Thanks for the suggestion, Wesley. Unfortunately, it's still running slow with US too. I tried several difference CV types too with no luck.
Do you happen to know which patches you have applied?
Thanks,
Gard
On 6/23/20, 2:14 PM, "Wesley Michael Botello-Smith" <wmsmith.uci.edu> wrote:
Interesting, I have run US using both INFE and NMROPT on pmemd.cuda and
they seemed more or less comparable in speed... I haven't tried SMD with CV
module though. My version of amber may be a few patches behind though...
maybe something changed?
Do you see the same slowdown with INFE US?
On Mon, Jun 22, 2020 at 2:58 PM Gard Nelson <Gard.Nelson.immunitybio.com>
wrote:
> Hi all,
>
> I want to run some free energy calulations. I’ve used the NMRopt module
> for simple distance-based restraints in the past, but some of the
> collective variables and methods in the free energy module would be really
> useful. I set up and ran a simple distance-based SMD calculation and found
> that it slowed my simulation down ~10x (4.41 ns/day) compared to unbiased
> MD (42.00 ns/day). I set up the same calculation (except US instead of SMD)
> using nmropt instead and found the slowdown to be ~10% (37.53 ns/day) which
> is about what I’d expect. Additionally, the slowdown with nmropt is grows
> with the number of atoms I include in the COM calculation, whereas with
> SMD, the slowdown is constant whether I use 2 atoms or 700.
>
> Is this expected behavior or am I doing something wrong? I’m using
> pmemd.cuda from amber18 with all the patches applied. I’ve pasted input
> files for each simulation below. Let me know if you need any more info.
>
> Thanks for your help!
> Gard
>
> SMD input
> mdin:
> NVT
> &cntrl
> imin=0,
> irest=1,
> ntx=5,
> ntxo=2,
> iwrap=1,
> ioutfm=1,
> ntb=1,
> ntp=0,
> cut=8.0,
> nstlim=5000000,
> dt=0.002,
> ntf=2,
> ntc=2,
> temp0=300.0,
> ntpr=5000,
> ntwx=50000,
> ntt=3,
> gamma_ln=5.0,
> ig=-1,
> infe=1,
> /
> &ewald
> /
> &smd
> output_file = 'smd.dat'
> output_freq = 5000
> cv_file = 'def_CV-SMD'
> /
>
> cv_file:
> &colvar
> cv_type = 'DISTANCE'
> cv_ni = 2
> cv_i = 5119,13218
> npath = 2
> path = 0.0,100.0
> path_mode = 'LINES'
> nharm = 1
> harm = 0.0
> /
>
>
> US input
> mdin:
> NVT
> &cntrl
> imin=0,
> irest=1,
> ntx=5,
> ntxo=2,
> iwrap=1,
> ioutfm=1,
> ntb=1,
> ntp=0,
> cut=8.0,
> nstlim=50000,
> dt=0.002,
> ntf=2,
> ntc=2,
> temp0=300.0,
> ntpr=5000,
> ntwx=5000,
> ntt=3,
> gamma_ln=5.0,
> ig=-1,
> nmropt=1
> /
> &ewald
> /
> &wt
> type='DUMPFREQ'
> istep1=5000
> /
> &wt
> type='END'
> /
> DISANG=d-rst
> DUMPAVE=01_NVT.US.dat
>
> DISANG:
> US window
> &rst
> iat=-1,-1,
> igr1=5119,5259,
> igr2=13218,13371,
> r1=40,
> r2=50,
> r3=50,
> r4=60,
> rk2=0,
> rk3=0,
> fxyz=1,1,1,
> outxyz=1
> /
>
> CONFIDENTIALITY NOTICE
> This e-mail message and any attachments are only for the use of the
> intended recipient and may contain information that is privileged,
> confidential or exempt from disclosure under applicable law. If you are not
> the intended recipient, any disclosure, distribution or other use of this
> e-mail message or attachments is prohibited. If you have received this
> e-mail message in error, please delete and notify the sender immediately.
> Thank you.
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> https://urldefense.com/v3/__http://lists.ambermd.org/mailman/listinfo/amber__;!!LM3lv1w8qtQ!DEnI02Rwmlg3V1GQ-l1K_4yc7UwI42IbwxSv0fwq4XhC3woHbNBZcqWPQLRUbQz-4sZnIA$
>
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
https://urldefense.com/v3/__http://lists.ambermd.org/mailman/listinfo/amber__;!!LM3lv1w8qtQ!DEnI02Rwmlg3V1GQ-l1K_4yc7UwI42IbwxSv0fwq4XhC3woHbNBZcqWPQLRUbQz-4sZnIA$
CONFIDENTIALITY NOTICE
This e-mail message and any attachments are only for the use of the intended recipient and may contain information that is privileged, confidential or exempt from disclosure under applicable law. If you are not the intended recipient, any disclosure, distribution or other use of this e-mail message or attachments is prohibited. If you have received this e-mail message in error, please delete and notify the sender immediately. Thank you.
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Wed Jun 24 2020 - 16:00:02 PDT
