsim/build/core/branchpred.cpp > ptlsim/build/core/branchpred.cpp:1: error: bad value (native) for -march= > switch > ptlsim/build/core/branchpred.cpp:1: error: bad value (native) for -mtune= > switch > Compiling ==> ptlsim/build/core/ooocore.cpp > ptlsim/build/core/ooocore.cpp:1: error: bad value (native) for -march= > switch > ptlsim/build/core/ooocore.cpp:1: error: bad value (native) for -mtune= > switch > Compiling ==> ptlsim/build/core/oooexec.cpp > ptlsim/build/core/oooexec.cpp:1: error: bad value (native) for -march= > switch > ptlsim/build/core/oooexec.cpp:1: error: bad value (native) for -mtune= > switch > Compiling ==> ptlsim/build/core/ooopipe.cpp > ptlsim/build/core/ooopipe.cpp:1: error: bad value (native) for -march= > switch > ptlsim/build/core/ooopipe.cpp:1: error: bad value (native) for -mtune= > switch > Compiling ==> ptlsim/build/core/seqcore.cpp > ptlsim/build/core/seqcore.cpp:1: error: bad value (native) for -march= > switch > ptlsim/build/core/seqcore.cpp:1: error: bad value (native) for -mtune= > switch > Compiling ==> ptlsim/build/lib/superstl.cpp > ptlsim/build/lib/superstl.cpp:1: error: bad value (native) for -march= > switch > ptlsim/build/lib/superstl.cpp:1: error: bad value (native) for -mtune= > switch > Compiling ==> ptlsim/build/lib/statelist.cpp > ptlsim/build/lib/statelist.cpp:1: error: bad value (native) for -march= > switch > ptlsim/build/lib/statelist.cpp:1: error: bad value (native) for -mtune= > switch > Compiling ==> ptlsim/build/sim/syscalls.cpp > ptlsim/build/sim/syscalls.cpp:1: error: bad value (native) for -march= > switch > ptlsim/build/sim/syscalls.cpp:1: error: bad value (native) for -mtune= > switch > Compiling ==> ptlsim/build/sim/config-parser.cpp > ptlsim/build/sim/config-parser.cpp:1: error: bad value (native) for -march= > switch > ptlsim/build/sim/config-parser.cpp:1: error: bad value (native) for -mtune= > switch > Compiling ==> ptlsim/build/stats/datastore.cpp > ptlsim/build/stats/datastore.cpp:1: error: bad value (native) for -march= > switch > ptlsim/build/stats/datastore.cpp:1: error: bad value (native) for -mtune= > switch > Compiling ==> ptlsim/build/x86/ptlhwdef.cpp > ptlsim/build/x86/ptlhwdef.cpp:1: error: bad value (native) for -march= > switch > ptlsim/build/x86/ptlhwdef.cpp:1: error: bad value (native) for -mtune= > switch > Compiling ==> ptlsim/build/lib/bson/bson.c > cc1: warning: command line option "-fno-rtti" is valid for C++/ObjC++ but > not for C > ptlsim/build/lib/bson/bson.c:1: error: bad value (native) for -march= > switch > ptlsim/build/lib/bson/bson.c:1: error: bad value (native) for -mtune= > switch > Compiling ==> ptlsim/build/lib/bson/md5.c > cc1: warning: command line option "-fno-rtti" is valid for C++/ObjC++ but > not for C > ptlsim/build/lib/bson/md5.c:1: error: bad value (native) for -march= switch > ptlsim/build/lib/bson/md5.c:1: error: bad value (native) for -mtune= switch > Compiling ==> ptlsim/build/lib/bson/numbers.c > cc1: warning: command line option "-fno-rtti" is valid for C++/ObjC++ but > not for C > ptlsim/build/lib/bson/numbers.c:1: error: bad value (native) for -march= > switch > ptlsim/build/lib/bson/numbers.c:1: error: bad value (native) for -mtune= > switch > Compiling ==> ptlsim/build/lib/bson/mongo.c > cc1: warning: command line option "-fno-rtti" is valid for C++/ObjC++ but > not for C > ptlsim/build/lib/bson/mongo.c:1: error: bad value (native) for -march= > switch > ptlsim/build/lib/bson/mongo.c:1: error: bad value (native) for -mtune= > switch > Compiling ==> ptlsim/build/sim/ptl-qemu.cpp > ptlsim/build/sim/ptl-qemu.cpp:1: error: bad value (native) for -march= > switch > ptlsim/build/sim/ptl-qemu.cpp:1: error: bad value (native) for -mtune= > switch > Compiling ==> ptlsim/build/sim/ptlsim.cpp > ptlsim/build/sim/ptlsim.cpp:1: error: bad value (native) for -march= switch > ptlsim/build/sim/ptlsim.cpp:1: error: bad value (native) for -mtune= switch > Compiling ==> ptlsim/build/tools/ptlstats.cpp > ptlsim/build/tools/ptlstats.cpp:1: error: bad value (native) for -march= > switch > ptlsim/build/tools/ptlstats.cpp:1: error: bad value (native) for -mtune= > switch > Compiling ==> ptlsim/build/x86/decode-complex.cpp > ptlsim/build/x86/decode-complex.cpp:1: error: bad value (native) for > -march= switch > ptlsim/build/x86/decode-complex.cpp:1: error: bad value (native) for > -mtune= switch > Compiling ==> ptlsim/build/x86/decode-core.cpp > ptlsim/build/x86/decode-core.cpp:1: error: bad value (native) for -march= > switch > ptlsim/build/x86/decode-core.cpp:1: error: bad value (native) for -mtune= > switch > scons: *** [ptlsim/build/stats.i] Error 1 > Compiling ==> ptlsim/build/x86/decode-fast.cpp > scons: *** [ptlsim/build/cache/bus.o] Error 1 > scons: *** [ptlsim/build/cache/cacheController.o] Error 1 > scons: *** [ptlsim/build/cache/cpuController.o] Error 1 > scons: *** [ptlsim/build/cache/dcache.o] Error 1 > scons: *** [ptlsim/build/cache/memoryController.o] Error 1 > scons: *** [ptlsim/build/cache/memoryHierarchy.o] Error 1 > scons: *** [ptlsim/build/cache/memoryRequest.o] Error 1 > scons: *** [ptlsim/build/cache/mesiBus.o] Error 1 > scons: *** [ptlsim/build/cache/mesiCache.o] Error 1 > scons: *** [ptlsim/build/cache/p2p.o] Error 1 > scons: *** [ptlsim/build/core/branchpred.o] Error 1 > scons: *** [ptlsim/build/core/ooocore.o] Error 1 > scons: *** [ptlsim/build/core/oooexec.o] Error 1 > scons: *** [ptlsim/build/core/ooopipe.o] Error 1 > scons: *** [ptlsim/build/core/seqcore.o] Error 1 > scons: *** [ptlsim/build/lib/superstl.o] Error 1 > scons: *** [ptlsim/build/lib/statelist.o] Error 1 > scons: *** [ptlsim/build/sim/syscalls.o] Error 1 > scons: *** [ptlsim/build/sim/config-parser.o] Error 1 > scons: *** [ptlsim/build/stats/datastore.o] Error 1 > scons: *** [ptlsim/build/x86/ptlhwdef.o] Error 1 > scons: *** [ptlsim/build/lib/bson/bson.o] Error 1 > scons: *** [ptlsim/build/lib/bson/md5.o] Error 1 > scons: *** [ptlsim/build/lib/bson/numbers.o] Error 1 > scons: *** [ptlsim/build/lib/bson/mongo.o] Error 1 > scons: *** [ptlsim/build/sim/ptl-qemu.o] Error 1 > scons: *** [ptlsim/build/sim/ptlsim.o] Error 1 > scons: *** [ptlsim/build/tools/ptlstats.o] Error 1 > scons: *** [ptlsim/build/x86/decode-complex.o] Error 1 > scons: *** [ptlsim/build/x86/decode-core.o] Error 1 > ptlsim/build/x86/decode-fast.cpp:1: error: bad value (native) for -march= > switch > ptlsim/build/x86/decode-fast.cpp:1: error: bad value (native) for -mtune= > switch > scons: *** [ptlsim/build/x86/decode-fast.o] Error 1 > [root@specclient1 avadhpatel-marss-0a29c42]# > > I tried installed and ran Scons from cygwin: > > > > Here is the error > > > > ~/marss/avadhpatel-marss-5d2a2a0 > > $ scons -Q > > running with -j8 > > 2 [main] python 3304 C:\cygwin\bin\python.exe: *** fatal error - > unable to > > remap \\?\C:\cygwin\lib\python2.6\lib-dynload\_multiprocessing.dll to same > addr > > ess as parent: 0x2A0000 != 0x370000 > > Stack trace: > > Frame Function Args > > 00228478 6102796B (00228478, 00000000, 00000000, 00000000) > > 00228768 6102796B (6117EC60, 00008000, 00000000, 61180977) > > 00229798 61004F1B (611A7FAC, 6124A83C, 002A0000, 00370000) > > End of stack trace > > 4 [main] python 7616 fork: child 3304 - died waiting for dll loading, > errn > > o 11 > > 102564 [main] python 7552 C:\cygwin\bin\python.exe: *** fatal error - > unable to > > remap \\?\C:\cygwin\lib\python2.6\lib-dynload\_multiprocessing.dll to same > addr > > ess as parent: 0x2A0000 != 0x370000 > > Stack trace: > > Frame Function Args > > 00228478 6102796B (00228478, 00000000, 00000000, 00000000) > > 00228768 6102796B (6117EC60, 00008000, 00000000, 61180977) > > 00229798 61004F1B (611A7FAC, 6124A83C, 002A0000, 00370000) > > End of stack trace > > 116374 [main] python 7616 fork: child 7552 - died waiting for dll loading, > errn > > o 11 > > 1245353 [main] python 6052 C:\cygwin\bin\python.exe: *** fatal error - > unable to > > remap \\?\C:\cygwin\lib\python2.6\lib-dynload\_multiprocessing.dll to same > addr > > ess as parent: 0x2A0000 != 0x370000 > > Stack trace: > > Frame Function Args > > 00227A38 6102796B (00227A38, 00000000, 00000000, 00000000) > > 00227D28 6102796B (6117EC60, 00008000, 00000000, 61180977) > > 00228D58 61004F1B (611A7FAC, 6124A83C, 002A0000, 00370000) > > End of stack trace > > 1258223 [main] python 7616 fork: child 6052 - died waiting for dll loading, > errn > > o 11 > > 1604931 [main] python 4216 C:\cygwin\bin\python.exe: *** fatal error - > unable to > > remap \\?\C:\cygwin\lib\python2.6\lib-dynload\_multiprocessing.dll to same > addr > > ess as parent: 0x2A0000 != 0x370000 > > Stack trace: > > Frame Function Args > > 002255A8 6102796B (002255A8, 00000000, 00000000, 00000000) > > 00225898 6102796B (6117EC60, 00008000, 00000000, 61180977) > > 002268C8 61004F1B (611A7FAC, 6124A83C, 002A0000, 00370000) > > End of stack trace > > 1619003 [main] python 7616 fork: child 4216 - died waiting for dll loading, > errn > > o 11 > > ERROR: You do not have g++ compiler installed!! > > > > scons: *** Return of non-existent variable ''ret_str'' > > File "/home/PCao/marss/avadhpatel-marss-5d2a2a0/qemu/SConfigure", line 30, > in ex > > it_with_error > > > > ~/marss/avadhpatel-marss-5d2a2a0 > > $ > > > > > > --- On *Tue, 5/10/11, DRAM Ninjas <[email protected]>* wrote: > > > From: DRAM Ninjas <[email protected]> > Subject: Re: [marss86-devel] marss on cygwin windows Vista ? > To: [email protected] > > Cc: [email protected] > Date: Tuesday, May 10, 2011, 2:00 PM > > erm, sorry ... I meant cygwin shell, not the scons shell ... > > On Tue, May 10, 2011 at 3:00 PM, DRAM Ninjas > <[email protected]<http://us.mc1302.mail.yahoo.com/mc/[email protected]>> > wrote: > > I would see if you can install scons via the cygwin installer and try it > within the scons shell (the bash shell, not the cmd prompt). > > > > Perhaps this will fix some of your issues? > > > > On Tue, May 10, 2011 at 12:57 PM, <[email protected]> wrote: > > Thanks for the pointer. > > > > I use scons outside of cygwin (start -> cmd) > > > > > > --- On *Tue, 5/10/11, DRAM Ninjas > <[email protected]<http://us.mc1302.mail.yahoo.com/mc/[email protected]> > >* wrote: > > > From: DRAM Ninjas > <[email protected]<http://us.mc1302.mail.yahoo.com/mc/[email protected]> > > > Subject: Re: [marss86-devel] marss on cygwin windows Vista ? > To: [email protected] > > Cc: > [email protected]<http://us.mc1302.mail.yahoo.com/mc/[email protected]> > Date: Tuesday, May 10, 2011, 10:11 AM > > > > The recommended OS for marss is Linux. I highly doubt you could do this on > Windows, in all honesty (that's not to say I've tried, but just from general > past experience with Cygwin). > > > > Are you using scons from the cygwin repositories? It seems like it's > failing before it can even execute scons ... > > On Mon, May 9, 2011 at 3:44 PM, [email protected] wrote: > > Hi, > > > > I'm trying to set up marss on cygwin windows Vista, run into problem > with "unable to remap". > > > > Have tried rebase and rebaseall and run into various errors ? > > > > What is the recommend OS for marss software ? Have anyone run it > successfully on cygwin Vista ? > > > > many thanks > > Paul > > > > $ scons -Q > running with -j8 > 3 [main] python 6244 C:\cygwin\bin\python.exe: *** fatal error - > unable to > remap \\?\C:\cygwin\lib\python2.6\lib-dynload\_random.dll to same address > as pa > rent: 0x370000 != 0x3F0000 > Stack trace: > Frame Function Args > 00228478 6102796B (00228478, 00000000, 00000000, 00000000) > 00228768 6102796B (6117EC60, 00008000, 00000000, 61180977) > 00229798 61004F1B (611A7FAC, 61249BC4, 00370000, 003F0000) > End of stack trace > 4 [main] python 7844 fork: child 6244 - died waiting for dll loading, > errn > o 11 > 2168764 [main] python 2168 C:\cygwin\bin\python.exe: *** fatal error - > unable to > remap \\?\C:\cygwin\lib\python2.6\lib-dynload\_random.dll to same address > as pa > rent: 0x370000 != 0x3F0000 > Stack trace: > Frame Function Args > 00227A38 6102796B (00227A38, 00000000, 00000000, 00000000) > 00227D28 6102796B (6117EC60, 00008000, 00000000, 61180977) > 00228D58 61004F1B (611A7FAC, 61249BC4, 00370000, 003F0000) > End of stack trace > 2228080 [main] python 7844 fork: child 2168 - died waiting for dll loading, > errn > o 11 > 5005287 [main] python 7488 C:\cygwin\bin\python.exe: *** fatal error - > unable to > remap \\?\C:\cygwin\lib\python2.6\lib-dynload\_random.dll to same address > as pa > rent: 0x370000 != 0x3C0000 > Stack trace: > Frame Function Args > 00226528 6102796B (00226528, 00000000, 00000000, 00000000) > 00226818 6102796B (6117EC60, 00008000, 00000000, 61180977) > 00227848 61004F1B (611A7FAC, 61249BC4, 00370000, 003C0000) > End of stack trace > 5019284 [main] python 7844 fork: child 7488 - died waiting for dll loading, > errn > o 11 > ERROR: g++ either does not exist or does not work > > scons: *** Return of non-existent variable ''ret_str'' > File "/home/PCao/marss/avadhpatel-marss-5d2a2a0/qemu/SConfigure", line 30, > in ex > it_with_error > > > _______________________________________________ > http://www.marss86.org > Marss86-Devel mailing list > [email protected]<http://us.mc1302.mail.yahoo.com/mc/[email protected]> > https://www.cs.binghamton.edu/mailman/listinfo/marss86-devel > > > > > -------------- next part -------------- An HTML attachment was scrubbed... URL: <https://www.cs.binghamton.edu/mailman/private/marss86-devel/attachments/20110511/6d277a75/attachment-0001.html> ------------------------------ Message: 2 Date: Thu, 12 May 2011 04:45:18 +0000 From: Hui Kang <[email protected]> To: "<[email protected]> <[email protected]>" <[email protected]> Subject: [marss86-devel] Switch to native mode and ptlstat results Message-ID: <[email protected]> Content-Type: text/plain; charset="us-ascii" Hi, I have a question about switching to the native QEMU mode after simulation. Suppose I have this simulation command simconfig -run -stopinsns 10m -stats my.stats It seems that I must add -kill-after-run to make the ptlsim stop completely such that ptlstat can be used to convert the stats file to a readable format. However, -kill-after-run will kill the QEMU process. So how can I switch to native mode in QEMU, as well as have the readable pltstat results. Thanks. - Hui -------------- next part -------------- An HTML attachment was scrubbed... URL: <https://www.cs.binghamton.edu/mailman/private/marss86-devel/attachments/20110512/7444d4c1/attachment-0001.html> ------------------------------ Message: 3 Date: Thu, 12 May 2011 05:21:08 +0000 From: Hui Kang <[email protected]> To: Erdem Aktas <[email protected]> Cc: "<[email protected]> <[email protected]>" <[email protected]> Subject: Re: [marss86-devel] Switch to native mode and ptlstat results Message-ID: <[email protected]> Content-Type: text/plain; charset="us-ascii" Hi, Erdem I do not think qemu switches back to the native mode after the simulation completes, even without -kill-after-run. The console has no response to my input. - Hui On May 12, 2011, at 1:16 AM, Erdem Aktas wrote: > Hi Hui; > > After the simulation stops , in your example after 10m instructions, > it will switch back to the native mode but I guess you don't need to > go back to the native mode. Just use -kill-after-run parameter. stats > files will be in the host machine where you run the qemu-system-x86_64 > command. > > -Erdem > > On Thu, May 12, 2011 at 12:45 AM, Hui Kang <[email protected]> wrote: >> Hi, >> I have a question about switching to the native QEMU mode after simulation. >> Suppose I have this simulation command >> simconfig -run -stopinsns 10m -stats my.stats >> It seems that I must add -kill-after-run to make the ptlsim stop completely >> such that ptlstat can be used to convert the stats file to a readable >> format. However, -kill-after-run will kill the QEMU process. >> So how can I switch to native mode in QEMU, as well as have the readable >> pltstat results. Thanks. >> - Hui >> _______________________________________________ >> http://www.marss86.org >> Marss86-Devel mailing list >> [email protected] >> https://www.cs.binghamton.edu/mailman/listinfo/marss86-devel >> >> ------------------------------ Message: 4 Date: Thu, 12 May 2011 01:46:07 -0400 From: Erdem Aktas <[email protected]> To: Hui Kang <[email protected]> Cc: [email protected] Subject: Re: [marss86-devel] Switch to native mode and ptlstat results Message-ID: <[email protected]> Content-Type: text/plain; charset=ISO-8859-1 Hi Hui; I just downloaded the marss and qemu image from the marss web page and tried and after 10m instructions (in case you gave -stopinsns 10m) I am able to use the QEMU console. But when you use -kill-after-run, you don't need to go back to the QEMU console. QEMU will be killed and you will have the stat files at the host machine. If you are still having the problems, could you write: 1- Are you using the latest version of marss 2- How do you run the qemu? 3-Which image are you running? 4-Are you running any benchmark in the guest OS? 5-Are you trying to go back to native mode or just trying to get the stat files? -Erdem On Thu, May 12, 2011 at 1:21 AM, Hui Kang <[email protected]> wrote: > Hi, Erdem > I do not think qemu switches back to the native mode after the simulation > completes, even without -kill-after-run. > The console has no response to my input. > > - Hui > > > On May 12, 2011, at 1:16 AM, Erdem Aktas wrote: > >> Hi Hui; >> >> After the simulation stops , in your example after 10m instructions, >> it will switch back to the native mode but I guess you don't need to >> go back to the native mode. Just use -kill-after-run parameter. stats >> files will be in the host machine where you run the qemu-system-x86_64 >> command. >> >> -Erdem >> >> On Thu, May 12, 2011 at 12:45 AM, Hui Kang <[email protected]> wrote: >>> Hi, >>> I have a question about switching to the native QEMU mode after simulation. >>> Suppose I have this simulation command >>> ? ? ? ? simconfig -run -stopinsns 10m -stats my.stats >>> It seems that I must add -kill-after-run to make the ptlsim stop completely >>> such that ptlstat can be used to convert the stats file to a readable >>> format. However, -kill-after-run will kill the QEMU process. >>> So how can I switch to native mode in QEMU, as well as have the readable >>> pltstat results. Thanks. >>> - Hui >>> _______________________________________________ >>> http://www.marss86.org >>> Marss86-Devel mailing list >>> [email protected] >>> https://www.cs.binghamton.edu/mailman/listinfo/marss86-devel >>> >>> > > ------------------------------ Message: 5 Date: Thu, 12 May 2011 13:59:11 +0000 From: Hui Kang <[email protected]> To: Erdem Aktas <[email protected]> Cc: "<[email protected]>" <[email protected]> Subject: Re: [marss86-devel] Switch to native mode and ptlstat results Message-ID: <[email protected]> Content-Type: text/plain; charset="us-ascii" On May 12, 2011, at 1:46 AM, Erdem Aktas wrote: Hi Hui; I just downloaded the marss and qemu image from the marss web page and tried and after 10m instructions (in case you gave -stopinsns 10m) I am able to use the QEMU console. But when you use -kill-after-run, you don't need to go back to the QEMU console. QEMU will be killed and you will have the stat files at the host machine. es, using -kill-after-run kills the QEMU process. However, I would like to switch to the native mode of the Qemu VM. And if I do not use the -kill-after-run option, Qemu does not quit and I can use ctrl-alt-1 to enter the VM mode, which seems stop. Is there any simconfig command to resume the VM? If you are still having the problems, could you write: 1- Are you using the latest version of marss Yes, from the git repository. 2- How do you run the qemu? qemu/qemu-system-x86_64 -m 256 -hda parsec_qcow.img 3-Which image are you running? parsec_image.tar.bz2<http://www.marss86.org/downloads/parsec_image.tar.bz2> 4-Are you running any benchmark in the guest OS? No, just do "vmstat 5" 5-Are you trying to go back to native mode or just trying to get the stat files? Actually I want to do both: go back to native mode, which should continue running and get the stat files. Thanks. - Hui -Erdem On Thu, May 12, 2011 at 1:21 AM, Hui Kang <[email protected]<mailto:[email protected]>> wrote: Hi, Erdem I do not think qemu switches back to the native mode after the simulation completes, even without -kill-after-run. The console has no response to my input. - Hui On May 12, 2011, at 1:16 AM, Erdem Aktas wrote: Hi Hui; After the simulation stops , in your example after 10m instructions, it will switch back to the native mode but I guess you don't need to go back to the native mode. Just use -kill-after-run parameter. stats files will be in the host machine where you run the qemu-system-x86_64 command. -Erdem On Thu, May 12, 2011 at 12:45 AM, Hui Kang <[email protected]<mailto:[email protected]>> wrote: Hi, I have a question about switching to the native QEMU mode after simulation. Suppose I have this simulation command simconfig -run -stopinsns 10m -stats my.stats It seems that I must add -kill-after-run to make the ptlsim stop completely such that ptlstat can be used to convert the stats file to a readable format. However, -kill-after-run will kill the QEMU process. So how can I switch to native mode in QEMU, as well as have the readable pltstat results. Thanks. - Hui _______________________________________________ http://www.marss86.org Marss86-Devel mailing list [email protected]<mailto:[email protected]> https://www.cs.binghamton.edu/mailman/listinfo/marss86-devel -------------- next part -------------- An HTML attachment was scrubbed... URL: <https://www.cs.binghamton.edu/mailman/private/marss86-devel/attachments/20110512/21126bb7/attachment-0001.html> ------------------------------ Message: 6 Date: Thu, 12 May 2011 16:47:06 +0200 From: Carlos Jos? Martinez Gavil?n <[email protected]> To: [email protected] Subject: [marss86-devel] consumption cpu and associatitity caches Message-ID: <[email protected]> Content-Type: text/plain; charset="iso-8859-1" Hi everyone, Where I can find the consumption of each CPU at the end of a simulation? where I can change the L2 cache associativity? Thanks a lot. -- Carlos J. Mart?nez -------------- next part -------------- An HTML attachment was scrubbed... URL: <https://www.cs.binghamton.edu/mailman/private/marss86-devel/attachments/20110512/0b24b67d/attachment-0001.html> ------------------------------ Message: 7 Date: Thu, 12 May 2011 15:12:29 +0000 From: Hui Kang <[email protected]> To: Erdem Aktas <[email protected]> Cc: "<[email protected]>" <[email protected]> Subject: Re: [marss86-devel] Switch to native mode and ptlstat results Message-ID: <[email protected]> Content-Type: text/plain; charset="us-ascii" Hi, Erdem By the way, when I add -kill-after-run, I can get the stats output successfully. However, I have to re-launch the Qemu again, which is not very convenient. - Hui On May 12, 2011, at 1:46 AM, Erdem Aktas wrote: > Hi Hui; > > I just downloaded the marss and qemu image from the marss web page and > tried and after 10m instructions (in case you gave -stopinsns 10m) I > am able to use the QEMU console. But when you use -kill-after-run, you > don't need to go back to the QEMU console. QEMU will be killed and you > will have the stat files at the host machine. > > If you are still having the problems, could you write: > 1- Are you using the latest version of marss > 2- How do you run the qemu? > 3-Which image are you running? > 4-Are you running any benchmark in the guest OS? > 5-Are you trying to go back to native mode or just trying to get the stat > files? > > -Erdem > > On Thu, May 12, 2011 at 1:21 AM, Hui Kang <[email protected]> wrote: >> Hi, Erdem >> I do not think qemu switches back to the native mode after the simulation >> completes, even without -kill-after-run. >> The console has no response to my input. >> >> - Hui >> >> >> On May 12, 2011, at 1:16 AM, Erdem Aktas wrote: >> >>> Hi Hui; >>> >>> After the simulation stops , in your example after 10m instructions, >>> it will switch back to the native mode but I guess you don't need to >>> go back to the native mode. Just use -kill-after-run parameter. stats >>> files will be in the host machine where you run the qemu-system-x86_64 >>> command. >>> >>> -Erdem >>> >>> On Thu, May 12, 2011 at 12:45 AM, Hui Kang <[email protected]> wrote: >>>> Hi, >>>> I have a question about switching to the native QEMU mode after simulation. >>>> Suppose I have this simulation command >>>> simconfig -run -stopinsns 10m -stats my.stats >>>> It seems that I must add -kill-after-run to make the ptlsim stop completely >>>> such that ptlstat can be used to convert the stats file to a readable >>>> format. However, -kill-after-run will kill the QEMU process. >>>> So how can I switch to native mode in QEMU, as well as have the readable >>>> pltstat results. Thanks. >>>> - Hui >>>> _______________________________________________ >>>> http://www.marss86.org >>>> Marss86-Devel mailing list >>>> [email protected] >>>> https://www.cs.binghamton.edu/mailman/listinfo/marss86-devel >>>> >>>> >> >> ------------------------------ Message: 8 Date: Thu, 12 May 2011 11:18:23 -0400 From: DRAM Ninjas <[email protected]> To: Hui Kang <[email protected]> Cc: "<[email protected]>" <[email protected]> Subject: Re: [marss86-devel] Switch to native mode and ptlstat results Message-ID: <[email protected]> Content-Type: text/plain; charset="iso-8859-1" You can mitigate this inconvenience to some extent by using the checkpointing scheme. Basically you can create a checkpoint right before the point where marss kicks into simulation mode. This way, you don't have to wait for the machine to boot up and you don't have to re-type the command at the command line. On Thu, May 12, 2011 at 11:12 AM, Hui Kang <[email protected]> wrote: > Hi, Erdem > By the way, when I add -kill-after-run, I can get the stats output > successfully. However, I have to re-launch the Qemu again, which is not > very convenient. > > - Hui > > On May 12, 2011, at 1:46 AM, Erdem Aktas wrote: > > > Hi Hui; > > > > I just downloaded the marss and qemu image from the marss web page and > > tried and after 10m instructions (in case you gave -stopinsns 10m) I > > am able to use the QEMU console. But when you use -kill-after-run, you > > don't need to go back to the QEMU console. QEMU will be killed and you > > will have the stat files at the host machine. > > > > If you are still having the problems, could you write: > > 1- Are you using the latest version of marss > > 2- How do you run the qemu? > > 3-Which image are you running? > > 4-Are you running any benchmark in the guest OS? > > 5-Are you trying to go back to native mode or just trying to get the stat > files? > > > > -Erdem > > > > On Thu, May 12, 2011 at 1:21 AM, Hui Kang <[email protected]> > wrote: > >> Hi, Erdem > >> I do not think qemu switches back to the native mode after the > simulation completes, even without -kill-after-run. > >> The console has no response to my input. > >> > >> - Hui > >> > >> > >> On May 12, 2011, at 1:16 AM, Erdem Aktas wrote: > >> > >>> Hi Hui; > >>> > >>> After the simulation stops , in your example after 10m instructions, > >>> it will switch back to the native mode but I guess you don't need to > >>> go back to the native mode. Just use -kill-after-run parameter. stats > >>> files will be in the host machine where you run the qemu-system-x86_64 > >>> command. > >>> > >>> -Erdem > >>> > >>> On Thu, May 12, 2011 at 12:45 AM, Hui Kang <[email protected]> > wrote: > >>>> Hi, > >>>> I have a question about switching to the native QEMU mode after > simulation. > >>>> Suppose I have this simulation command > >>>> simconfig -run -stopinsns 10m -stats my.stats > >>>> It seems that I must add -kill-after-run to make the ptlsim stop > completely > >>>> such that ptlstat can be used to convert the stats file to a readable > >>>> format. However, -kill-after-run will kill the QEMU process. > >>>> So how can I switch to native mode in QEMU, as well as have the > readable > >>>> pltstat results. Thanks. > >>>> - Hui > >>>> _______________________________________________ > >>>> 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 > -------------- next part -------------- An HTML attachment was scrubbed... URL: <https://www.cs.binghamton.edu/mailman/private/marss86-devel/attachments/20110512/e2e0027e/attachment-0001.html> ------------------------------ _______________________________________________ http://www.marss86.org Marss86-Devel mailing list [email protected] https://www.cs.binghamton.edu/mailman/listinfo/marss86-devel End of Marss86-Devel Digest, Vol 15, Issue 10 ********************************************* _______________________________________________ http://www.marss86.org Marss86-Devel mailing list [email protected] https://www.cs.binghamton.edu/mailman/listinfo/marss86-devel
