Re: Query regarding performance

From: Daipayan Sarkar (
Date: Tue Oct 25 2022 - 10:09:16 CDT

Please provide more context. What is the error message you see in the NAMD log?


From: Nicholus Bhattacharjee <>
Date: Tuesday, October 25, 2022 at 8:43 AM
To: Daipayan Sarkar <>
Cc: "" <>, René Hafner TUK <>
Subject: Re: namd-l: Query regarding performance

Hello Daipayan,

Thanks for the message. Yes now it shows ns/day. But I have another little problem. I have fixedAtoms on in my file and namd3 doesn't work with it when I try to run. I thought there might be something else in the user guide. But namd3 user guide also shows there is fixedAtoms parameter. Kindly let me know how to do this.

P.S. it works with extraBonds but not fixedAtoms


On Tue, Oct 25, 2022, 12:16 PM Daipayan Sarkar <<>> wrote:
Hi Nicholus,

Do you have “CUDASOAintegrate on” in your NAMD configuration script? Same applies for the keyword margin, check
I say this because of you mentioned in your email that namd3 log gives you “0.030 days/ns”. Usually, the benchmark lines in namd3 log would be in “ns/day”. Please check and confirm if CUDASOAintegrate is turned on. Also, it seems like your system is minimized and you can perform equilibration dynamics using minimization output. So, when using NAMD3, comment out minimize in your NAMD configuration script.

In addition, please follow the steps listed here under section getting good performance with NAMD v3 (;!!DZ3fjg!5kOeaj6HIs_H3JN6QVDSv5kn16lb8km3Cfs9SZ6oJE1k5fURGWprivEBHXKcraCNHgtDFOfisE1bllhOfh0$ ). Note, the performance here is reported when benchmarking on A100 GPUs and will be different when using V100 GPUs for a similar system size, but I expect it to get better than what you currently have.


From: Nicholus Bhattacharjee <<>>
Date: Tuesday, October 25, 2022 at 4:21 AM
To: Daipayan Sarkar <<>>
Cc: "<>" <<>>, René Hafner TUK <<>>
Subject: Re: namd-l: Query regarding performance

Hello Rene and Daipayan,

I am now using namd3. The system is little bigger 85k instead of 80k earlier. Also I have change the following parameters

Timestep 2 (it was 1 earlier)
Cutoff. 12 (it was 14)
Pairlistdist. 14 (it was 16)
fullElectFrequency 2 (it was 4)

The benchmark time in the logfile is showing around 0.030 days/ns which should have given me around 32 ns per day. This value for the earlier simulation with namd2 was around 0.080 days/ns.

However, both the simulations are giving me around 12 ns per day. Considering the benchmark time value for earlier simulation it makes sense. But I don't understand why I am getting the same for my current simulation.

Kindly let me know if I am doing something wrong.

Thank you


On Tue, Oct 4, 2022, 1:46 PM Daipayan Sarkar <<>> wrote:
Hi Nicholus,

Just to confirm are you using any collective variables based on your choice of using NAMD 2.14? If not, as Rene suggests, NAMD3 should offer better performance.


From: "<>" <<>> on behalf of René Hafner TUK <<>>
Reply-To: "<>" <<>>, René Hafner TUK <<>>
Date: Tuesday, October 4, 2022 at 5:50 AM
To: "<>" <<>>
Cc: Nicholus Bhattacharjee <<>>
Subject: Re: namd-l: Query regarding performance

Multiple timestep scheme:

timestep 2; # timestep of 2fs
nonbondedFreq 1 ; # evaluation of nonbonded interactions in every step
fullElectFrequency 2; # electrostratics only evaluated every 2nd step

The factor of 2 may be explained by the power of the V100 vs. P40 but I have no P40 at hand.

If you want/need to run for a real long time then consider HMR and/or NAMD3 if applicable, see below.


On 10/4/2022 11:07 AM, Nicholus Bhattacharjee wrote:
Hello Rene,

Thanks for the reply. It seems to be a huge difference.

I am using the following for 80k system

namd 2.14 multicore cuda
CPU 32 Xeon 6242
GPU Nvidia Tesla p40 (x1)
Timestep 1fs

What do you mean by nonbeval,fullelecteval
Is it

nonbindedFreq 1
fullElectFrequency 4

I am getting 12.5 ns/day

With increasing the timestep to 2 I can get around 25 ns/day. But still less than what you mentioned.

Please let me know.


On Tue, Oct 4, 2022, 10:45 AM René Hafner TUK <<>> wrote:

Hi Nicholus,

this depends on:

  * exact NAMD Version (2.x or 3.alphaY)
  * GPU type
  * CPU type
  * timestepping scheme
  * atom number

example one on a cluster I have access to:

  * using NAMD2.14 multicore CUDA (but selfcompiled)
  * GPU: 1xV100
  * CPU: 24xCores on XEON_SP_6126
  * timestep,nonbeval,fullelecteval: 2-1-2
  * with ~80k atoms

Results in: ~56ns/day (averaged value over multiple runs)

Though there may be space to tweak for you (cannot compare without knowing your GPU and namd version).

Depending on the system, simulation features you need and quantities you're interested in you may try out HMR ( and/or NAMD3 ( ).



On 10/3/2022 10:13 AM, Nicholus Bhattacharjee wrote:
Hello everyone,

I am running a system of around 80k atoms with namd cuda. I am using 32 CPUs with 1 GPU. I am getting a performance of around 12.5 ns/day. I would like to know if this is fine or I am getting a bad performance.

Thank you in advance.


Dipl.-Phys. René Hafner
TU Kaiserslautern
Dipl.-Phys. René Hafner
TU Kaiserslautern

This archive was generated by hypermail 2.1.6 : Tue Dec 13 2022 - 14:32:44 CST