Stas and I added this and gave no time on Blues.

 

Thanks!

-John

 

From: allocations-admins <allocations-admins-bounces@lcrc.anl.gov> on behalf of "Bair, Raymond A." <rbair@anl.gov>
Reply-To: LCRC Allocations Admins <allocations-admins@lcrc.anl.gov>
Date: Friday, September 1, 2017 at 11:48 AM
To: LCRC Allocations Admins <allocations-admins@lcrc.anl.gov>
Subject: [allocations-admins] Request for interim 3DDES project allocation of 1M core hours on Bebop

 

The allocations committee seems OK with 3DDES request for 1M Bebop core-hours in Sept.  Please make the allocation, with the understanding that Ramesh will spend most of the time on KNL.

 

Regards,

 

Ray

 

-----------------------------------------

Ray Bair

Argonne National Laboratory

and the University of Chicago

 

From: "Wagner, Albert F." <wagner@anl.gov>
Date: Thursday, August 31, 2017 at 3:26 PM
To: Raymond Bair <rbair@anl.gov>
Subject: Re: [LCRC alloc-com] Request for interim 3DDES project allocation of 1M core hours on Bebop

 

Ray,

 

I am OK with this.

 

Al

 

From: alloc-com <alloc-com-bounces@lists.lcrc.anl.gov> on behalf of "Bair, Raymond A." <rbair@anl.gov>
Date: Thursday, August 31, 2017 at 3:21 PM
To: "alloc-com@lcrc.anl.gov" <alloc-com@lcrc.anl.gov>
Subject: Re: [LCRC alloc-com] Request for interim 3DDES project allocation of 1M core hours on Bebop

 

Are there any additional comments on this request?  Since he can spend most of the 1M core-hours time on KNL nodes, which are ore lightly used at this time, it should not load down Bebop.

 

Regards,

 

Ray

 

-----------------------------------------

Ray Bair

Argonne National Laboratory

and the University of Chicago

 

From: Raymond Bair <rbair@anl.gov>
Date: Wednesday, August 30, 2017 at 12:30 PM
To: "Pieper, Steven C." <spieper@anl.gov>, "alloc-com@lcrc.anl.gov" <alloc-com@lcrc.anl.gov>
Cc: Raymond Bair <rbair@anl.gov>
Subject: Re: [LCRC alloc-com] Request for interim 3DDES project allocation of 1M core hours on Bebop

 

Ramesh writes . . .

 

The current set of jobs that I have ready to go (in fact, I was running them during the EU period), have meshes partitioned for 4096 cores. So, they can be run either on 128x32 cores on (BDW) or 64x64 cores (on KNL) although the KNL would take a longer wall clock time. For my runs, I can do a 25% (BDW) 75%(KNL) split, since there are ways to accelerate the solution on the KNL nodes.

 

 

-----------------------------------------

Ray Bair

Argonne National Laboratory

and the University of Chicago

 

From: "Pieper, Steven C." <spieper@anl.gov>
Organization: Physics, Argonne National Laboratory
Reply-To: "Pieper, Steven C." <spieper@anl.gov>
Date: Wednesday, August 30, 2017 at 11:08 AM
To: Raymond Bair <rbair@anl.gov>
Subject: Re: [LCRC alloc-com] Request for interim 3DDES project allocation of 1M core hours on Bebop

 

I guess OK although it is ~1/16 of the total Broadwell hours
available for Sept.

Steve





Steven C. Pieper, Senior Physicist Emeritus:                spieper@anl.gov 
Argonne National Laboratory, Physics Division, Bldg. 203, Argonne, IL 60439
Phone:  630-252-4232         Fax -6008        Secretary, Debra Beres, -4100
http://www.phy.anl.gov/theory/staff/scp.html

On 08/30/17 11:02, Bair, Raymond A. wrote:

LCRC Allocations Committee,

 

Ramesh Balakrishnan would like to continue his early user work on Bebop, transitioning into a larger project in FY2018 (project 3DDES).  For the remainder of FY2017, her requests 1m core-hours.  See his email below.

 

Clearly the time is available on Bebop, and you will act on the full FY2018 proposal (also below) at your September meeting.  Do you concur with providing the 1M core-hour interim allocation?  Please respond by Thursday, August 31.

 

Regards,

 

Ray

 

-----------------------------------------

Ray Bair

Argonne National Laboratory

and the University of Chicago

 

From: Argonne Service Desk <argonne@service-now.com>
Reply-To: "argonne@service-now.com"<argonne@service-now.com>
Date: Monday, August 28, 2017 at 3:56 PM
To: Raymond Bair <rbair@anl.gov>, "Aithal, Shashikant M." <aithal@cels.anl.gov>
Subject: Request AREQ0175370 Watch List: Request for an interim 3DDES project allocation of 1M core hours on Bebop from (August 28, September 30], 2017 Queue: HPC (CE

 

 

You can reply directly to this email. Please enter your response above this line.


 

Hello,

 

You have been added to the watch list for this request and will receive updates as work on the ticket progresses.

 

If you have additional information that may help in fulfilling this request, please reply to this message,

 

Number: AREQ0175370

Customer: Ramesh Balakrishnan

Short Description: Request for an interim 3DDES project allocation of 1M core hours on Bebop from (August 28, September 30], 2017 Queue: HPC (CELS)

Details:

received from: bramesh@alcf.anl.gov

 

Hello All,

Based on my e-discussions with Shashi Aithal, I would like to request an allocation of 1M core hours on Bebop to continue my Early User work till September 30, 2017. I have submitted a project request for an allocation that I understand will start from October 1, 2017 (if the proposal is accepted). If possible, please grant me 1 TB of project space for this short term allocation of 1M core hours for the same project name, i.e. 3DDES. Do let me know if you have any questions.

 

Thanks,

Ramesh

 

Thank you,
Argonne Service Desk
630-252-9999, option 2
help@anl.gov

 

Ref:MSG1993806

 

 

 

 

THE FOLOLWING IS THE FY2018 REQEUST.

 

Hello,

 

A new project on the LCRC cluster has been requested.  Please forward

the information on to the LCRC Allocation sub-committee.

 

       Applicant's name:  Ramesh Balakrishnan 

Applicant's institution:  ANL 

   Applicant's division:  LCF 

           Project Name:  3DDES 

          Project title:  Detached Eddy Simulation of Separated High Reynolds Number Flows 

     Associated funding:  ASCR Funding via ALCF 

          Other Systems:  Theta, Mira, Bebop 

                Science:  I am requesting 10M core hours, for the whole year, on Bebop,  to develop detached eddy simulation models that can serve as a wall model for moderately separated flow in thick boundary layers. Examples of such flow are the atmospheric boundary layer (ABL) over complex terrain, large eddy breakup devices (LEBU), and separated flow past airfoils at high angles of attack (i.e. greater than 10 degrees). 

    Project description:  Preliminary simulations of the ABL over Complex Terrain have been carried out on Bebop during the Early User Period, where the PI had an opportunity to run and test out modified version(s) of OpenFOAM, SU2, and FDL3DI on Bebop. The simulations were run on meshes ranging from 28M grid points to 140M grid points, and on a maximum of 4096 cores of Bebop. The codes make use of MPI and can optionally use OpenMP. In addition, one can vectorize the code for the KNL nodes (in which case OpenMP is not needed at all). During the early user period, the vectorization and OpenMP options were not tested out. The flat MPI codes do scale well on both the KNL and Broadwell nodes, although the time-to-solution is about five to six times slower on the KNL nodes (when compared to the Broadwell nodes). This is something that I plan to investigate in more detail as part of this allocation request.  

   Industry partnership:  None at this time. 

            Project URL:   

   Requested allocation:  10000000 

                     Q1:  2500000 

                     Q2:  2500000 

                     Q3:  2500000 

                     Q4:  2500000 

          Justification:  modified OpenFOAM: 35,000 finite volumes per core (on smallest number of cores)

=======================================

cores               time/time_step

4096 (KNL)           27  seconds (64 cores/node x 64 nodes)

2048 (KNL)           43  seconds (64 cores/node x 32 nodes)

1024  (KNL)           75  seconds (64 cores/node x 16 nodes)

 

4096 (BDW)            2 seconds (32 cores/node x 128 nodes)

2048 (BDW)            5 seconds (32 cores/node x   64 nodes)

 

Please note that the 10M core hours (for the whole year) is requested on Bebop.

1024 (BDW)           12 seconds (32 cores/node x   32 nodes)

 

modified SU2: 12,000 finite volumes per core (on smallest # of cores)

=================================

 

solver             cores                  time/time_step

FV                   1024 (KNL)              11 seconds

FE_DG (p4)     1024 (KNL)              16 seconds (same # finite volumes/core)

 

FV                    2048 (KNL)              8 seconds

FE_DG (p4)     2048 (KNL)             11 seconds  (same # finite volumes per core)     

   Storage requirements:  5TB

I will be generating quite a few time slices of the simulation. 

 

 







_______________________________________________
alloc-com mailing list
alloc-com@lists.lcrc.anl.gov
https://lists.lcrc.anl.gov/mailman/listinfo/alloc-com