Did that project buy the license? ----------------------------------------- Ray Bair Computing, Environment, and Life Sciences Argonne National Laboratory and the University of Chicago TCS Building 240, Room 4122 9700 South Cass Avenue Argonne, IL 60439 email: rbair(at)anl.gov Phone: (630)252-5751 On 5/27/15, 9:03 AM, "Low, John J." <[email protected]> wrote:
Ray,
Amber requires a $2000 license fee for computing centers in nonprofits.
John J. Low Principal Computational Science Specialist Computing, Environment and Life Sciences Building 240, 2143 9700 South Cass Avenue Argonne National Laboratory Argonne, IL 60439. 630-252-0045 www.linkedin.com/pub/john-low/15/8b0/5aa/
________________________________________ From: [email protected] [[email protected]] on behalf of Bair, Raymond A. [[email protected]] Sent: Wednesday, May 27, 2015 8:44 AM To: LCRC Allocations Admins Subject: Re: [allocations-admins] [LCRC Accounts] Project Allocation Request
John,
Did you see their comments about how the polarization model in Amber worked better for them?
----------------------------------------- Ray Bair Computing, Environment, and Life Sciences Argonne National Laboratory and the University of Chicago TCS Building 240, Room 4122 9700 South Cass Avenue Argonne, IL 60439 email: rbair(at)anl.gov Phone: (630)252-5751
On 5/26/15, 4:31 PM, "Low, John J." <[email protected]> wrote:
Ray,
The authors of GROMACS are claiming a 3-5 speedup on gpus over a 6-core haswell desktop (like my home system). Ellis, et al. could use the gpu nodes on Blues or Fusion nodes. These nodes don’t seem to be heavily used.
See http://www.gromacs.org/GPU_acceleration.
The GROMACS GPU benchmarks show good performance for models containing 100,000 or more atoms. If he is planning to use a model of that size, then he could benefit by using the GPUs.
John J. Low Principal Computational Science Specialist Computing, Environment and Life Sciences Building 240, 2143 9700 South Cass Avenue Argonne National Laboratory Argonne, IL 60439. 630-252-0045 www.linkedin.com/pub/john-low/15/8b0/5aa/
-----Original Message----- From: <Bair>, "Bair, Raymond A." <[email protected]> Reply-To: LCRC Allocations Admins <[email protected]> Date: Tuesday, May 26, 2015 at 10:58 AM To: "Ellis, Ross J." <[email protected]> Cc: LCRC Allocations Admins <[email protected]> Subject: Re: [allocations-admins] [LCRC Accounts] Project Allocation Request
Dear Ross,
At this time Blues is very full, so we have been asking projects to move to Fusion when they request large amounts of additional time in FY2015. Can your team use Fusion (8 cores per node) to do the proposed calculations?
Also, for this size of a request we will need to poll the LCRC Allocations Committee. FYI, Al Wagner is your point of contact on this committee.
Regards,
Ray
----------------------------------------- Ray Bair Computing, Environment, and Life Sciences Argonne National Laboratory and the University of Chicago TCS Building 240, Room 4122 9700 South Cass Avenue Argonne, IL 60439 email: rbair(at)anl.gov Phone: (630)252-5751
On 5/26/15, 10:46 AM, "[email protected]" <[email protected]> wrote:
Hello,
A change in allocation has been requested:
Requester: ellisrj (Ross Ellis) Project: DMDOHEMA_1 Title: Metal Ion Transport through Nano-Structured Lipophilic Systems: Atomistic MD Simulations Description: In the last few months, we have performed some preliminary simulation runs in understanding the supra-molecular aggregate structure in organic phases under the project DMDOHEMA_1 on LCRC. Some findings have been unveiled, e.g., the presences of water, acid, Eu(NO3)3 all affect the supra-molecular aggregate structure of extractant (i.e., DMDOHEMA) in organic phased. These findings agree with the existing experimental data. However, due to the small size of those simulation systems, the aggregation behaviors are not converged well. Therefore, in the renewal period of project DMDOHEMA_1, we are aiming to enlarge our simulation boxes so that such findings can be solidified. Moreover, one already finished large simulation supports that larger simulation box decreases the calculated error bar of the aggregation number of DMDOHEMA. With the simulations on large enough simulation boxes, the effects of water, acid, Eu(NO3)3 will be distinguished, which will surely provi de significant insight for the rational design of higher efficient extractants. Our project is expected to highly spark the development of rare earth element related technologies. In this project, classical molecular dynamics (MD) simulations will be performed at the all-atom resolution, in addition to the concurrent solvent extraction experiments at ANL. All-atom MD simulations have been well proven in predicting supra-molecular aggregate structures. The open source package GROMACS will be employed for the simulations. GROMACS is one of the highly recommended packages for MD simulations. GROMACS has been reported to be able to scale well up to around 1,000 CPU processors. Based on our experience on LCRC, the simulations can be LINEARLY scaled up to 256 CPU cores, under which condition the simulation speed is around 3 hours/ns for a system of the size of around 17 Ë 17 Ë 17 = 5000 nm3. Two users will use this allocation time, Dr. Ross J. Ellis and Dr. Baofu Qiao, both at the CSE division at ANL. In the first phase of this project, we are studying the influence of a variety of components, i.e., water, acid, Eu(NO3)3, on the aggregate structures of DMDOHEMA extractants dissolved in n-heptane organic phases. As aforementioned, this project is currently ongoing. With the allocation from LCRC, we are able to collects some conclusions in the near fure. In the second phase, we plan to study the effect of extractant on the supra-molecula aggregate structure. The extraction of DMDBTDMA, which is the preceding extractant of DMDOHEMA, will be simulated and compared with the simulation results of DMDOHEMA systems. According to the available experimental data, the effects of some factors will be explore, such as the concentration of water, the concentration of Eu(NO3)3 complexes, the concentration of acidic molecule of HNO3.
Current: undetermined amount Justification: Table. Benchmark of the simulation speed on LCRC #CPU cores 16 32 64 256 Speed (ns/day) 0.51 1.0 2.0 7.7
Based on the benchmark simulations, we can conclude that the scaling on LCRC can be linearly up to at least 256 CPU cores for our simulations.
Requested: 400000
A specific reason has been given: We are currently using the AMBER molecular dynamics simulation package for polarizable simulations. We have observed some better agreements with experimental data using polarazable force field. Under AMBER, the program sander.MPI is used for the polarizable simulations. Based on our tests, the optimal performance is obtained when using 6 node with 16 cores per node. See the following for the results. With these test results, a total amount of allocation of 400,000 core hours is requests.
#node (ppn=16): 2 3 4 5 6 7 8 speed (ns/day) : 0.56 0.83 0.78 1.14 1.33 1.39 0.99
This needs to be approved and the final allocation amount decided upon.
Thank You, The LCRC Accounts System
_______________________________________________ allocations-admins mailing list [email protected] https://lists.lcrc.anl.gov/mailman/listinfo/allocations-admins
_______________________________________________ allocations-admins mailing list [email protected] https://lists.lcrc.anl.gov/mailman/listinfo/allocations-admins
_______________________________________________ allocations-admins mailing list [email protected] https://lists.lcrc.anl.gov/mailman/listinfo/allocations-admins _______________________________________________ allocations-admins mailing list [email protected] https://lists.lcrc.anl.gov/mailman/listinfo/allocations-admins