Intel(R) Xeon Phi(tm) Coprocessor -- Cluster training - call for demand!

[BelindaLiviero]

Intel is evaluating to offer a 4 hour web-based basic tutorial covering the fundamental principles of how to integrate an Intel Xeon Phi coprocessor into a Linux based cluster.

During the course each attendant would have remote access to a Linux server and be able to do each step as shown in the outline below.  The course will be given free of charge. Requirements are an Internet connection, a web browser, and Putty.   We are settling on the sharing technology we will be using, and will publish that at a later date.

If you are interested in such an offer please reply to this forum thread -- you have the ability to reply privately, if you don't want to be identified.

If we have enough interest, we'll pull it together!

Topics:

• Finding information on Intel Xeon Phi coprocessor on the web
• Download the driver software
• Unpacking the driver software package, explanation of components
• Discussion on prerequisites of the compute server (for instance what software needs to be installed, reserved IP addresses, user names, network file systems)
• Basic concepts (host, host OS, host kernel, coprocessor, Intel(R) MPSS stack, layout of files, boot image of the uOS, ramfs of the uOS)
• Recompiling HOST kernel packages; diagnose output  and understand errors (necessary to work with nonstandard kernels)
• Install a minimal set of MPSS rpm packages using rpm
• Create a default MPSS configuration (using “micctrl --initdefaults “)
• Startup (aka boot) the coprocessor
• Connect via minicom to the coprocessor (this allows one to connect to the Intel Xeon Phi coprocessor WITHOUT figuring out network problems)
• modify uOS filesystem by overlaying an /etc/passwd file; reboot the coprocessor
• Create a bridgded network on the host
• Configure the coprocessor for bridged networking by modifying micX.conf  directly
• Reboot card and connect via ssh
• Set up a ssh key-pair; diagnose ssh gotchas
• Mount a NFS file system on the coprocessor
• Configure a user known in the cluster by modifying the /etc/passwd file of the coprocessor
• Group up with neighbor – run MPI benchmark natively over Ethernet
• Recompile the MPSS OFED package to support a nonstandard kernel on the HOST
• Install MPSS-OFED rpms
• Start OFED on the coprocessor
• Group up with neighbor – run MPI benchmark natively over InfiniBand
• Create a minimal startup script wrapping everything up; this startup script can be used by a batch scheduling system to restart a coprocessor on behalf of a user before running a job.
• Where to find more resources or ask questions

In future (advanced) courses we might want to talk about topics like:

• Change micX.conf configuration to adopt to diskless clients
• The tools (micinfo, micrasd, ….)
• Logging – syslog, sar (how to enable logging, typical output and problems to look out for)
• cron: maintenance  log rotation,  detect issues like missing daemons, problems
• Adding more programs/libraries/files to the coprocessors (including pros and cons)
• Troubleshooting startups/shutdowns
• Upgrading the MPSS stack on an established installation
• Diagnose system and coprocessor health
• Customizing installation scripts

[ashish_s_]

Can you share any document to create a Intel MIC cluster based on Infiniband?

Thanks,

Ashish k Singh

[Jonathan_B_1]

Hi,

I am also interested in a futur webinar. Thank you.

Jonathan

[Frances_R_Intel]

Ashish,

Funny you should ask. This is one of the things covered in the training - which is getting closer to having a fixed date.

You can also try the following sources:

Intel® Manycore Platform Software Stack (Intel® MPSS) User's Guide - can be downloaded from the same location that you get the code for the MPSS. The directions in chapter 2 show how to install the software. This is the basic information you need. What isn't always clear is that you start the opensmd service on only one node in your cluster and let it map out your cluster for you. Then section 18.2 talks about IP over InfiniBand if you also want to set that up

Intel® Xeon Phi™ Coprocessor System Software Developers Guide - section 2.2.9 has way more information than you probably want to know at this point but when you feel comfortable with the basic InfiniBand, check out this discussion of about InfiniBand, MPI, CCL Direct and verbs.

Once you install the MPSS with OFED support, the directory /usr/share/doc/ofed-driver-2.6.32-279.el6.x86_64-{mpss_version} will contain a document showing how to set up Lustre file systems over InfiniBand.

There is also Configuring Intel® Xeon Phi™ coprocessors inside a cluster which shows an example of configuring a cluster with InfiniBand and MPSS 2.1 (so it is a bit old.)

You can also search in this forum for InfiniBand to see what other users have found.

Finally, there is an interesting paper from Colfax International, CONFIGURATION AND BENCHMARKS OF PEER-TO-PEER MPI COMMUNICATION OVER GIGABIT ETHERNET AND INFINIBAND IN A CLUSTER WITH INTEL XEON PHI COPROCESSORS

I would suggest that you do the basic install described in chapter 2 of the User's Guide and read through "Configuring Intel® Xeon Phi™ coprocessors inside a cluster" and possibly the Colfax International paper, then start a new forum thread for specific questions you may have. Starting a new thread makes it easier to track your particular questions.

[ashish_s_]

Thanks Roth,

This information was also a great help. The whitepaper has cleared my doubts about infiniband based MIC cluster.

Regards,

Ashish K Singh

[Nimisha_R_Intel]

I am interested in a future webinar too.

Thanks,

Nimisha

[ashish_s_]

Hi,

I am working on Intel MIC cluster with Mellanox Infiniband. MPSS-3.3 is supporting two Intel MIC cards with one Intel HCA card. Is this same configuration supported for Mellanox HCA too?

Thnaks,

Ashish K Singh 

[Frances_R_Intel]

Ashish - yes, The reason it is not explicitly mentioned in the readme file is that support for this configuration with a Mellanox adapter was also available in earlier releases of the MPSS. But support for the this configuration with an Intel adapter is new.

[BelindaLiviero]

All, another webinar is available on this topic -- scheduled for August 12, 8-noon PST.   If you are interested in attending, please register with the 'gotomeeting' link below :

Intel® Xeon Phi™ Cluster Integration – A hands-on Introduction

Join us for a Webinar on August 12     Space is limited. Reserve your Webinar seat now at: https://www1.gotomeeting.com/register/197769169   In this 4-hour course led by an instructor the attendants will learn how to integrate intel Xeon Phi coprocessors into a cluster.  Each attendant will connect via ssh to a cluster, and be able to run all necessary commands alongside the presentation. Title: Intel® Xeon Phi™ Cluster Integration – A hands-on Introduction Date: Tuesday, August 12, 2014 Time: 8:00 AM - 12:00 PM PDT After registering you will receive a confirmation email containing information about joining the Webinar. System Requirements PC-based attendees Required: Windows® 8, 7, Vista, XP or 2003 Server   Mac®-based attendees Required: Mac OS® X 10.6 or newer   Mobile attendees Required: iPhone®, iPad®, Android™ phone or Android tablet

[ashish_s_]

HI,

I have registered, but did not get any email to attend the Intel Phi cluster training. Can you please send me the link to connect?

Thanks,

AKS

[Nimisha_R_Intel]

Hi,

Even I have registered and when I use the link for joining the webinar it says Approval Required.

Thanks,

Nimisha

[BelindaLiviero]

today's webinar is full and we have more sign-ups than seats - we are working to schedule 2-3 more sessions and I will update this post with details

[ashish_s_]

Kindly arrange training for the same again. I didn't get the approval to attend this training on 12th august.

Thanks,

AKS

[Patrick_S_]

I'm also interested in another training.

[Randy_B_]

I am interested in this training as well. I hope there is another one soon.

[Patrick_S_]

Is it possible that the next training session will be at a time which is more suitable for people here in europe?

[BelindaLiviero]

Two new instances of the “Intel® Xeon Phi™ Coprocessor cluster training” webinars  have been scheduled (September 16 and September 24); details are below.   

Please pick the (one) session that works best for you.   Intel employees:  please be aware that customers receive priority seating.

September 16  8:00am-12:00 PST

Intel® Xeon Phi™ Cluster Integration – A hands-on Introduction Space is limited.   Reserve your Webinar seat now at: https://www1.gotomeeting.com/register/471047704 Tuesday, September 16, 2014  8:00 AM - 12:00 AM PDT

September 24  8:00am-12:00 PST

Intel® Xeon Phi™ Cluster Integration – A hands-on Introduction Space is limited.  Reserve your Webinar seat now at: https://www1.gotomeeting.com/register/477933208

Wednesday, September 24, 2014  8:00 AM - 12:00 PM PDT

[Thomas_M_]

Yes, I am very interested too to join this Web training about INtel MIC and Linux Cluster.

[BelindaLiviero]

We don't have any more sessions scheduled at this point, but if there is enough demand we can add more.     Please let us know if you want this training

[Fredrik_N_]

Oh yes.  Count me in please.

[BelindaLiviero]

We have added another date for training as there has been additional demand -- This will be on December 9 - sign up details below:

up

	Intel® Xeon Phi™ Cluster Integration – A hands-on Introduction   Join us for a Webinar on December 9     Space is limited. Reserve your Webinar seat now at: https://www1.gotomeeting.com/register/850206849   In this 4-hour course led by an instructor the attendants will learn how to integrate intel Xeon Phi coprocessors into a cluster.  Each attendant will connect via ssh to a cluster, and be able to run all necessary commands alongside the presentation. Title: Intel® Xeon Phi™ Cluster Integration – A hands-on Introduction Date: Tuesday, December 9, 2014 Time: 8:00 AM - 12:00 PM PST After registering you will receive a confirmation email containing information about joining the Webinar. System Requirements PC-based attendees Required: Windows® 8, 7, Vista, XP or 2003 Server   Mac®-based attendees Required: Mac OS® X 10.6 or newer   Mobile attendees Required: iPhone®, iPad®, Android™ phone or Android tablet

[BelindaLiviero]

there are seats available still for the December 9 webinar!   Please register if interested:

Join us for a Webinar on December 9

Space is limited.
Reserve your Webinar seat now at:
https://www1.gotomeeting.com/register/850206849