I just wanted to make everyone aware that we're holding a tutorial again at ASPLOS this year (Sunday 3/2) in the morning. We're planning to provide public access to the M5 repository in conjunction with the tutorial. We have some exciting new features in v2.0, including: - New cache model - More abstract cache coherence that is general enough to allow multi- level configurations - Inter-object protocol for memory-system objects - Python creation of object parameters - Support for ARM, SPARC, and MIPS - Performance improvements - Creating and using Simpoints Even if you came to last year's tutorial, you can come again to learn about these new features. We'll also be discussing: - M5 structure - Specifying configurations - Running experiments - Statistics - Debugging techniques - Future development directions So, if you're new to M5, you're going to learn enough to get started. Our tutorial page: http://www.m5sim.org/wiki/index.php/ASPLOS_2008 ASPLOS web page (registration, schedule, etc.): http://research.microsoft.com/asplos08/ Hope to see you there! Ali

Any new stuff going to appear much before ASPLOS? I wouldn't mind exercising my finely honed skill at breaking things to extend testing ... (I won't be going to ASPLOS so don't worry about spoiling it for me.) On Feb 6, 2008 2:51 PM, Ali Saidi <saidi(a)umich.edu> wrote: > I just wanted to make everyone aware that we're holding a tutorial > again at ASPLOS this year (Sunday 3/2) in the morning. We're planning > to provide public access to the M5 repository in conjunction with the > tutorial. > > We have some exciting new features in v2.0, including: > > - New cache model > - More abstract cache coherence that is general enough to allow multi- > level configurations > - Inter-object protocol for memory-system objects > - Python creation of object parameters > - Support for ARM, SPARC, and MIPS > - Performance improvements > - Creating and using Simpoints > > Even if you came to last year's tutorial, you can come again to learn > about these new features. > > We'll also be discussing: > - M5 structure > - Specifying configurations > - Running experiments > - Statistics > - Debugging techniques > - Future development directions > > So, if you're new to M5, you're going to learn enough to get started. > > Our tutorial page: > http://www.m5sim.org/wiki/index.php/ASPLOS_2008 > > ASPLOS web page (registration, schedule, etc.): > http://research.microsoft.com/asplos08/ > > Hope to see you there! > > Ali > _______________________________________________ > m5-users mailing list > m5-users(a)m5sim.org > http://m5sim.org/cgi-bin/mailman/listinfo/m5-users > -- Philip Machanick (Visiting Fellow, School of ITEE, University of Queensland) 39 Cunningham Street, Taringa, Qld 4068, Australia http://opinion-nation.blogspot.com/ +61-7-3871-0963 (office 3365-1190) skype philipmach

Another question regarding a Formula that is based on a Counter's final value. For example: Counter c; Formula f; regStats() { f = scalar(c)/constant(1000); } However, type scalar doesn't exist and Scalar doesn't work either. Can anyone give me some hints? I am following the documentation on the statistics website, which says " If you want to use the value of bar or num at the moment the formula is calculated (i.e. the end), use the scalar() castIf you want to use the value of bar or num at the moment the formula is calculated (i.e. the end), use the scalar() cast " Thanks for helping! Jiayuan _________________________________________________________________ Need to know the score, the latest news, or you need your Hotmail®-get your "fix". http://www.msnmobilefix.com/Default.aspx

Wow, I had to dig back in the history to figure out how this worked. scalar() itself was removed in May 2004, so I guess we should update the documentation. I did create a stat type called Value for sucking in C++ variables (though, it's use is discouraged.) You can see it used in src/sim/stat_control.cc It's much better to create a statistic type and use that. (Especially if you want to do your binning stuff.) If there is just a variable initialized though, you can use it how we do for the frequency and such. Do you mind updating the documentation as you work on this stuff? Thanks, Nate On Feb 5, 2008 10:48 PM, jiayuan meng <jerrygonair(a)hotmail.com> wrote: > > > Another question regarding a Formula that is based on a Counter's final > value. For example: > > Counter c; > Formula f; > > regStats() > { > f = scalar(c)/constant(1000); > } > > However, type scalar doesn't exist and Scalar doesn't work either. Can > anyone give me some hints? > > I am following the documentation on the statistics website, which says > " > If you want to use the value of bar or num at the moment the formula is > calculated (i.e. the end), use the scalar() castIf you want to use the value > of bar or num at the moment the formula is calculated (i.e. the end), use > the scalar() cast > " > > Thanks for helping! > > Jiayuan > > ________________________________ > Need to know the score, the latest news, or you need your Hotmail(R)-get your > "fix". Check it out. > > _______________________________________________ > m5-users mailing list > m5-users(a)m5sim.org > http://m5sim.org/cgi-bin/mailman/listinfo/m5-users >

Wow, thanks for the mid-night reply! Yeah, I just updated the formula session in the statistics documantation page. Thanks! Jiayuan ----- Original Message ----- From: "nathan binkert" <nate(a)binkert.org> To: "M5 users mailing list" <m5-users(a)m5sim.org> Sent: 2008年2月6日 3:03 PM Subject: Re: [m5-users] does the scalar() cast exist for Formula calculation? > Wow, I had to dig back in the history to figure out how this worked. > scalar() itself was removed in May 2004, so I guess we should update > the documentation. I did create a stat type called Value for sucking > in C++ variables (though, it's use is discouraged.) > > You can see it used in src/sim/stat_control.cc > > It's much better to create a statistic type and use that. (Especially > if you want to do your binning stuff.) If there is just a variable > initialized though, you can use it how we do for the frequency and > such. > > Do you mind updating the documentation as you work on this stuff? > > Thanks, > > Nate > > On Feb 5, 2008 10:48 PM, jiayuan meng <jerrygonair(a)hotmail.com> wrote: >> >> >> Another question regarding a Formula that is based on a Counter's final >> value. For example: >> >> Counter c; >> Formula f; >> >> regStats() >> { >> f = scalar(c)/constant(1000); >> } >> >> However, type scalar doesn't exist and Scalar doesn't work either. Can >> anyone give me some hints? >> >> I am following the documentation on the statistics website, which says >> " >> If you want to use the value of bar or num at the moment the formula is >> calculated (i.e. the end), use the scalar() castIf you want to use the value >> of bar or num at the moment the formula is calculated (i.e. the end), use >> the scalar() cast >> " >> >> Thanks for helping! >> >> Jiayuan >> >> ________________________________ >> Need to know the score, the latest news, or you need your Hotmail(R)-get your >> "fix". Check it out. >> >> _______________________________________________ >> m5-users mailing list >> m5-users(a)m5sim.org >> http://m5sim.org/cgi-bin/mailman/listinfo/m5-users >> > _______________________________________________ > m5-users mailing list > m5-users(a)m5sim.org > http://m5sim.org/cgi-bin/mailman/listinfo/m5-users >

Thanks a lot. The statistics package is on my long list of things to clean up. I know what I want to do, I'm just not fast enough to get it done. Nate On Feb 6, 2008 5:39 AM, Jiayuan Meng <jerrygonair(a)hotmail.com> wrote: > Wow, thanks for the mid-night reply! > Yeah, I just updated the formula session in the statistics documantation page. > > Thanks! > > Jiayuan > > > ----- Original Message ----- > From: "nathan binkert" <nate(a)binkert.org> > To: "M5 users mailing list" <m5-users(a)m5sim.org> > Sent: 2008年2月6日 3:03 PM > Subject: Re: [m5-users] does the scalar() cast exist for Formula calculation? > > > > Wow, I had to dig back in the history to figure out how this worked. > > scalar() itself was removed in May 2004, so I guess we should update > > the documentation. I did create a stat type called Value for sucking > > in C++ variables (though, it's use is discouraged.) > > > > You can see it used in src/sim/stat_control.cc > > > > It's much better to create a statistic type and use that. (Especially > > if you want to do your binning stuff.) If there is just a variable > > initialized though, you can use it how we do for the frequency and > > such. > > > > Do you mind updating the documentation as you work on this stuff? > > > > Thanks, > > > > Nate > > > > On Feb 5, 2008 10:48 PM, jiayuan meng <jerrygonair(a)hotmail.com> wrote: > >> > >> > >> Another question regarding a Formula that is based on a Counter's final > >> value. For example: > >> > >> Counter c; > >> Formula f; > >> > >> regStats() > >> { > >> f = scalar(c)/constant(1000); > >> } > >> > >> However, type scalar doesn't exist and Scalar doesn't work either. Can > >> anyone give me some hints? > >> > >> I am following the documentation on the statistics website, which says > >> " > >> If you want to use the value of bar or num at the moment the formula is > >> calculated (i.e. the end), use the scalar() castIf you want to use the value > >> of bar or num at the moment the formula is calculated (i.e. the end), use > >> the scalar() cast > >> " > >> > >> Thanks for helping! > >> > >> Jiayuan > >> > >> ________________________________ > >> Need to know the score, the latest news, or you need your Hotmail(R)-get your > >> "fix". Check it out. > >> > >> _______________________________________________ > >> m5-users mailing list > >> m5-users(a)m5sim.org > >> http://m5sim.org/cgi-bin/mailman/listinfo/m5-users > >> > > _______________________________________________ > > m5-users mailing list > > m5-users(a)m5sim.org > > http://m5sim.org/cgi-bin/mailman/listinfo/m5-users > > > _______________________________________________ > m5-users mailing list > m5-users(a)m5sim.org > http://m5sim.org/cgi-bin/mailman/listinfo/m5-users >

I just wanted to remind everyone about the tutorial at ASPLOS. We're about two weeks away now. Thanks, Ali On Feb 5, 2008, at 11:51 PM, Ali Saidi wrote: > I just wanted to make everyone aware that we're holding a tutorial > again at ASPLOS this year (Sunday 3/2) in the morning. We're > planning to provide public access to the M5 repository in > conjunction with the tutorial. > > We have some exciting new features in v2.0, including: > > - New cache model > - More abstract cache coherence that is general enough to allow > multi-level configurations > - Inter-object protocol for memory-system objects > - Python creation of object parameters > - Support for ARM, SPARC, and MIPS > - Performance improvements > - Creating and using Simpoints > > Even if you came to last year's tutorial, you can come again to > learn about these new features. > > We'll also be discussing: > - M5 structure > - Specifying configurations > - Running experiments > - Statistics > - Debugging techniques > - Future development directions > > So, if you're new to M5, you're going to learn enough to get started. > > Our tutorial page: > http://www.m5sim.org/wiki/index.php/ASPLOS_2008 > > ASPLOS web page (registration, schedule, etc.): > http://research.microsoft.com/asplos08/ > > Hope to see you there! > > Ali > _______________________________________________ > m5-users mailing list > m5-users(a)m5sim.org > http://m5sim.org/cgi-bin/mailman/listinfo/m5-users >

By the way, I read that someone requested for this tutorial to be recorded. Are there any plans for this? This will be very helpful for us (me and my colleagues who plan to use m5) since we cannot go to ASPLOS for a variety of reasons (primarily and frankly, difficulty for applying for a US visa coming from a third world country and the cost of going to Seattle, probably equal to at least six months of what we get for a round trip). On 15/02/2008, Ali Saidi <saidi(a)umich.edu> wrote: > > I just wanted to remind everyone about the tutorial at ASPLOS. We're > about two weeks away now. > > Thanks, > > Ali > > > On Feb 5, 2008, at 11:51 PM, Ali Saidi wrote: > > > I just wanted to make everyone aware that we're holding a tutorial > > again at ASPLOS this year (Sunday 3/2) in the morning. We're > > planning to provide public access to the M5 repository in > > conjunction with the tutorial. > > > > We have some exciting new features in v2.0, including: > > > > - New cache model > > - More abstract cache coherence that is general enough to allow > > multi-level configurations > > - Inter-object protocol for memory-system objects > > - Python creation of object parameters > > - Support for ARM, SPARC, and MIPS > > - Performance improvements > > - Creating and using Simpoints > > > > Even if you came to last year's tutorial, you can come again to > > learn about these new features. > > > > We'll also be discussing: > > - M5 structure > > - Specifying configurations > > - Running experiments > > - Statistics > > - Debugging techniques > > - Future development directions > > > > So, if you're new to M5, you're going to learn enough to get started. > > > > Our tutorial page: > > http://www.m5sim.org/wiki/index.php/ASPLOS_2008 > > > > ASPLOS web page (registration, schedule, etc.): > > http://research.microsoft.com/asplos08/ > > > > Hope to see you there! > > > > Ali > > _______________________________________________ > > m5-users mailing list > > m5-users(a)m5sim.org > > http://m5sim.org/cgi-bin/mailman/listinfo/m5-users > > > > _______________________________________________ > m5-users mailing list > m5-users(a)m5sim.org > http://m5sim.org/cgi-bin/mailman/listinfo/m5-users >

I agree that this is a great idea. Perhaps this is something you could ask ASPLOS organizers (the steering committee). -- Sasha On Fri, 15 Feb 2008, Allan Noriel Estrella wrote: > By the way, I read that someone requested for this tutorial to be recorded. > Are there any plans for this? This will be very helpful for us (me and my > colleagues who plan to use m5) since we cannot go to ASPLOS for a variety of > reasons (primarily and frankly, difficulty for applying for a US visa coming > from a third world country and the cost of going to Seattle, probably equal > to at least six months of what we get for a round trip). > > On 15/02/2008, Ali Saidi <saidi(a)umich.edu> wrote: >> >> I just wanted to remind everyone about the tutorial at ASPLOS. We're >> about two weeks away now. >> >> Thanks, >> >> Ali >> >> >> On Feb 5, 2008, at 11:51 PM, Ali Saidi wrote: >> >>> I just wanted to make everyone aware that we're holding a tutorial >>> again at ASPLOS this year (Sunday 3/2) in the morning. We're >>> planning to provide public access to the M5 repository in >>> conjunction with the tutorial. >>> >>> We have some exciting new features in v2.0, including: >>> >>> - New cache model >>> - More abstract cache coherence that is general enough to allow >>> multi-level configurations >>> - Inter-object protocol for memory-system objects >>> - Python creation of object parameters >>> - Support for ARM, SPARC, and MIPS >>> - Performance improvements >>> - Creating and using Simpoints >>> >>> Even if you came to last year's tutorial, you can come again to >>> learn about these new features. >>> >>> We'll also be discussing: >>> - M5 structure >>> - Specifying configurations >>> - Running experiments >>> - Statistics >>> - Debugging techniques >>> - Future development directions >>> >>> So, if you're new to M5, you're going to learn enough to get started. >>> >>> Our tutorial page: >>> http://www.m5sim.org/wiki/index.php/ASPLOS_2008 >>> >>> ASPLOS web page (registration, schedule, etc.): >>> http://research.microsoft.com/asplos08/ >>> >>> Hope to see you there! >>> >>> Ali >>> _______________________________________________ >>> m5-users mailing list >>> m5-users(a)m5sim.org >>> http://m5sim.org/cgi-bin/mailman/listinfo/m5-users >>> >> >> _______________________________________________ >> m5-users mailing list >> m5-users(a)m5sim.org >> http://m5sim.org/cgi-bin/mailman/listinfo/m5-users >> >

Just to let everyone know we did record the tutorial and hope to make it available in some form soon. Right now we just have 40GB of raw video that needs to be edited and re-compressed into a more useful form. I think there are probably two options: (1) Extract the audio from the presentation and add audio to a PDF of the slides; (2) edit/ re-compress the video+audio. The video capture doesn't capture much more than the projection of the slides on the screen so I'm leaning toward (1), but if anyone has some comments I would like to hear them. Also, if anyone would be willing to volunteer to do either options please let me know as I've never done either so it make take some time. Thanks, Ali On Feb 14, 2008, at 1:11 PM, Alexandra (Sasha) Fedorova wrote: > I agree that this is a great idea. Perhaps this is something you > could ask ASPLOS organizers (the steering committee). > > -- Sasha > > On Fri, 15 Feb 2008, Allan Noriel Estrella wrote: > >> By the way, I read that someone requested for this tutorial to be >> recorded. >> Are there any plans for this? This will be very helpful for us (me >> and my >> colleagues who plan to use m5) since we cannot go to ASPLOS for a >> variety of >> reasons (primarily and frankly, difficulty for applying for a US >> visa coming >> from a third world country and the cost of going to Seattle, >> probably equal >> to at least six months of what we get for a round trip). >> >> On 15/02/2008, Ali Saidi <saidi(a)umich.edu> wrote: >>> >>> I just wanted to remind everyone about the tutorial at ASPLOS. We're >>> about two weeks away now. >>> >>> Thanks, >>> >>> Ali >>> >>> >>> On Feb 5, 2008, at 11:51 PM, Ali Saidi wrote: >>> >>>> I just wanted to make everyone aware that we're holding a tutorial >>>> again at ASPLOS this year (Sunday 3/2) in the morning. We're >>>> planning to provide public access to the M5 repository in >>>> conjunction with the tutorial. >>>> >>>> We have some exciting new features in v2.0, including: >>>> >>>> - New cache model >>>> - More abstract cache coherence that is general enough to allow >>>> multi-level configurations >>>> - Inter-object protocol for memory-system objects >>>> - Python creation of object parameters >>>> - Support for ARM, SPARC, and MIPS >>>> - Performance improvements >>>> - Creating and using Simpoints >>>> >>>> Even if you came to last year's tutorial, you can come again to >>>> learn about these new features. >>>> >>>> We'll also be discussing: >>>> - M5 structure >>>> - Specifying configurations >>>> - Running experiments >>>> - Statistics >>>> - Debugging techniques >>>> - Future development directions >>>> >>>> So, if you're new to M5, you're going to learn enough to get >>>> started. >>>> >>>> Our tutorial page: >>>> http://www.m5sim.org/wiki/index.php/ASPLOS_2008 >>>> >>>> ASPLOS web page (registration, schedule, etc.): >>>> http://research.microsoft.com/asplos08/ >>>> >>>> Hope to see you there! >>>> >>>> Ali >>>> _______________________________________________ >>>> m5-users mailing list >>>> m5-users(a)m5sim.org >>>> http://m5sim.org/cgi-bin/mailman/listinfo/m5-users >>>> >>> >>> _______________________________________________ >>> m5-users mailing list >>> m5-users(a)m5sim.org >>> http://m5sim.org/cgi-bin/mailman/listinfo/m5-users >>> >> > _______________________________________________ > m5-users mailing list > m5-users(a)m5sim.org > http://m5sim.org/cgi-bin/mailman/listinfo/m5-users >