Re: [gmx-users] Water tutorial
Hi Arpita http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin/gmx-tutorials/lysozyme/ http://www.gromacs.org/Documentation/Tutorials Cheers Carmen - Original Message - From: "Arpita Srivastava" To: ; Sent: Thursday, January 28, 2016 3:57 PM Subject: [gmx-users] Water tutorial Dear Sir, I am a new user of GROMACS software.I need a tutorial to understand the working of software. Please send me a tutorial of putting water molecules in a box and simulating the system. Thank you. -- Gromacs Users mailing list * Please search the archive at http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before posting! * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists * For (un)subscribe requests visit https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or send a mail to gmx-users-requ...@gromacs.org. -- Gromacs Users mailing list * Please search the archive at http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before posting! * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists * For (un)subscribe requests visit https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or send a mail to gmx-users-requ...@gromacs.org.
Re: [gmx-users] GPU low performance
I report the changes made to improve the performance of a molecular dynamics on a protein of 1925 running on GPU INVIDIA K20 Tesla : a.. To limit the number of cores used in the calculation (option -pin on) and to have a better performance: "gmx_mpi mdrun ... -ntomp 16 -pin on" where ntomp is the number of OpenMP threads a.. clock frequency using the NVDIA management tool is been increased from the default 705 MHz to 758 MHz. a.. to reduce runtime to calculate energies every step in mdp file: nstcalcenergy option = -1 The actual performance is about 7ns /day against 2ns/day without these changes. Carmen - Original Message - From: "Szilárd Páll" To: "Carmen Di Giovanni" Cc: "Discussion list for GROMACS users" Sent: Friday, February 20, 2015 1:25 AM Subject: Re: [gmx-users] GPU low performance Please consult the manual an wiki. -- Szilárd On Thu, Feb 19, 2015 at 6:44 PM, Carmen Di Giovanni wrote: Szilard, about: Fatal error 1) Setting the number of thread-MPI threads is only supported with thread-MPI and Gromacs was compiled without thread-MPI For more information and tips for troubleshooting, please check the GROMACS website at http://www.gromacs.org/Documentation/Errors --- The error is quite clearly explains that you're trying to use mdrun's built-in thread-MPI parallelization, but you have a binary that does not support it. Use the MPI launching syntax instead. Can you help me about the MPI launching syntax? What is the suitable command ? A previous poster has already pointed you to the "Acceleration and parallelization" page which, I believe describes the matter in detail. 2) Have you looked at the the performance table at the end of the log? You are wasting a large amount of runtime calculating energies every step and this overhead comes in multiple places in the code - one of them being the non-timed code parts which typically take <3%. As can I reduce runtime to calculate the energies every step? I must to modify something in mdp file ? This is discussed throughly in the manual, you should be looking for the nstcalcenergy option. Thank you in advance Carmen -- Carmen Di Giovanni, PhD Dept. of Pharmaceutical and Toxicological Chemistry "Drug Discovery Lab" University of Naples "Federico II" Via D. Montesano, 49 80131 Naples Tel.: ++39 081 678623 Fax: ++39 081 678100 Email: cdigi...@unina.it Quoting Szilárd Páll : On Thu, Feb 19, 2015 at 11:32 AM, Carmen Di Giovanni wrote: Dear Szilárd, 1) the output of command nvidia-smi -ac 2600,758 is [root@localhost test_gpu]# nvidia-smi -ac 2600,758 Applications clocks set to "(MEM 2600, SM 758)" for GPU :03:00.0 Warning: persistence mode is disabled on this device. This settings will go back to default as soon as driver unloads (e.g. last application like nvidia-smi or cuda application terminates). Run with [--help | -h] switch to get more information on how to enable persistence mode. run nvidia-smi -pm 1 if you want to avoid that. Setting applications clocks is not supported for GPU :82:00.0. Treating as warning and moving on. All done. 2) I decreased nlists to 20 However when I do the command: gmx_mpi mdrun -deffnm nvt -ntmpi 8 -gpu_id give me a fatal error: GROMACS: gmx mdrun, VERSION 5.0 Executable: /opt/SW/gromacs-5.0/build/mpi-cuda/bin/gmx_mpi Library dir: /opt/SW/gromacs-5.0/share/top Command line: gmx_mpi mdrun -deffnm nvt -ntmpi 8 -gpu_id Back Off! I just backed up nvt.log to ./#nvt.log.8# Reading file nvt.tpr, VERSION 5.0 (single precision) Changing nstlist from 10 to 40, rlist from 1 to 1.097 --- Program gmx_mpi, VERSION 5.0 Source code file: /opt/SW/gromacs-5.0/src/programs/mdrun/runner.c, line: 876 Fatal error: Setting the number of thread-MPI threads is only supported with thread-MPI and Gromacs was compiled without thread-MPI For more information and tips for troubleshooting, please check the GROMACS website at http://www.gromacs.org/Documentation/Errors --- The error is quite clearly explains that you're trying to use mdrun's built-in thread-MPI parallelization, but you have a binary that does not support it. Use the MPI launching syntax instead. Halting program gmx_mpi gcq#223: "Jesus Not Only Saves, He Also Frequently Makes Backups." (Myron Bradshaw) -- MPI_ABORT was invoked on rank 0 in communicator MPI_COMM_WORLD with errorcode -1. NOTE: invoking MPI_ABORT causes Open MPI to kill all MPI processes. You may or may not see output from other processes, depending on exactly when Open MPI kills them.
Re: [gmx-users] GTX980 performance
Szilárd thank you for the useful adivices about the configuration of the new server machine. I report the changes made to improve the performance of a molecular dynamics on a protein of 1925 running on GPU INVIDIA K20 Tesla : a.. To limit the number of cores used in the calculation (option -pin on) and to have a better performance: "gmx_mpi mdrun ... -ntomp 16 -pin on" where ntomp is the number of OpenMP threads a.. clock frequency using the NVDIA management tool is been increased from the default 705 MHz to 758 MHz. a.. to reduce runtime to calculate energies every step in mdp file: nstcalcenergy option = -1 The actual performance is about 7ns /day against 2ns/day without these changes. Carmen - Original Message - From: "Szilárd Páll" To: "Carmen Di Giovanni" Cc: "Discussion list for GROMACS users" Sent: Thursday, February 26, 2015 2:37 PM Subject: Re: GTX980 performance On Wed, Feb 25, 2015 at 1:21 PM, Carmen Di Giovanni wrote: A special thank you to Szilárd Páll for the good advices in "GPU low performance discussion". The perfomance calculation is much improved after its suggestions. I'm glad it helped. Could you post the changes you made to your mdp/command line and the results these gave? It would allow others to learn from it. Dear GROMACS users and developers, we are thinking to buy a new tyan server machine with these features: SERVER SYSTEM TYAN FT48 - Tower/Rack,Dual Xeon ,8xSATA N. 2 CPU INTEL XEON E5-2620 2.0Ghz - 6 CORE 15MCache LGA2011 N. 4 Scheda Video Nvidia GTX980, 4GB GDDR5,PCIE 3.0 , 2DVI, HD N. 4 DDR3 8 GB 1600 Mhz HARD DISK 1 TB SATA 3 WD I known that the GTX980 offer good performance for GROMACS 5.0 What are your views about this ? That CPU-GPU combination will give heavily CPU-bound GROMACS runs, those GTX 980s are 1.5-2x faster than what you can use with those CPUs - conversely, if you'd get 6-core 3 GHz CPUs, you'll see a huge, nearly 50% improvement in performance. This will change in the futur, but at least with GROMACS v5.1 and earlier, the performance on this machine won't be much higher than with a single fast CPU and one GTX 980. For better performance with GROMACS, consider getting better CPUs in this machine or for the same (or less) money get two workstations with i7 4930K or 4960X CPUs. -- Szilárd Thank you in advance Carmen Carmen Di Giovanni, PhD Postdoctoral Researcher Dept. of Pharmacy "Drug Discovery Lab" University of Naples "Federico II" Via D. Montesano, 49 80131 Naples Tel.: ++39 081 678623 Fax: ++39 081 678100 Email: cdigi...@unina.it -- Gromacs Users mailing list * Please search the archive at http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before posting! * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists * For (un)subscribe requests visit https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or send a mail to gmx-users-requ...@gromacs.org.
Re: [gmx-users] GTX980 performance
Szilárd thank you for the useful adivices about the configuration of the new server machine. I report the changes made to improve the performance of a molecular dynamics on a protein of 1925 running on GPU INVIDIA K20 Tesla : a.. To limit the number of cores used in the calculation (option -pin on) and to have a better performance: "gmx_mpi mdrun ... -ntomp 16 -pin on" where ntomp is the number of OpenMP threads a.. clock frequency using the NVDIA management tool is been increased from the default 705 MHz to 758 MHz. a.. to reduce runtime to calculate energies every step in mdp file: nstcalcenergy option = -1 The actual performance is about 7ns /day against 2ns/day without these changes. Carmen - Original Message - From: "Szilárd Páll" To: "Carmen Di Giovanni" Cc: "Discussion list for GROMACS users" Sent: Thursday, February 26, 2015 2:37 PM Subject: Re: GTX980 performance On Wed, Feb 25, 2015 at 1:21 PM, Carmen Di Giovanni wrote: A special thank you to Szilárd Páll for the good advices in "GPU low performance discussion". The perfomance calculation is much improved after its suggestions. I'm glad it helped. Could you post the changes you made to your mdp/command line and the results these gave? It would allow others to learn from it. Dear GROMACS users and developers, we are thinking to buy a new tyan server machine with these features: SERVER SYSTEM TYAN FT48 - Tower/Rack,Dual Xeon ,8xSATA N. 2 CPU INTEL XEON E5-2620 2.0Ghz - 6 CORE 15MCache LGA2011 N. 4 Scheda Video Nvidia GTX980, 4GB GDDR5,PCIE 3.0 , 2DVI, HD N. 4 DDR3 8 GB 1600 Mhz HARD DISK 1 TB SATA 3 WD I known that the GTX980 offer good performance for GROMACS 5.0 What are your views about this ? That CPU-GPU combination will give heavily CPU-bound GROMACS runs, those GTX 980s are 1.5-2x faster than what you can use with those CPUs - conversely, if you'd get 6-core 3 GHz CPUs, you'll see a huge, nearly 50% improvement in performance. This will change in the futur, but at least with GROMACS v5.1 and earlier, the performance on this machine won't be much higher than with a single fast CPU and one GTX 980. For better performance with GROMACS, consider getting better CPUs in this machine or for the same (or less) money get two workstations with i7 4930K or 4960X CPUs. -- Szilárd Thank you in advance Carmen Carmen Di Giovanni, PhD Postdoctoral Researcher Dept. of Pharmacy "Drug Discovery Lab" University of Naples "Federico II" Via D. Montesano, 49 80131 Naples Tel.: ++39 081 678623 Fax: ++39 081 678100 Email: cdigi...@unina.it -- Gromacs Users mailing list * Please search the archive at http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before posting! * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists * For (un)subscribe requests visit https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or send a mail to gmx-users-requ...@gromacs.org.
[gmx-users] GTX980 performance
A special thank you to Szilárd Páll for the good advices in "GPU low performance discussion". The perfomance calculation is much improved after its suggestions. Dear GROMACS users and developers, we are thinking to buy a new tyan server machine with these features: a.. SERVER SYSTEM TYAN FT48 - Tower/Rack,Dual Xeon ,8xSATA a.. N. 2 CPU INTEL XEON E5-2620 2.0Ghz - 6 CORE 15MCache LGA2011 a.. N. 4 Scheda Video Nvidia GTX980, 4GB GDDR5,PCIE 3.0 , 2DVI, HD a.. N. 4 DDR3 8 GB 1600 Mhz b.. HARD DISK 1 TB SATA 3 WD I known that the GTX980 offer good performance for GROMACS 5.0 What are your views about this ? Thank you in advance Carmen Carmen Di Giovanni, PhD Postdoctoral Researcher Dept. of Pharmacy "Drug Discovery Lab" University of Naples "Federico II" Via D. Montesano, 49 80131 Naples Tel.: ++39 081 678623 Fax: ++39 081 678100 Email: cdigi...@unina.it -- Gromacs Users mailing list * Please search the archive at http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before posting! * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists * For (un)subscribe requests visit https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or send a mail to gmx-users-requ...@gromacs.org.
Re: [gmx-users] GPU low performance
Szilard, about: Fatal error 1) Setting the number of thread-MPI threads is only supported with thread-MPI and Gromacs was compiled without thread-MPI For more information and tips for troubleshooting, please check the GROMACS website at http://www.gromacs.org/Documentation/Errors --- The error is quite clearly explains that you're trying to use mdrun's built-in thread-MPI parallelization, but you have a binary that does not support it. Use the MPI launching syntax instead. Can you help me about the MPI launching syntax? What is the suitable command ? 2) Have you looked at the the performance table at the end of the log? You are wasting a large amount of runtime calculating energies every step and this overhead comes in multiple places in the code - one of them being the non-timed code parts which typically take <3%. As can I reduce runtime to calculate the energies every step? I must to modify something in mdp file ? Thank you in advance Carmen -- Carmen Di Giovanni, PhD Dept. of Pharmaceutical and Toxicological Chemistry "Drug Discovery Lab" University of Naples "Federico II" Via D. Montesano, 49 80131 Naples Tel.: ++39 081 678623 Fax: ++39 081 678100 Email: cdigi...@unina.it Quoting Szilárd Páll : On Thu, Feb 19, 2015 at 11:32 AM, Carmen Di Giovanni wrote: Dear Szilárd, 1) the output of command nvidia-smi -ac 2600,758 is [root@localhost test_gpu]# nvidia-smi -ac 2600,758 Applications clocks set to "(MEM 2600, SM 758)" for GPU :03:00.0 Warning: persistence mode is disabled on this device. This settings will go back to default as soon as driver unloads (e.g. last application like nvidia-smi or cuda application terminates). Run with [--help | -h] switch to get more information on how to enable persistence mode. run nvidia-smi -pm 1 if you want to avoid that. Setting applications clocks is not supported for GPU :82:00.0. Treating as warning and moving on. All done. 2) I decreased nlists to 20 However when I do the command: gmx_mpi mdrun -deffnm nvt -ntmpi 8 -gpu_id give me a fatal error: GROMACS: gmx mdrun, VERSION 5.0 Executable: /opt/SW/gromacs-5.0/build/mpi-cuda/bin/gmx_mpi Library dir: /opt/SW/gromacs-5.0/share/top Command line: gmx_mpi mdrun -deffnm nvt -ntmpi 8 -gpu_id Back Off! I just backed up nvt.log to ./#nvt.log.8# Reading file nvt.tpr, VERSION 5.0 (single precision) Changing nstlist from 10 to 40, rlist from 1 to 1.097 --- Program gmx_mpi, VERSION 5.0 Source code file: /opt/SW/gromacs-5.0/src/programs/mdrun/runner.c, line: 876 Fatal error: Setting the number of thread-MPI threads is only supported with thread-MPI and Gromacs was compiled without thread-MPI For more information and tips for troubleshooting, please check the GROMACS website at http://www.gromacs.org/Documentation/Errors --- The error is quite clearly explains that you're trying to use mdrun's built-in thread-MPI parallelization, but you have a binary that does not support it. Use the MPI launching syntax instead. Halting program gmx_mpi gcq#223: "Jesus Not Only Saves, He Also Frequently Makes Backups." (Myron Bradshaw) -- MPI_ABORT was invoked on rank 0 in communicator MPI_COMM_WORLD with errorcode -1. NOTE: invoking MPI_ABORT causes Open MPI to kill all MPI processes. You may or may not see output from other processes, depending on exactly when Open MPI kills them. - 4) I don't understand as I can reduce the "Rest" time Have you looked at the the performance table at the end of the log? You are wasting a large amount of runtime calculating energies every step and this overhead comes in multiple places in the code - one of them being the non-timed code parts which typically take <3%. Cheers, -- Szilard Carmen -- Carmen Di Giovanni, PhD Dept. of Pharmaceutical and Toxicological Chemistry "Drug Discovery Lab" University of Naples "Federico II" Via D. Montesano, 49 80131 Naples Tel.: ++39 081 678623 Fax: ++39 081 678100 Email: cdigi...@unina.it Quoting Szilárd Páll : Please keep the mails on the list. On Wed, Feb 18, 2015 at 6:32 PM, Carmen Di Giovanni wrote: nvidia-smi -q -g 0 ==NVSMI LOG== Timestamp : Wed Feb 18 18:30:01 2015 Driver Version : 340.24 Attached GPUs : 2 GPU :03:00.0 Product Name: Tesla K20c [... Clocks Graphics: 705 MHz SM : 705 MHz Memory
Re: [gmx-users] GPU low performance
-- Carmen Di Giovanni, PhD Dept. of Pharmaceutical and Toxicological Chemistry "Drug Discovery Lab" University of Naples "Federico II" Via D. Montesano, 49 80131 Naples Tel.: ++39 081 678623 Fax: ++39 081 678100 Email: cdigi...@unina.it Hi Szilard, I post also the output of the command: gmx_mpi mdrun -deffnm nvt -nb gpu -ntomp 16 -pin on . Back Off! I just backed up nvt.log to ./#nvt.log.10# Reading file nvt.tpr, VERSION 5.0 (single precision) Changing nstlist from 10 to 40, rlist from 1 to 1.097 Using 1 MPI process Using 16 OpenMP threads 2 GPUs detected on host localhost.localdomain: #0: NVIDIA Tesla K20c, compute cap.: 3.5, ECC: yes, stat: compatible #1: NVIDIA GeForce GTX 650, compute cap.: 3.0, ECC: no, stat: compatible 1 GPU auto-selected for this run. Mapping of GPU to the 1 PP rank in this node: #0 NOTE: potentially sub-optimal launch configuration, gmx_mpi started with less PP MPI process per node than GPUs available. Each PP MPI process can use only one GPU, 1 GPU per node will be used ---- Quoting Carmen Di Giovanni : Dear all, the full log file is too big. However in the middle part of it, there are only informations about the energies at each time. The first part is alrady posted. So I post the final part of it: - Step Time Lambda 10002.00.0 Writing checkpoint, step 1000 at Mon Dec 29 13:16:22 2014 Energies (kJ/mol) G96AngleProper Dih. Improper Dih. LJ-14 Coulomb-14 9.34206e+034.14342e+032.79172e+03 -1.75465e+027.99811e+04 LJ (SR) Coulomb (SR) Coul. recip. PotentialKinetic En. 1.01135e+06 -7.13064e+062.01349e+04 -6.00306e+061.08201e+06 Total Energy Conserved En.Temperature Pressure (bar) Constr. rmsd -4.92106e+06 -5.86747e+062.99426e+021.29480e+022.16280e-05 <== ### ==> < A V E R A G E S > <== ### ==> Statistics over 1001 steps using 1001 frames Energies (kJ/mol) G96AngleProper Dih. Improper Dih. LJ-14 Coulomb-14 9.45818e+034.30665e+032.92407e+03 -1.75556e+028.02473e+04 LJ (SR) Coulomb (SR) Coul. recip. PotentialKinetic En. 1.01284e+06 -7.13138e+062.01510e+04 -6.00163e+061.08407e+06 Total Energy Conserved En.Temperature Pressure (bar) Constr. rmsd -4.91756e+06 -5.38519e+062.8e+021.37549e+020.0e+00 Total Virial (kJ/mol) 3.42887e+051.63625e+011.23658e+02 1.67406e+013.42916e+05 -4.27834e+01 1.23997e+02 -4.29636e+013.42881e+05 Pressure (bar) 1.37573e+027.50214e-02 -1.03916e-01 7.22048e-021.37623e+02 -1.66417e-02 -1.06444e-01 -1.52990e-021.37453e+02 M E G A - F L O P S A C C O U N T I N G NB=Group-cutoff nonbonded kernelsNxN=N-by-N cluster Verlet kernels RF=Reaction-Field VdW=Van der Waals QSTab=quadratic-spline table W3=SPC/TIP3p W4=TIP4p (single or pairs) V&F=Potential and force V=Potential only F=Force only Computing: M-Number M-Flops % Flops - Pair Search distance check16343508.605344 147091577.448 0.0 NxN Ewald Elec. + LJ [V&F] 5072118956.506304 542716728346.17498.1 1,4 nonbonded interactions 95860.009586 8627400.863 0.0 Calc Weights 13039741.303974 469430686.943 0.1 Spread Q Bspline 278181147.818112 556362295.636 0.1 Gather F Bspline 278181147.818112 1669086886.909 0.3 3D-FFT 880787450.909824 7046299607.279 1.3 Solve PME 163837.90950410485626.208 0.0 Shift-X 108664.934658 651989.608 0.0 Angles 86090.00860914463121.446 0.0 Propers 31380.003138 7186020.719 0.0 Impropers28790.002879 5988320.599 0.0 Virial 4347030.43470378246547.825 0.0 Stop-CM4346580.86931643465808.693 0.0 Calc-Ekin 4346580.869316 117357683.472 0.0 Lincs59130.017739 3547801.064 0.0 Lincs-Mat 1033080.309924 4132321.240 0.0 Constraint-V 4406580.88131635252647.051 0.0 Constraint-Vir 4347450.434745 104338810.434 0.0 Set
Re: [gmx-users] GPU low performance
Dear Szilárd, 1) the output of command nvidia-smi -ac 2600,758 is [root@localhost test_gpu]# nvidia-smi -ac 2600,758 Applications clocks set to "(MEM 2600, SM 758)" for GPU :03:00.0 Warning: persistence mode is disabled on this device. This settings will go back to default as soon as driver unloads (e.g. last application like nvidia-smi or cuda application terminates). Run with [--help | -h] switch to get more information on how to enable persistence mode. Setting applications clocks is not supported for GPU :82:00.0. Treating as warning and moving on. All done. 2) I decreased nlists to 20 However when I do the command: gmx_mpi mdrun -deffnm nvt -ntmpi 8 -gpu_id give me a fatal error: GROMACS: gmx mdrun, VERSION 5.0 Executable: /opt/SW/gromacs-5.0/build/mpi-cuda/bin/gmx_mpi Library dir: /opt/SW/gromacs-5.0/share/top Command line: gmx_mpi mdrun -deffnm nvt -ntmpi 8 -gpu_id Back Off! I just backed up nvt.log to ./#nvt.log.8# Reading file nvt.tpr, VERSION 5.0 (single precision) Changing nstlist from 10 to 40, rlist from 1 to 1.097 --- Program gmx_mpi, VERSION 5.0 Source code file: /opt/SW/gromacs-5.0/src/programs/mdrun/runner.c, line: 876 Fatal error: Setting the number of thread-MPI threads is only supported with thread-MPI and Gromacs was compiled without thread-MPI For more information and tips for troubleshooting, please check the GROMACS website at http://www.gromacs.org/Documentation/Errors --- Halting program gmx_mpi gcq#223: "Jesus Not Only Saves, He Also Frequently Makes Backups." (Myron Bradshaw) -- MPI_ABORT was invoked on rank 0 in communicator MPI_COMM_WORLD with errorcode -1. NOTE: invoking MPI_ABORT causes Open MPI to kill all MPI processes. You may or may not see output from other processes, depending on exactly when Open MPI kills them. - 3) I don't understand as I can reduce the "Rest" time Carmen -- Carmen Di Giovanni, PhD Dept. of Pharmaceutical and Toxicological Chemistry "Drug Discovery Lab" University of Naples "Federico II" Via D. Montesano, 49 80131 Naples Tel.: ++39 081 678623 Fax: ++39 081 678100 Email: cdigi...@unina.it - Original Message - From: "Szilárd Páll" To: "Discussion list for GROMACS users" ; "Carmen Di Giovanni" Sent: Wednesday, February 18, 2015 6:38 PM Subject: Re: [gmx-users] GPU low performance Please keep the mails on the list. On Wed, Feb 18, 2015 at 6:32 PM, Carmen Di Giovanni wrote: nvidia-smi -q -g 0 ==NVSMI LOG== Timestamp : Wed Feb 18 18:30:01 2015 Driver Version : 340.24 Attached GPUs : 2 GPU :03:00.0 Product Name: Tesla K20c [... Clocks Graphics: 705 MHz SM : 705 MHz Memory : 2600 MHz Applications Clocks Graphics: 705 MHz Memory : 2600 MHz Default Applications Clocks Graphics: 705 MHz Memory : 2600 MHz Max Clocks Graphics: 758 MHz SM : 758 MHz Memory : 2600 MHz This is the relevant part I was looking for. The Tesla K20c supports setting a so-called application clock which is essentially means that you can bump its clock frequency using the NVDIA management tool nvidia-smi from the default 705 MHz to 758 MHz. Use the command: nvidia-smi -ac 2600,758 This should give you another 7% or so (I didn't remember the correct max clock before, that's why I guessing 5%). Cheers, Szilard Clock Policy Auto Boost : N/A Auto Boost Default : N/A Compute Processes Process ID : 19441 Name: gmx_mpi Used GPU Memory : 110 MiB [carmendigi@localhost test_gpu]$ -- Carmen Di Giovanni, PhD Dept. of Pharmaceutical and Toxicological Chemistry "Drug Discovery Lab" University of Naples "Federico II" Via D. Montesano, 49 80131 Naples Tel.: ++39 081 678623 Fax: ++39 081 678100 Email: cdigi...@unina.it Quoting Szilárd Páll : As I suggested above please use pastebin.com or similar! -- Szilárd On Wed, Feb 18, 2015 at 6:09 PM, Carmen Di Giovanni wrote: Dear Szilàrd, it's not possible attach the full log file in the forum mail because it is too big. I send it by your private mail address. Thank you in advance Ca
Re: [gmx-users] GPU low performance
Dear Szilárd, 1) the output of command nvidia-smi -ac 2600,758 is [root@localhost test_gpu]# nvidia-smi -ac 2600,758 Applications clocks set to "(MEM 2600, SM 758)" for GPU :03:00.0 Warning: persistence mode is disabled on this device. This settings will go back to default as soon as driver unloads (e.g. last application like nvidia-smi or cuda application terminates). Run with [--help | -h] switch to get more information on how to enable persistence mode. Setting applications clocks is not supported for GPU :82:00.0. Treating as warning and moving on. All done. 2) I decreased nlists to 20 However when I do the command: gmx_mpi mdrun -deffnm nvt -ntmpi 8 -gpu_id give me a fatal error: GROMACS: gmx mdrun, VERSION 5.0 Executable: /opt/SW/gromacs-5.0/build/mpi-cuda/bin/gmx_mpi Library dir: /opt/SW/gromacs-5.0/share/top Command line: gmx_mpi mdrun -deffnm nvt -ntmpi 8 -gpu_id Back Off! I just backed up nvt.log to ./#nvt.log.8# Reading file nvt.tpr, VERSION 5.0 (single precision) Changing nstlist from 10 to 40, rlist from 1 to 1.097 --- Program gmx_mpi, VERSION 5.0 Source code file: /opt/SW/gromacs-5.0/src/programs/mdrun/runner.c, line: 876 Fatal error: Setting the number of thread-MPI threads is only supported with thread-MPI and Gromacs was compiled without thread-MPI For more information and tips for troubleshooting, please check the GROMACS website at http://www.gromacs.org/Documentation/Errors --- Halting program gmx_mpi gcq#223: "Jesus Not Only Saves, He Also Frequently Makes Backups." (Myron Bradshaw) -- MPI_ABORT was invoked on rank 0 in communicator MPI_COMM_WORLD with errorcode -1. NOTE: invoking MPI_ABORT causes Open MPI to kill all MPI processes. You may or may not see output from other processes, depending on exactly when Open MPI kills them. - 4) I don't understand as I can reduce the "Rest" time Carmen -- Carmen Di Giovanni, PhD Dept. of Pharmaceutical and Toxicological Chemistry "Drug Discovery Lab" University of Naples "Federico II" Via D. Montesano, 49 80131 Naples Tel.: ++39 081 678623 Fax: ++39 081 678100 Email: cdigi...@unina.it Quoting Szilárd Páll : Please keep the mails on the list. On Wed, Feb 18, 2015 at 6:32 PM, Carmen Di Giovanni wrote: nvidia-smi -q -g 0 ==NVSMI LOG== Timestamp : Wed Feb 18 18:30:01 2015 Driver Version : 340.24 Attached GPUs : 2 GPU :03:00.0 Product Name: Tesla K20c [... Clocks Graphics: 705 MHz SM : 705 MHz Memory : 2600 MHz Applications Clocks Graphics: 705 MHz Memory : 2600 MHz Default Applications Clocks Graphics: 705 MHz Memory : 2600 MHz Max Clocks Graphics: 758 MHz SM : 758 MHz Memory : 2600 MHz This is the relevant part I was looking for. The Tesla K20c supports setting a so-called application clock which is essentially means that you can bump its clock frequency using the NVDIA management tool nvidia-smi from the default 705 MHz to 758 MHz. Use the command: nvidia-smi -ac 2600,758 This should give you another 7% or so (I didn't remember the correct max clock before, that's why I guessing 5%). Cheers, Szilard Clock Policy Auto Boost : N/A Auto Boost Default : N/A Compute Processes Process ID : 19441 Name: gmx_mpi Used GPU Memory : 110 MiB [carmendigi@localhost test_gpu]$ -- Carmen Di Giovanni, PhD Dept. of Pharmaceutical and Toxicological Chemistry "Drug Discovery Lab" University of Naples "Federico II" Via D. Montesano, 49 80131 Naples Tel.: ++39 081 678623 Fax: ++39 081 678100 Email: cdigi...@unina.it Quoting Szilárd Páll : As I suggested above please use pastebin.com or similar! -- Szilárd On Wed, Feb 18, 2015 at 6:09 PM, Carmen Di Giovanni wrote: Dear Szilàrd, it's not possible attach the full log file in the forum mail because it is too big. I send it by your private mail address. Thank you in advance Carmen -- Carmen Di Giovanni, PhD Dept. of Pharmaceutical and Toxicological Chemistry "Drug Discovery Lab" University of Naples "Federico II" Via D. Montesano, 49 8
Re: [gmx-users] GPU low performance
2.8191107630.452 2.2 Constraints1 32 1001 34210.1422846293.908 5.8 Rest 92338.7817682613.897 15.6 - Total 593302.894 49362976.023 100.0 - Breakdown of PME mesh computation - PME spread/gather 1 32 2002 144767.207 12044674.424 24.4 PME 3D-FFT 1 32 2002 39499.1573286341.501 6.7 PME solve Elec 1 32 10019947.340 827621.589 1.7 - GPU timings - Computing: Count Wall t (s) ms/step % - Pair list H2D 250001 935.7513.743 0.2 X / q H2D 1001 11509.2091.151 2.8 Nonbonded F+ene k. 975 377111.949 38.67892.0 Nonbonded F+ene+prune k. 250001 12049.010 48.196 2.9 F D2H 10018129.2920.813 2.0 - Total 409735.211 40.974 100.0 - Force evaluation time GPU/CPU: 40.974 ms/24.437 ms = 1.677 For optimal performance this ratio should be close to 1! NOTE: The GPU has >20% more load than the CPU. This imbalance causes performance loss, consider using a shorter cut-off and a finer PME grid. Core t (s) Wall t (s)(%) Time: 18713831.228 593302.894 3154.2 6d20h48:22 (ns/day)(hour/ns) Performance:2.9138.240 Finished mdrun on rank 0 Mon Dec 29 13:16:24 2014 --- thank you in advance Carmen -- Carmen Di Giovanni, PhD Dept. of Pharmaceutical and Toxicological Chemistry "Drug Discovery Lab" University of Naples "Federico II" Via D. Montesano, 49 80131 Naples Tel.: ++39 081 678623 Fax: ++39 081 678100 Email: cdigi...@unina.it Quoting Szilárd Páll : We need a *full* log file, not parts of it! You can try running with "-ntomp 16 -pin on" - it may be a bit faster not not use HyperThreading. -- Szilárd On Wed, Feb 18, 2015 at 5:20 PM, Carmen Di Giovanni wrote: Justin, the problem is evident for all calculations. This is the log file of a recent run: Log file opened on Mon Dec 22 16:28:00 2014 Host: localhost.localdomain pid: 8378 rank ID: 0 number of ranks: 1 GROMACS:gmx mdrun, VERSION 5.0 GROMACS is written by: Emile Apol Rossen Apostolov Herman J.C. Berendsen Par Bjelkmar Aldert van Buuren Rudi van DrunenAnton Feenstra Sebastian Fritsch Gerrit GroenhofChristoph Junghans Peter Kasson Carsten Kutzner Per LarssonJustin A. Lemkul Magnus LundborgPieter Meulenhoff Erik Marklund Teemu Murtola Szilard Pall Sander Pronk Roland Schulz Alexey ShvetsovMichael Shirts Alfons Sijbers Peter Tieleman Christian Wennberg Maarten Wolf and the project leaders: Mark Abraham, Berk Hess, Erik Lindahl, and David van der Spoel Copyright (c) 1991-2000, University of Groningen, The Netherlands. Copyright (c) 2001-2014, The GROMACS development team at Uppsala University, Stockholm University and the Royal Institute of Technology, Sweden. check out http://www.gromacs.org for more information. GROMACS is free software; you can redistribute it and/or modify it under the terms of the GNU Lesser General Public License as published by the Free Software Foundation; either version 2.1 of the License, or (at your option) any later version. GROMACS: gmx mdrun, VERSION 5.0 Executable: /opt/SW/gromacs-5.0/build/mpi-cuda/bin/gmx_mpi Library dir: /opt/SW/gromacs-5.0/share/top Command line: gmx_mpi mdrun -deffnm prod_20ns Gromacs version:VERSION 5.0 Precision: single Memory model: 64 bit MPI library:MPI OpenMP support: enabled GPU support:enabled invsqrt routine:gmx_software_invsqrt(x) SIMD instructions: AVX_256 FFT library:fftw-3.3.3-sse2 RDTSCP usage: enabled C++11 compilation: disabled TNG support:enabled Tracing support:disabled Built on: Thu Jul 31 18:30:37 CEST 2014 Built by: root@localhost.localdomain [CMAKE] Build OS/arch: Linux 2.6.32-431.el6.x86_64 x86_64 Build CPU
Re: [gmx-users] GPU low performance
6e+032.97359e+03 -1.93107e+028.05534e+04 LJ (SR) Coulomb (SR) Coul. recip. PotentialKinetic En. 1.01340e+06 -7.13271e+062.01361e+04 -6.00175e+061.09887e+06 Total Energy Conserved En.Temperature Pressure (bar) Constr. rmsd -4.90288e+06 -4.90288e+063.04092e+021.70897e+022.16683e-05 step 80: timed with pme grid 128 128 128, coulomb cutoff 1.200: 6279.0 M-cycles step 160: timed with pme grid 112 112 112, coulomb cutoff 1.306: 6962.2 M-cycles step 240: timed with pme grid 100 100 100, coulomb cutoff 1.463: 8406.5 M-cycles step 320: timed with pme grid 128 128 128, coulomb cutoff 1.200: 6424.0 M-cycles step 400: timed with pme grid 120 120 120, coulomb cutoff 1.219: 6369.1 M-cycles step 480: timed with pme grid 112 112 112, coulomb cutoff 1.306: 7309.0 M-cycles step 560: timed with pme grid 108 108 108, coulomb cutoff 1.355: 7521.2 M-cycles step 640: timed with pme grid 104 104 104, coulomb cutoff 1.407: 8369.8 M-cycles optimal pme grid 128 128 128, coulomb cutoff 1.200 Step Time Lambda 25005.00.0 Energies (kJ/mol) G96AngleProper Dih. Improper Dih. LJ-14 Coulomb-14 9.72545e+034.33046e+032.98087e+03 -1.95794e+028.05967e+04 LJ (SR) Coulomb (SR) Coul. recip. PotentialKinetic En. 1.01293e+06 -7.13110e+062.01689e+04 -6.00057e+061.08489e+06 Total Energy Conserved En.Temperature Pressure (bar) Constr. rmsd -4.91567e+06 -4.90300e+063.00225e+021.36173e+022.25998e-05 Step Time Lambda 5000 10.00.0 --- Thank you in advance -- Carmen Di Giovanni, PhD Dept. of Pharmaceutical and Toxicological Chemistry "Drug Discovery Lab" University of Naples "Federico II" Via D. Montesano, 49 80131 Naples Tel.: ++39 081 678623 Fax: ++39 081 678100 Email: cdigi...@unina.it Quoting Justin Lemkul : On 2/18/15 11:09 AM, Barnett, James W wrote: What's your exact command? A full .log file would be even better; it would tell us everything we need to know :) -Justin Have you reviewed this page: http://www.gromacs.org/Documentation/Acceleration_and_parallelization James "Wes" Barnett Ph.D. Candidate Chemical and Biomolecular Engineering Tulane University Boggs Center for Energy and Biotechnology, Room 341-B From: gromacs.org_gmx-users-boun...@maillist.sys.kth.se on behalf of Carmen Di Giovanni Sent: Wednesday, February 18, 2015 10:06 AM To: gromacs.org_gmx-users@maillist.sys.kth.se Subject: Re: [gmx-users] GPU low performance I post the message of a md run : Force evaluation time GPU/CPU: 40.974 ms/24.437 ms = 1.677 For optimal performance this ratio should be close to 1! NOTE: The GPU has >20% more load than the CPU. This imbalance causes performance loss, consider using a shorter cut-off and a finer PME grid. As can I solved this problem ? Thank you in advance -- Carmen Di Giovanni, PhD Dept. of Pharmaceutical and Toxicological Chemistry "Drug Discovery Lab" University of Naples "Federico II" Via D. Montesano, 49 80131 Naples Tel.: ++39 081 678623 Fax: ++39 081 678100 Email: cdigi...@unina.it Quoting Justin Lemkul : On 2/18/15 10:30 AM, Carmen Di Giovanni wrote: Daear all, I'm working on a machine with an INVIDIA Teska K20. After a minimization on a protein of 1925 atoms this is the mesage: Force evaluation time GPU/CPU: 2.923 ms/116.774 ms = 0.025 For optimal performance this ratio should be close to 1! Minimization is a poor indicator of performance. Do a real MD run. NOTE: The GPU has >25% less load than the CPU. This imbalance causes performance loss. Core t (s) Wall t (s) (%) Time: 3289.010 205.891 1597.4 (steps/hour) Performance: 8480.2 Finished mdrun on rank 0 Wed Feb 18 15:50:06 2015 Cai I improve the performance? At the moment in the forum I didn't full informations to solve this problem. In attachment there is the log. file The list does not accept attachments. If you wish to share a file, upload it to a file-sharing service and provide a URL. The full .log is quite important for understanding your hardware, optimizations, and seeing full details of the performance breakdown. But again, base your assessment on MD, not EM. -Justin -- == Justin A. Lemkul, Ph.D. Ruth L. Kirschstein NRSA Postdoctoral Fellow Department of Pharmaceutical Sciences School of Pharmacy Health Sciences Facility II, Room 629 University of Maryland, Baltimore 20 Penn St. Baltimore, MD 21201 jalem...@outerbanks.umaryland.edu | (410) 706-7441 http://mackerell.umaryland.edu/~jalemkul =
Re: [gmx-users] GPU low performance
Dear James, this is the command: gmx_mpi mdrun -s prod_30ns.tpr -deffnm prod_30ns -gpu_id 0 where gpu_id = 0 is INVIDIA Tesla K20 -- Carmen Di Giovanni, PhD Dept. of Pharmaceutical and Toxicological Chemistry "Drug Discovery Lab" University of Naples "Federico II" Via D. Montesano, 49 80131 Naples Tel.: ++39 081 678623 Fax: ++39 081 678100 Email: cdigi...@unina.it Quoting "Barnett, James W" : What's your exact command? Have you reviewed this page: http://www.gromacs.org/Documentation/Acceleration_and_parallelization James "Wes" Barnett Ph.D. Candidate Chemical and Biomolecular Engineering Tulane University Boggs Center for Energy and Biotechnology, Room 341-B From: gromacs.org_gmx-users-boun...@maillist.sys.kth.se on behalf of Carmen Di Giovanni Sent: Wednesday, February 18, 2015 10:06 AM To: gromacs.org_gmx-users@maillist.sys.kth.se Subject: Re: [gmx-users] GPU low performance I post the message of a md run : Force evaluation time GPU/CPU: 40.974 ms/24.437 ms = 1.677 For optimal performance this ratio should be close to 1! NOTE: The GPU has >20% more load than the CPU. This imbalance causes performance loss, consider using a shorter cut-off and a finer PME grid. As can I solved this problem ? Thank you in advance -- Carmen Di Giovanni, PhD Dept. of Pharmaceutical and Toxicological Chemistry "Drug Discovery Lab" University of Naples "Federico II" Via D. Montesano, 49 80131 Naples Tel.: ++39 081 678623 Fax: ++39 081 678100 Email: cdigi...@unina.it Quoting Justin Lemkul : On 2/18/15 10:30 AM, Carmen Di Giovanni wrote: Daear all, I'm working on a machine with an INVIDIA Teska K20. After a minimization on a protein of 1925 atoms this is the mesage: Force evaluation time GPU/CPU: 2.923 ms/116.774 ms = 0.025 For optimal performance this ratio should be close to 1! Minimization is a poor indicator of performance. Do a real MD run. NOTE: The GPU has >25% less load than the CPU. This imbalance causes performance loss. Core t (s) Wall t (s) (%) Time: 3289.010 205.891 1597.4 (steps/hour) Performance: 8480.2 Finished mdrun on rank 0 Wed Feb 18 15:50:06 2015 Cai I improve the performance? At the moment in the forum I didn't full informations to solve this problem. In attachment there is the log. file The list does not accept attachments. If you wish to share a file, upload it to a file-sharing service and provide a URL. The full .log is quite important for understanding your hardware, optimizations, and seeing full details of the performance breakdown. But again, base your assessment on MD, not EM. -Justin -- == Justin A. Lemkul, Ph.D. Ruth L. Kirschstein NRSA Postdoctoral Fellow Department of Pharmaceutical Sciences School of Pharmacy Health Sciences Facility II, Room 629 University of Maryland, Baltimore 20 Penn St. Baltimore, MD 21201 jalem...@outerbanks.umaryland.edu | (410) 706-7441 http://mackerell.umaryland.edu/~jalemkul == -- Gromacs Users mailing list * Please search the archive at http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before posting! * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists * For (un)subscribe requests visit https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or send a mail to gmx-users-requ...@gromacs.org. -- Gromacs Users mailing list * Please search the archive at http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before posting! * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists * For (un)subscribe requests visit https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or send a mail to gmx-users-requ...@gromacs.org. -- Gromacs Users mailing list * Please search the archive at http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before posting! * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists * For (un)subscribe requests visit https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or send a mail to gmx-users-requ...@gromacs.org. -- Gromacs Users mailing list * Please search the archive at http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before posting! * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists * For (un)subscribe requests visit https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or send a mail to gmx-users-requ...@gromacs.org.
Re: [gmx-users] GPU low performance
I post the message of a md run : Force evaluation time GPU/CPU: 40.974 ms/24.437 ms = 1.677 For optimal performance this ratio should be close to 1! NOTE: The GPU has >20% more load than the CPU. This imbalance causes performance loss, consider using a shorter cut-off and a finer PME grid. As can I solved this problem ? Thank you in advance -- Carmen Di Giovanni, PhD Dept. of Pharmaceutical and Toxicological Chemistry "Drug Discovery Lab" University of Naples "Federico II" Via D. Montesano, 49 80131 Naples Tel.: ++39 081 678623 Fax: ++39 081 678100 Email: cdigi...@unina.it Quoting Justin Lemkul : On 2/18/15 10:30 AM, Carmen Di Giovanni wrote: Daear all, I'm working on a machine with an INVIDIA Teska K20. After a minimization on a protein of 1925 atoms this is the mesage: Force evaluation time GPU/CPU: 2.923 ms/116.774 ms = 0.025 For optimal performance this ratio should be close to 1! Minimization is a poor indicator of performance. Do a real MD run. NOTE: The GPU has >25% less load than the CPU. This imbalance causes performance loss. Core t (s) Wall t (s) (%) Time: 3289.010 205.891 1597.4 (steps/hour) Performance: 8480.2 Finished mdrun on rank 0 Wed Feb 18 15:50:06 2015 Cai I improve the performance? At the moment in the forum I didn't full informations to solve this problem. In attachment there is the log. file The list does not accept attachments. If you wish to share a file, upload it to a file-sharing service and provide a URL. The full .log is quite important for understanding your hardware, optimizations, and seeing full details of the performance breakdown. But again, base your assessment on MD, not EM. -Justin -- == Justin A. Lemkul, Ph.D. Ruth L. Kirschstein NRSA Postdoctoral Fellow Department of Pharmaceutical Sciences School of Pharmacy Health Sciences Facility II, Room 629 University of Maryland, Baltimore 20 Penn St. Baltimore, MD 21201 jalem...@outerbanks.umaryland.edu | (410) 706-7441 http://mackerell.umaryland.edu/~jalemkul == -- Gromacs Users mailing list * Please search the archive at http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before posting! * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists * For (un)subscribe requests visit https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or send a mail to gmx-users-requ...@gromacs.org. -- Gromacs Users mailing list * Please search the archive at http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before posting! * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists * For (un)subscribe requests visit https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or send a mail to gmx-users-requ...@gromacs.org.
[gmx-users] GPU low performance
Daear all, I'm working on a machine with an INVIDIA Teska K20. After a minimization on a protein of 1925 atoms this is the mesage: Force evaluation time GPU/CPU: 2.923 ms/116.774 ms = 0.025 For optimal performance this ratio should be close to 1! NOTE: The GPU has >25% less load than the CPU. This imbalance causes performance loss. Core t (s) Wall t (s) (%) Time: 3289.010 205.891 1597.4 (steps/hour) Performance: 8480.2 Finished mdrun on rank 0 Wed Feb 18 15:50:06 2015 Cai I improve the performance? At the moment in the forum I didn't full informations to solve this problem. In attachment there is the log. file thank you in advance Carmen Di Giovanni -- Carmen Di Giovanni, PhD Dept. of Pharmaceutical and Toxicological Chemistry "Drug Discovery Lab" University of Naples "Federico II" Via D. Montesano, 49 80131 Naples Tel.: ++39 081 678623 Fax: ++39 081 678100 Email: cdigi...@unina.it Carmen Di Giovanni, PhD Postdoctoral Researcher Dept. of Pharmacy "Drug Discovery Lab" University of Naples "Federico II" Via D. Montesano, 49 80131 Naples Tel.: ++39 081 678623 Fax: ++39 081 678100 Email: cdigi...@unina.it -- Gromacs Users mailing list * Please search the archive at http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before posting! * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists * For (un)subscribe requests visit https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or send a mail to gmx-users-requ...@gromacs.org.
Re: [gmx-users] GPU Installation
Many thanks Mark - Original Message - From: "Mark Abraham" To: "Discussion list for GROMACS users" Sent: Monday, February 16, 2015 1:15 PM Subject: Re: [gmx-users] GPU Installation Hi, Technically yes, but usefully, no. I suspect all GPU MD implementation need several tens of thousands of particles per GPU to work efficiently, and the more GPUs the more issues there are to manage (for software and hardware). The GROMACS implementation needs a comparably tasty CPU, which you probably can't get for 8 GTX Titans. So for GROMACS, I would think more in terms of 4 Titans per dual-socket Haswell node. Mark On Mon, Feb 16, 2015 at 11:30 AM, Carmen Di Giovanni wrote: Dear all, I would like to known if Gromacs can be installed on 8 GPU GTX Titan in parallel Thank you in advance Carmen Carmen Di Giovanni, PhD Postdoctoral Researcher Dept. of Pharmacy "Drug Discovery Lab" University of Naples "Federico II" Via D. Montesano, 49 80131 Naples Tel.: ++39 081 678623 Fax: ++39 081 678100 Email: cdigi...@unina.it -- Gromacs Users mailing list * Please search the archive at http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before posting! * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists * For (un)subscribe requests visit https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or send a mail to gmx-users-requ...@gromacs.org. -- Gromacs Users mailing list * Please search the archive at http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before posting! * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists * For (un)subscribe requests visit https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or send a mail to gmx-users-requ...@gromacs.org. -- Gromacs Users mailing list * Please search the archive at http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before posting! * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists * For (un)subscribe requests visit https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or send a mail to gmx-users-requ...@gromacs.org.
[gmx-users] GPU Installation
Dear all, I would like to known if Gromacs can be installed on 8 GPU GTX Titan in parallel Thank you in advance Carmen Carmen Di Giovanni, PhD Postdoctoral Researcher Dept. of Pharmacy "Drug Discovery Lab" University of Naples "Federico II" Via D. Montesano, 49 80131 Naples Tel.: ++39 081 678623 Fax: ++39 081 678100 Email: cdigi...@unina.it -- Gromacs Users mailing list * Please search the archive at http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before posting! * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists * For (un)subscribe requests visit https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or send a mail to gmx-users-requ...@gromacs.org.