- codes-ross-users - mailman3.cels.anl.gov

Re: [codes-ross-users] Fat-tree configuration support oversubscription >1?
by noah wolfe 01 Feb '17

01 Feb '17

Hi Peixin, The current implementation of the CODES fat-tree model only supports 1:1 port mapping for leaf level switches to maintain the standard full bisection fat-tree layout (i.e. half the ports go down to hosts and the other half go up to the next level switches). However, the mapping of ports can be adjusted for leaf level switches (and all levels of switches) in the switch_init() function. Perhaps adding another ft_type for oversubscription of leaf switches would be the best approach. The mapping of hosts to switches which is done in the ft_terminal_init() would also need to be adjusted for the increased number of host connections to leaf switches. Both functions are found in src/networks/model-net/fattree.c. By the way, recently updated documentation for the CODES fat-tree model can be found on the wiki: https://xgitlab.cels.anl.gov/codes/codes/wikis/codes-fattree Thanks, Noah From: <codes-ross-users-bounces(a)lists.mcs.anl.gov <mailto:[email protected]>> on behalf of Peixin Qiao <pqiao(a)hawk.iit.edu <mailto:[email protected]>> Date: Tuesday, January 31, 2017 at 12:02 PM To: "codes-ross-users(a)lists.mcs.anl.gov <mailto:[email protected]>" <codes-ross-users(a)lists.mcs.anl.gov <mailto:[email protected]>> Subject: [codes-ross-users] Fat-tree configuration support oversubscription >1? Hi there, When I do research about fat-tree topology uisng CODES, the fat-tree.conf shows that the topology is with an oversubscription of 1:1, which indicates that all hosts may potentially communicate with arbitrary other hosts at the full bandwidth of their network interface. But the Stampede systems, which I simulated is with an oversubscription of 1.25:1, which means that each leaf switch with 36 ports is directly connected to 20 hosts and 16 ports in the core switch. So, does CODES support this configuration? And how should I modify the configuration file? Best Regards, Peixin Qiao

1 0

CODES install error
by Peixin Qiao 25 Jan '17

25 Jan '17

Hi CODES researchers, When I install and make test CODES on our server, it report some errors, which is shown in attachment: test-suite.log. Best Regards, Peixin

2 1

CODES activities at SC
by Mubarak, Misbah 09 Nov '16

09 Nov '16

Hi All, If you are going to SC'16 next week, here is a list of CODES related activities in case you would like to attend. Thanks to Phil for putting together the list. * workshop paper: Visual Data-Analytics of Large-Scale Parallel Discrete-Event Simulations<http://www.dcs.warwick.ac.uk/pmbs/pmbs/PMBS/Schedule.html>, by Caitlin Ross, Christopher D. Carothers, Misbah Mubarak, Philip Carns, Robert Ross, Jianping Kelvin Li, and Kwan-Liu Ma. PMBS workshop, Sunday Nov. 13, 4:00pm-4:30pm. * workshop poster: Performance Analysis and Visualization of Dragonfly Network Simulations<http://www.womeninhpc.org/women-in-hpc-at-sc16/workshop/sc-2016-workshop-sp…>, by Caitlin Ross. WHPC workshop, Sunday Nov. 13. * conference paper: Watch Out for the Bully! Job Interference Study on Dragonfly Network<http://sc16.supercomputing.org/presentation/?id=pap427&sess=sess158>, by Xu Yang, John Jenkins, Misbah Mubarak, Robert B. Ross, and Zhiling Lan. Thursday, Nov. 17, 11:00am - 11:30am. * conference poster: Design and Evaluation of Topology-Aware Scatter and AllGather Algorithms for Dragonfly Networks<http://sc16.supercomputing.org/presentation/?id=spost146&sess=sess318>, by Nathanael Cheriere. Tuesday, Nov. 15th, 5:15pm-7:00pm. We look forward to seeing many of you next week! Regards, Misbah

1 0

update on dragonfly model: Beta version
by Mubarak, Misbah 04 Nov '16

04 Nov '16

Another example configuration based on the Theta dragonfly interconnect has been added. I have also modified the network config files to reflect more realistic bandwidth numbers so you might want to update your repos. Thanks, Misbah From: Misbah Mubarak <mmubarak(a)anl.gov<mailto:[email protected]>> Date: Wednesday, November 2, 2016 at 3:27 PM To: "codes-core(a)lists.mcs.anl.gov<mailto:[email protected]>" <codes-core(a)lists.mcs.anl.gov<mailto:[email protected]>> Subject: New dragonfly model: Beta version Hi All, The new dragonfly model based on Cray style interconnects is now available in the CODES master repo. There is documentation available on generating the network configuration files, routing algorithms and running the simulation available at: src/networks/model-net/doc/README.dragonfly-custom.txt. Along with minimal and non-minimal routings, the model supports local adaptive, global adaptive and progressive adaptive routings. You can use the available network configurations in the repo. Currently Cray Edison configuration is around along with an example custom dragonfly configuration. Thanks to Nikhil and the LLNL team for contributing towards the topology setup and network configurations. Feel free to provide your feedback. Feature requests, issues and bug reports can be done at: https://xgitlab.cels.anl.gov/codes/codes/issues Regards, Misbah

1 0

New dragonfly model: Beta version
by Mubarak, Misbah 02 Nov '16

02 Nov '16

Hi All, The new dragonfly model based on Cray style interconnects is now available in the CODES master repo. There is documentation available on generating the network configuration files, routing algorithms and running the simulation available at: src/networks/model-net/doc/README.dragonfly-custom.txt. Along with minimal and non-minimal routings, the model supports local adaptive, global adaptive and progressive adaptive routings. You can use the available network configurations in the repo. Currently Cray Edison configuration is around along with an example custom dragonfly configuration. Thanks to Nikhil and the LLNL team for contributing towards the topology setup and network configurations. Feel free to provide your feedback. Feature requests, issues and bug reports can be done at: https://xgitlab.cels.anl.gov/codes/codes/issues Regards, Misbah

2 1

codes model net problems
by 于雷（Yu Lei） 30 Sep '16

30 Sep '16

1/for the method: model_net_event_return model_net_event_annotated( int net_id, char const * annotation, char const * category, tw_lpid final_dest_lp, uint64_t message_size, tw_stime offset, int remote_event_size, void const * remote_event, int self_event_size, void const * self_event, tw_lp *sender); we think that two lps with same network type but with different annotation can not communicate with this method. But we do some tests which prove that the communication can be established. We don't know why? 2/for simplenet model, it needs a parameter "start_up". How can we obtain this parameter? Thanks, Lei

3 9

Summer of CODES! CODES 0.5.2!
by Jenkins, Jonathan P. 15 Jul '16

15 Jul '16

Hi all, It's been a busy week! First, thanks to everyone who attended this year's Summer of CODES workshop! Most of the talk slides have been posted at http://press3.mcs.anl.gov/summerofcodes2016/workshop-proceedings/ . I'll post more as I receive them. Some pictures of the workshop in action can be found at http://press3.mcs.anl.gov/summerofcodes2016/media/ . Second, we released CODES 0.5.2! Info and links at http://press3.mcs.anl.gov/codes/2016/07/15/codes-0-5-2-released/ . Thanks everyone for making the release happen! Thanks, John

1 0

CODES 0.5.1 Released!
by Jenkins, Jonathan P. 09 Jun '16

09 Jun '16

Hi all, We are happy to announce the release of CODES 0.5.1! This release is a hotfix for latency calculation on the dragonfly, along with a few trivia missed in the past release. Downloads can be found at http://www.mcs.anl.gov/projects/codes/downloads/. Thanks, John

1 0

CODES 0.5.0 Released!
by Jenkins, Jonathan P. 24 May '16

24 May '16

Hi all, CODES 0.5.0 has been released! Read the announcement here: http://press3.mcs.anl.gov/codes/2016/05/24/codes-0-5-0-released/. Thanks, John

1 0

AVL tree size
by Jenkins, Jonathan P. 17 Feb '16

17 Feb '16

Hi all, Does the AVL tree size parameter have effects on performance, or is it simply the number of nodes for AVL operations that are allocated at startup? Thanks, John

3 3