Re: [m5-dev] possible contributions to M5
Got it! I'll take a look at m5threads and prepare the patches to upload! Thanks! Jiayuan On Thu, Jul 15, 2010 at 9:51 AM, Steve Reinhardt ste...@gmail.com wrote: On Wed, Jul 14, 2010 at 6:46 AM, Jiayuan Meng jerryh...@gmail.com wrote: Thank you all for the suggestions! === I have the following questions: === 1. how would M5 support plugins with pseudo instructions in the ISA? Following Gabe's thoughts: * M5 can provide general, customizable pseudo instructions One possibility is to have this pseudo instruction take several parameters, with one of the parameter being an operation ID which developers can link the ID to the function to be called (like how system calls are handled). Right now this is just a gleam in Gabe's eye, I believe. With the current code base, pseudo instructions require code modifications in the ISA description so they can't be done without changing the main repository. * As more people are considering to use M5, M5 can become more popular with a plugin website for people to download plugins/patches approved by your team to customize M5. The fact is that for now, those contributions either go into the mainline or they don't appear at all. We're generally pretty willing to host plugins/extras as separate repos on repo.m5sim.org. The m5threads and encumbered repos are already examples of this (sort of). I hadn't really considered hosting a patch queue and not a repository that's added in via EXTRAS, but we could consider that. 2. how can I get m5threads? I searched online and and looked at m5 user discussions, but can't find a place to download it. http://repo.m5sim.org/m5threads And to summarize your opinions r.t. potential contributions === *** It's green light for the following two items: 4. Batch Simulation System : I will document briefly and send a tarball to Nate 1. Multi-threaded TimingSimpleCPU: *** The following items are wanted features but they have several concerns: 2. SIMD cores (depend on #3) : requires two additional instructions in the ISA 3. Parallel Thread API in SE mode : requires one instruction in the ISA, and it appends the system call list. Need to be compared with M5threads. *** The following items are back burners and might be useful: directory-based cache coherence, interconnects Perhaps the best next step for all of these is to upload patches to reviews.m5sim.org? That would let us look at the code directly and give you feedback rather than asking questions about how it works. Let me know if I missed anything. Sounds good to me... thank you! Steve ___ m5-dev mailing list m5-dev@m5sim.org http://m5sim.org/mailman/listinfo/m5-dev ___ m5-dev mailing list m5-dev@m5sim.org http://m5sim.org/mailman/listinfo/m5-dev
Re: [m5-dev] possible contributions to M5
Thank you all for the suggestions! === I have the following questions: === 1. how would M5 support plugins with pseudo instructions in the ISA? Following Gabe's thoughts: * M5 can provide general, customizable pseudo instructions One possibility is to have this pseudo instruction take several parameters, with one of the parameter being an operation ID which developers can link the ID to the function to be called (like how system calls are handled). * As more people are considering to use M5, M5 can become more popular with a plugin website for people to download plugins/patches approved by your team to customize M5. The fact is that for now, those contributions either go into the mainline or they don't appear at all. 2. how can I get m5threads? I searched online and and looked at m5 user discussions, but can't find a place to download it. And to summarize your opinions r.t. potential contributions === *** It's green light for the following two items: 4. Batch Simulation System : I will document briefly and send a tarball to Nate 1. Multi-threaded TimingSimpleCPU: *** The following items are wanted features but they have several concerns: 2. SIMD cores (depend on #3) : requires two additional instructions in the ISA 3. Parallel Thread API in SE mode : requires one instruction in the ISA, and it appends the system call list. Need to be compared with M5threads. *** The following items are back burners and might be useful: directory-based cache coherence, interconnects Let me know if I missed anything. Thanks! Jiayuan ___ m5-dev mailing list m5-dev@m5sim.org http://m5sim.org/mailman/listinfo/m5-dev
Re: [m5-dev] possible contributions to M5
2. SIMD cores: Based on TimingSimpleCPU. This sounds pretty interesting too. Does the ISA matter much, or do you think it could be pretty generic? Did you add new instructions to do things? Yes, I amended the Alpha ISA and added two instructions to mark the begin and end of branch divergences. I assume we can do the same thing for other ISAs. 3. A parallel API for writing parallel benchmarks that run in system emulation mode. So, this doesn't depend on #2? I assume that this is in syscall emulation mode? This is pretty interesting. The one issue is that there are already two? systems in m5 for doing parallel simulations and we don't want to just throw the kitchen sink into m5. Have you looked at the current m5threads stuff? Is there any overlap? No it doesn't. Actually, #2 depends on this one. Yes, it's in syscall emulation mode. Thanks for the pointer, I'll take a look at m5threads. 4. A batch-simulation tool, which creates, manages, organizes, and analyzes simulation tasks in batches. Written in Python. Good for space exploration and sensitivity study. Sure, I will write some documentation and send you a tarball. Thanks, Jiayuan ___ m5-dev mailing list m5-dev@m5sim.org http://m5sim.org/mailman/listinfo/m5-dev
Re: [m5-dev] possible contributions to M5
1. An in-order CPU module with multi-threading (switching threads upon cache accesses) based on TimingSimpleCPU This is probably #1 on the priority list Is this inorder model detailed (in terms of pipeline stages, branch prediciton, Functional Units, etc.)? No it's not detailed (so I guess it does not antiquate the actual inorderCPU model that you are developing). It's simply have multiple active thread contexts on a TimingSimpleCPU and the CPU switches among them upon memory accesses. Is it functional within Full system mode? I've only used it for system emulation mode. If your team is interested and can help me with fully system mode, I am willing to bring it to full system mode afterwards. Thanks, Jiayuan ___ m5-dev mailing list m5-dev@m5sim.org http://m5sim.org/mailman/listinfo/m5-dev
Re: [m5-dev] possible contributions to M5
Hi Steve, Personally I would say 2,3,4,1... but I can understand how others might have different priorities. (Actually I haven't seen two people with the same order yet!) Maybe I should first re-take Stats101 to measure the best average of the ordering :) I think a key question that Nate already hinted at is how related are items 1-3... I can see #1 being somewhat independent, but it sounds like #2 #3 might be interdependent? Do they also depend on #1? Yes, #2 depends on #3. #1 is standalone. All of them are only tested in system emulation mode. Thanks, Jiayuan ___ m5-dev mailing list m5-dev@m5sim.org http://m5sim.org/mailman/listinfo/m5-dev
[m5-dev] possible contributions to M5
Dear M5 Team, I'll soon start to work on integrating our M5-based simulator, MV5, into M5. I've read some documentation on the current status of M5, and I'd like to share my plan/ideas with you and hear your suggestions. Since M5 will soon replace its entire memory system modeling with RubyGems, it seems the interconnect and directory coherence modules in MV5 will have little value for future M5. I'm wondering if the following contributions would be interesting to you. 1. An in-order CPU module with multi-threading (switching threads upon cache accesses) based on TimingSimpleCPU. 2. SIMD cores: Based on TimingSimpleCPU. It models GPU-like branch handling using the re-convergence stack. It is implemented as an array of scalar threads rather than wider vector registers. It requires a parallel API written in a similar style to OpenMP, with manually instrumentation to signal the end of a branch in the parallel section of code. 3. A parallel API for writing parallel benchmarks that run in system emulation mode. It can be used in a similar way to Pthread. It also provide OpenMP-like programming interface, and it can work with the SIMD cores. It also comes with various scheduling policies. 4. A batch-simulation tool, which creates, manages, organizes, and analyzes simulation tasks in batches. Written in Python. Good for space exploration and sensitivity study. I manage tens of thousands of simulations over up to 100 machines using this tool. It can work with a cluster of machines with a shared file system. It also supports the PBS queuing system. What do you think about the above contributions? Can you prioritize them? Thanks, Jiayuan ___ m5-dev mailing list m5-dev@m5sim.org http://m5sim.org/mailman/listinfo/m5-dev
Re: [m5-dev] Any interest in directory coherence based on Jiayuan's Patch
Dear all, Just curious --- Is anyone going to ICS or coming around New York around June 10 or 11th? If so, we can meet and talk :) I know it's been a long struggle --- we have made a couple attempts in the past to get the patches into M5's source tree, but we didn't make much progress and I think the main reason is we need time to communicate and digest the patches before they can be integrated. I fully respect the work of Gems and Ruby and it would be great to combine them with M5. In the meantime, I think the patches offer good alternatives and can become a integral tool for the M5 user community. They are designed, programmed and used in an M5 style and philosophy, and they have been tested by various users and now support full-system simulation with Parsec (Lots of thanks to Rick!). I am open to sit down for a week or two focusing on the integration, and I think two weeks would be sufficient for coding and initial testings. To re-kick off the ball and move on, I can make slides to show the structure/design of the patches, send them to you for feedback, and then we can arrange some time to discuss more details. Would you be interested to resume the process? Cheers, Jiayuan On Thu, May 21, 2009 at 8:23 PM, Rick Strong rstr...@cs.ucsd.edu wrote: I don't know what is going on with ruby at the moment, so I guess it is up to the m5-community if they want the patch to get in or not. So let me know. Thanks, -Rick Ali Saidi wrote: Hi Rick, I think there might be some interest depending on how fast progress with Ruby is made and the quality of the patches. Thanks, Ali On May 1, 2009, at 6:52 PM, Rick Strong wrote: Hi all, I am finishing tests with Jiayuan's directory coherence patch. I have it working with the MemTest class, my memory Debugger for multiple levels, mshr settings and applications in fullsys and syscall mode. I am now putting it through its paces on parsec. It is up-to-date with m5-dev, but I had to take out prefetcher support as I don't fully understand that system yet. Would there be interest in getting this into M5-dev? If so, what tests should I still put it through? Thanks, -Rick ___ m5-dev mailing list m5-dev@m5sim.org http://m5sim.org/mailman/listinfo/m5-dev ___ m5-dev mailing list m5-dev@m5sim.org http://m5sim.org/mailman/listinfo/m5-dev ___ m5-dev mailing list m5-dev@m5sim.org http://m5sim.org/mailman/listinfo/m5-dev ___ m5-dev mailing list m5-dev@m5sim.org http://m5sim.org/mailman/listinfo/m5-dev
Re: [m5-dev] submit patches: Mesh2D, directory coherence, and multithreading
Thanks Steve! By the way, I'll send out a revised version of the banked cache code later this week. Jiayuan - Original Message - From: Steve Reinhardt To: M5 Developer List Sent: 2008年10月13日 10:33 PM Subject: Re: [m5-dev] submit patches: Mesh2D, directory coherence,and multithreading This looks great, thanks. I started looking at your patches back when you sent them out but ran out of time and haven't gotten back to it. I'm on vacation right now, but I'll take a look at the banked cache code specifically when I get back to work (later this week or early next). Steve On Sun, Oct 12, 2008 at 1:52 PM, jiayuan meng [EMAIL PROTECTED] wrote: Hey all, As a first step to submit my first patch, here is a brief PPT presentation for the banked cache module. Let me know if you have any questions. I will refine the code and submit the full patch after learning your opinions. http://www.cs.virginia.edu/~jm6dg/fractal/slides/M5%20Multibanked%20Cache.ppt Thanks! Jiayuan Date: Sun, 21 Sep 2008 21:01:31 -0700 From: [EMAIL PROTECTED] To: m5-dev@m5sim.org Subject: Re: [m5-dev] submit patches: Mesh2D, directory coherence, and multithreading This all sounds fantastic. We'll do what we can to help! Nate Thank you for your reviewing! Here is my tentative plan for submitting my patches: I'll work on the patches in the following order: 1. BankedCache 2. On-chip network 3. Directory-based Coherence 4. multithreading for simplecpus. For each patch, shall we work in the following protocol? a). I will present the code and we will discuss over emails/conference calls. b). I will refine the code according to our discussion. c). the refined code will be sent to you for another review. if necessary, we can setup another conference call. d). can you help me with the regression test, while I go on to work on the next patch? We can kick this off in October. Considering other deadlines and vacations, it may take three to four months to complete all patches. What do you think? Any suggestions? Thanks! Jiayuan Hey Nathan and all, Thanks for the support! Here is a preview version, you can download the patches from: http://www.cs.virginia.edu/~jm6dg/fractal/m5patches.htm I'm open to any comments/suggestio ns. Let me know if you have any questions. Thanks! Jiayuan - Original Message - From: nathan binkert To: M5 Developer List Sent: 2008年7月12日 7:50 AM Subject: Re: [m5-dev] submit patches: Mesh2D, directory coherence,and multithreading Wow, that's fantastic. The best thing you can do is get your code working with the new repository using mercurial and mercurial queues. If you do that, then we can work with you to make sure that the code meets all of the requirements (doesn't break things, is reasonably organized, meets the style guidelines, etc.) Did y ou use a revision control system as you did your work? Thanks, Nate 2008/7/11 jiayuan meng : Dear all, I appreciate all your support to help me with M5. I really love it. I'm a graduate student working with Prof. Kevin Skadron at University of Virginia. And now, I'd like to submit patches to M5 that enables it to simulate CMP architecture. I have a mesh2D model that works well with an MSI directory coherence protocol. Both the interconnection and directory coherence models are extendable. Another patch is to support multithreading for simple CPUs . The cpu can switch to another HW thread context if it has a data access. However, atomic modes are not well tested. And I haven't considered checkpoints. The mercurial patches are based on changeset 5291:456a50cf2a67, a version after 2.0beta4 but before 2.0beta5. As an external developer, how do you think I should submit the patches? An do you think it would be necessary to update the patches so they are based on a more recent changeset? If so, which changeset? Just hope it might be useful. Thanks! Jiayuan Meng gt; Need to know now? Get instant answers with Windows Live Messenger. IM on your terms. ___ m5-dev mailing list m5-dev@m5sim.org http://m5sim.org/mailman/listinfo/m5-dev ___ m5-dev mailing list m5-dev@m5sim.org http://m5sim.org/mailman/listinfo/m5-dev ___ m5-dev mailing list m5-dev@m5sim.org http
Re: [m5-dev] submit patches: Mesh2D, directory coherence, and multithreading
Great! so I will go on to prepare for the presentation about banked caches. I'll let you know when I'm ready to present. Thanks! Jiayuan Date: Sun, 21 Sep 2008 21:01:31 -0700 From: [EMAIL PROTECTED] To: m5-dev@m5sim.org Subject: Re: [m5-dev] submit patches: Mesh2D, directory coherence,and multithreading This all sounds fantastic. We'll do what we can to help! Nate Thank you for your reviewing! Here is my tentative plan for submitting my patches: I'll work on the patches in the following order: 1. BankedCache 2. On-chip network 3. Directory-based Coherence 4. multithreading for simplecpus. For each patch, shall we work in the following protocol? a). I will present the code and we will discuss over emails/conference calls. b). I will refine the code according to our discussion. c). the refined code will be sent to you for another review. if necessary, we can setup another conference call. d). can you help me with the regression test, while I go on to work on the next patch? We can kick this off in October. Considering other deadlines and vacations, it may take three to four months to complete all patches. What do you think? Any suggestions? Thanks! Jiayuan Hey Nathan and all, Thanks for the support! Here is a preview version, you can download the patches from: http://www.cs.virginia.edu/~jm6dg/fractal/m5patches.htm I'm open to any comments/suggestions. Let me know if you have any questions. Thanks! Jiayuan - Original Message - From: nathan binkert [EMAIL PROTECTED] To: M5 Developer List m5-dev@m5sim.org Sent: 2008年7月12日 7:50 AM Subject: Re: [m5-dev] submit patches: Mesh2D, directory coherence,and multithreading Wow, that's fantastic. The best thing you can do is get your code working with the new repository using mercurial and mercurial queues. If you do that, then we can work with you to make sure that the code meets all of the requirements (doesn't break things, is reasonably organized, meets the style guidelines, etc.) Did you use a revision control system as you did your work? Thanks, Nate 2008/7/11 jiayuan meng [EMAIL PROTECTED]: Dear all, I appreciate all your support to help me with M5. I really love it. I'm a graduate student working with Prof. Kevin Skadron at University of Virginia. And now, I'd like to submit patches to M5 that enables it to simulate CMP architecture. I have a mesh2D model that works well with an MSI directory coherence protocol. Both the interconnection and directory coherence models are extendable. Another patch is to support multithreading for simple CPUs. The cpu can switch to another HW thread context if it has a data access. However, atomic modes are not well tested. And I haven't considered checkpoints. The mercurial patches are based on changeset 5291:456a50cf2a67, a version after 2.0beta4 but before 2.0beta5. As an external developer, how do you think I should submit the patches? An do you think it would be necessary to update the patches so they are based on a more recent changeset? If so, which changeset? Just hope it might be useful. Thanks! Jiayuan Meng Need to know now? Get instant answers with Windows Live Messenger. IM on your terms. ___ m5-dev mailing list m5-dev@m5sim.org http://m5sim.org/mailman/listinfo/m5-dev ___ m5-dev mailing list m5-dev@m5sim.org http://m5sim.org/mailman/listinfo/m5-dev ___ m5-dev mailing list m5-dev@m5sim.org http://m5sim.org/mailman/listinfo/m5-dev ___ m5-dev mailing list m5-dev@m5sim.org http://m5sim.org/mailman/listinfo/m5-dev ___ m5-dev mailing list m5-dev@m5sim.org http://m5sim.org/mailman/listinfo/m5-dev _ See how Windows Mobile brings your life together—at home, work, or on the go. http://clk.atdmt.com/MRT/go/msnnkwxp1020093182mrt/direct/01/___ m5-dev mailing list m5-dev@m5sim.org http://m5sim.org/mailman/listinfo/m5-dev
Re: [m5-dev] submit patches: Mesh2D, directory coherence, and multithreading
Dear all, Thank you for your reviewing! Here is my tentative plan for submitting my patches: I'll work on the patches in the following order: 1. BankedCache 2. On-chip network 3. Directory-based Coherence 4. multithreading for simplecpus. For each patch, shall we work in the following protocol? a). I will present the code and we will discuss over emails/conference calls. b). I will refine the code according to our discussion. c). the refined code will be sent to you for another review. if necessary, we can setup another conference call. d). can you help me with the regression test, while I go on to work on the next patch? We can kick this off in October. Considering other deadlines and vacations, it may take three to four months to complete all patches. What do you think? Any suggestions? Thanks! Jiayuan Hey Nathan and all, Thanks for the support! Here is a preview version, you can download the patches from: http://www.cs.virginia.edu/~jm6dg/fractal/m5patches.htm I'm open to any comments/suggestions. Let me know if you have any questions. Thanks! Jiayuan - Original Message - From: nathan binkert [EMAIL PROTECTED] To: M5 Developer List m5-dev@m5sim.org Sent: 2008年7月12日 7:50 AM Subject: Re: [m5-dev] submit patches: Mesh2D, directory coherence,and multithreading Wow, that's fantastic. The best thing you can do is get your code working with the new repository using mercurial and mercurial queues. If you do that, then we can work with you to make sure that the code meets all of the requirements (doesn't break things, is reasonably organized, meets the style guidelines, etc.) Did you use a revision control system as you did your work? Thanks, Nate 2008/7/11 jiayuan meng [EMAIL PROTECTED]: Dear all, I appreciate all your support to help me with M5. I really love it. I'm a graduate student working with Prof. Kevin Skadron at University of Virginia. And now, I'd like to submit patches to M5 that enables it to simulate CMP architecture. I have a mesh2D model that works well with an MSI directory coherence protocol. Both the interconnection and directory coherence models are extendable. Another patch is to support multithreading for simple CPUs. The cpu can switch to another HW thread context if it has a data access. However, atomic modes are not well tested. And I haven't considered checkpoints. The mercurial patches are based on changeset 5291:456a50cf2a67, a version after 2.0beta4 but before 2.0beta5. As an external developer, how do you think I should submit the patches? An do you think it would be necessary to update the patches so they are based on a more recent changeset? If so, which changeset? Just hope it might be useful. Thanks! Jiayuan Meng Need to know now? Get instant answers with Windows Live Messenger. IM on your terms. ___ m5-dev mailing list m5-dev@m5sim.org http://m5sim.org/mailman/listinfo/m5-dev ___ m5-dev mailing list m5-dev@m5sim.org http://m5sim.org/mailman/listinfo/m5-dev ___ m5-dev mailing list m5-dev@m5sim.org http://m5sim.org/mailman/listinfo/m5-dev ___ m5-dev mailing list m5-dev@m5sim.org http://m5sim.org/mailman/listinfo/m5-dev
Re: [m5-dev] submit patches: Mesh2D, directory coherence, and multithreading
Hey Nathan and all, Thanks for the support! Here is a preview version, you can download the patches from: http://www.cs.virginia.edu/~jm6dg/fractal/m5patches.htm I'm open to any comments/suggestions. Let me know if you have any questions. Thanks! Jiayuan - Original Message - From: nathan binkert [EMAIL PROTECTED] To: M5 Developer List m5-dev@m5sim.org Sent: 2008年7月12日 7:50 AM Subject: Re: [m5-dev] submit patches: Mesh2D, directory coherence,and multithreading Wow, that's fantastic. The best thing you can do is get your code working with the new repository using mercurial and mercurial queues. If you do that, then we can work with you to make sure that the code meets all of the requirements (doesn't break things, is reasonably organized, meets the style guidelines, etc.) Did you use a revision control system as you did your work? Thanks, Nate 2008/7/11 jiayuan meng [EMAIL PROTECTED]: Dear all, I appreciate all your support to help me with M5. I really love it. I'm a graduate student working with Prof. Kevin Skadron at University of Virginia. And now, I'd like to submit patches to M5 that enables it to simulate CMP architecture. I have a mesh2D model that works well with an MSI directory coherence protocol. Both the interconnection and directory coherence models are extendable. Another patch is to support multithreading for simple CPUs. The cpu can switch to another HW thread context if it has a data access. However, atomic modes are not well tested. And I haven't considered checkpoints. The mercurial patches are based on changeset 5291:456a50cf2a67, a version after 2.0beta4 but before 2.0beta5. As an external developer, how do you think I should submit the patches? An do you think it would be necessary to update the patches so they are based on a more recent changeset? If so, which changeset? Just hope it might be useful. Thanks! Jiayuan Meng Need to know now? Get instant answers with Windows Live Messenger. IM on your terms. ___ m5-dev mailing list m5-dev@m5sim.org http://m5sim.org/mailman/listinfo/m5-dev ___ m5-dev mailing list m5-dev@m5sim.org http://m5sim.org/mailman/listinfo/m5-dev ___ m5-dev mailing list m5-dev@m5sim.org http://m5sim.org/mailman/listinfo/m5-dev
Re: [m5-dev] submit patches: Mesh2D, directory coherence, and multithreading
Hey all, FYI, I've finished most part of the patch, but wish it can be tested a bit more before I submit it to the repository. Since there are some deadlines coming, I hope submitting it in late Aug or early Sept won't bother you much? Thanks! Jiayuan Date: Sat, 12 Jul 2008 10:29:17 -0700 From: [EMAIL PROTECTED] To: m5-dev@m5sim.org Subject: Re: [m5-dev] submit patches: Mesh2D, directory coherence,and multithreading No, please base your patches on m5 (the development repo) rather than m5-stable, as that's where we'll want to integrate them. Thanks, Steve 2008/7/12 jiayuan meng [EMAIL PROTECTED]: I'm going to pull the repository of m5_stable and base my patches upon this version, will that be fine? Date: Fri, 11 Jul 2008 16:50:19 -0700 From: [EMAIL PROTECTED] To: m5-dev@m5sim.org Subject: Re: [m5-dev] submit patches: Mesh2D, directory coherence, and multithreading Wow, that's fantastic. The best thing you can do is get your code working with the new repository using mercurial and mercurial queues. If you do that, then we can work with you to make sure that the code meets all of the requirements (doesn't break things, is reasonably organized, meets the style guidelines, etc.) Did you use a revision control system as you did your work? Thanks, Nate 2008/7/11 jiayuan meng [EMAIL PROTECTED]: Dear all, I appreciate all your support to help me with M5. I really love it. I'm a graduate student working with Prof. Kevin Skadron at University of Virginia. And now, I'd like to submit patches to M5 that enables it to simulate CMP architecture. I have a mesh2D model that works well with an MSI directory coherence protocol. Both the interconnection and directory coherence models are extendable. Another patch is to support multithreading for simple CPUs. The cpu can switch to another HW thread context if it has a data access. However, atomic modes are not well tested. And I haven't considered checkpoints. The mercurial patches are based on changeset 5291:456a50cf2a67, a version after 2.0beta4 but before 2.0beta5. As an external developer, how do you think I should submit the patches? An do you think it would be necessary to update the patches so they are based on a more recent changeset? If so, which changeset? Just hope it might be useful. Thanks! Jiayuan Meng Need to know now? Get instant answers with Windows Live Messenger. IM on your terms. ___ m5-dev mailing list m5-dev@m5sim.org http://m5sim.org/mailman/listinfo/m5-dev ___ m5-dev mailing list m5-dev@m5sim.org http://m5sim.org/mailman/listinfo/m5-dev It's a talkathon – but it's not just talk. Check out the i'm Talkathon. ___ m5-dev mailing list m5-dev@m5sim.org http://m5sim.org/mailman/listinfo/m5-dev ___ m5-dev mailing list m5-dev@m5sim.org http://m5sim.org/mailman/listinfo/m5-dev _ Time for vacation? WIN what you need- enter now! http://www.gowindowslive.com/summergiveaway/?ocid=tag_jlyhm___ m5-dev mailing list m5-dev@m5sim.org http://m5sim.org/mailman/listinfo/m5-dev
Re: [m5-dev] submit patches: Mesh2D, directory coherence, and multithreading
I'm going to pull the repository of m5_stable and base my patches upon this version, will that be fine? Date: Fri, 11 Jul 2008 16:50:19 -0700 From: [EMAIL PROTECTED] To: m5-dev@m5sim.org Subject: Re: [m5-dev] submit patches: Mesh2D, directory coherence,and multithreading Wow, that's fantastic. The best thing you can do is get your code working with the new repository using mercurial and mercurial queues. If you do that, then we can work with you to make sure that the code meets all of the requirements (doesn't break things, is reasonably organized, meets the style guidelines, etc.) Did you use a revision control system as you did your work? Thanks, Nate 2008/7/11 jiayuan meng [EMAIL PROTECTED]: Dear all, I appreciate all your support to help me with M5. I really love it. I'm a graduate student working with Prof. Kevin Skadron at University of Virginia. And now, I'd like to submit patches to M5 that enables it to simulate CMP architecture. I have a mesh2D model that works well with an MSI directory coherence protocol. Both the interconnection and directory coherence models are extendable. Another patch is to support multithreading for simple CPUs. The cpu can switch to another HW thread context if it has a data access. However, atomic modes are not well tested. And I haven't considered checkpoints. The mercurial patches are based on changeset 5291:456a50cf2a67, a version after 2.0beta4 but before 2.0beta5. As an external developer, how do you think I should submit the patches? An do you think it would be necessary to update the patches so they are based on a more recent changeset? If so, which changeset? Just hope it might be useful. Thanks! Jiayuan Meng Need to know now? Get instant answers with Windows Live Messenger. IM on your terms. ___ m5-dev mailing list m5-dev@m5sim.org http://m5sim.org/mailman/listinfo/m5-dev ___ m5-dev mailing list m5-dev@m5sim.org http://m5sim.org/mailman/listinfo/m5-dev _ It’s a talkathon – but it’s not just talk. http://www.imtalkathon.com/?source=EML_WLH_Talkathon_JustTalk___ m5-dev mailing list m5-dev@m5sim.org http://m5sim.org/mailman/listinfo/m5-dev
Re: [m5-dev] submit patches: Mesh2D, directory coherence, and multithreading
Thanks! Yeah, I used mercurial queue to manage the patches. Ali and Steve helped me with that. I'll see if I can pull out the most recent version, and please give me some time to update the patch for the new changeset. Jiayuan Date: Fri, 11 Jul 2008 16:50:19 -0700 From: [EMAIL PROTECTED] To: m5-dev@m5sim.org Subject: Re: [m5-dev] submit patches: Mesh2D, directory coherence,and multithreading Wow, that's fantastic. The best thing you can do is get your code working with the new repository using mercurial and mercurial queues. If you do that, then we can work with you to make sure that the code meets all of the requirements (doesn't break things, is reasonably organized, meets the style guidelines, etc.) Did you use a revision control system as you did your work? Thanks, Nate 2008/7/11 jiayuan meng [EMAIL PROTECTED]: Dear all, I appreciate all your support to help me with M5. I really love it. I'm a graduate student working with Prof. Kevin Skadron at University of Virginia. And now, I'd like to submit patches to M5 that enables it to simulate CMP architecture. I have a mesh2D model that works well with an MSI directory coherence protocol. Both the interconnection and directory coherence models are extendable. Another patch is to support multithreading for simple CPUs. The cpu can switch to another HW thread context if it has a data access. However, atomic modes are not well tested. And I haven't considered checkpoints. The mercurial patches are based on changeset 5291:456a50cf2a67, a version after 2.0beta4 but before 2.0beta5. As an external developer, how do you think I should submit the patches? An do you think it would be necessary to update the patches so they are based on a more recent changeset? If so, which changeset? Just hope it might be useful. Thanks! Jiayuan Meng Need to know now? Get instant answers with Windows Live Messenger. IM on your terms. ___ m5-dev mailing list m5-dev@m5sim.org http://m5sim.org/mailman/listinfo/m5-dev ___ m5-dev mailing list m5-dev@m5sim.org http://m5sim.org/mailman/listinfo/m5-dev _ Need to know now? Get instant answers with Windows Live Messenger. http://www.windowslive.com/messenger/connect_your_way.html?ocid=TXT_TAGLM_WL_messenger_072008___ m5-dev mailing list m5-dev@m5sim.org http://m5sim.org/mailman/listinfo/m5-dev
Re: [m5-dev] submit patches: Mesh2D, directory coherence, and multithreading
great! Thanks for the support! Jiayuan Date: Fri, 11 Jul 2008 18:30:40 -0700 From: [EMAIL PROTECTED] To: m5-dev@m5sim.org Subject: Re: [m5-dev] submit patches: Mesh2D, directory coherence,and multithreading Hi, Great work. I noticed you added support for multithreading for simple CPUs. I wouldn't mind figuring out checkpoints with the multithreaded simple CPUs as it would free me from EIO trace madness. Let me know when the patches are in m5. -Rick jiayuan meng wrote: Thanks! Yeah, I used mercurial queue to manage the patches. Ali and Steve helped me with that. I'll see if I can pull out the most recent version, and please give me some time to update the patch for the new changeset. Jiayuan Date: Fri, 11 Jul 2008 16:50:19 -0700 From: [EMAIL PROTECTED] To: m5-dev@m5sim.org Subject: Re: [m5-dev] submit patches: Mesh2D, directory coherence, and multithreading Wow, that's fantastic. The best thing you can do is get your code working with the new repository using mercurial and mercurial queues. If you do that, then we can work with you to make sure that the code meets all of the requirements (doesn't break things, is reasonably organized, meets the style guidelines, etc.) Did you use a revision control system as you did your work? Thanks, Nate 2008/7/11 jiayuan meng [EMAIL PROTECTED]: Dear all, I appreciate all your support to help me with M5. I really love it. I'm a graduate student working with Prof. Kevin Skadron at University of Virginia. And now, I'd like to submit patches to M5 that enables it to simulate CMP architecture. I have a mesh2D model that works well with an MSI directory coherence protocol. Both the interconnection and directory coherence models are extendable. Another patch is to support multithreading for simple CPUs. The cpu can switch to another HW thread context if it has a data access. However, atomic modes are not well tested. And I haven't considered checkpoints. The mercurial patches are based on changeset 5291:456a50cf2a67, a version after 2.0beta4 but before 2.0beta5. As an external developer, how do you think I should submit the patches? An do you think it would be necessary to update the patches so they are based on a more recent changeset? If so, which changeset? Just hope it might be useful. Thanks! Jiayuan Meng Need to know now? Get instant answers with Windows Live Messenger. IM on your terms. ___ m5-dev mailing list m5-dev@m5sim.org http://m5sim.org/mailman/listinfo/m5-dev ___ m5-dev mailing list m5-dev@m5sim.org http://m5sim.org/mailman/listinfo/m5-dev Need to know now? Get instant answers with Windows Live Messenger. IM on your terms. http://www.windowslive.com/messenger/connect_your_way.html?ocid=TXT_TAGLM_WL_messenger_072008 ___ m5-dev mailing list m5-dev@m5sim.org http://m5sim.org/mailman/listinfo/m5-dev _ Making the world a better place one message at a time. http://www.imtalkathon.com/?source=EML_WLH_Talkathon_BetterPlace___ m5-dev mailing list m5-dev@m5sim.org http://m5sim.org/mailman/listinfo/m5-dev