AMBER Archive (2002)

Subject: gB vs. explicit (fwd)

From: amber_at_heimdal.compchem.ucsf.edu
Date: Tue Dec 03 2002 - 16:57:53 CST


---------- Forwarded message ----------
Date: Tue, 3 Dec 2002 14:39:51 -0600
From: Kristina Furse <kristina.e.furse_at_vanderbilt.edu>
To: amber list <amber_at_heimdal.compchem.ucsf.edu>
Subject: gB vs. explicit

Hello,

I've been running sander calculations on a large protein (cyclooxygenase dimer
with 2 heme's, 2 arachidonic acids and 8 bog detergents--1118 residues, 18,422
atoms). In the interest of saving compute cycles, I've been using generalized
Born and keeping the 200 crystallographic waters held in place with 1kcal
positional restraints. I recently set up an explicit water simulation as a
control (truncated octahedron, 11A buffer, 6 Na+, SPC/E water...):

  Solute vdw bounding box: 103.139 68.888 84.395
  Total bounding box for atom centers: 130.591 130.591 130.591
  Volume: 1139331.597 A^3 (oct)
  Total mass 646857.976 amu, Density 0.943 g/cc
  Added 28407 residues.

Scary big (29,726 residues, 104,234 atoms)--so I assumed this calculation
would take forever. But I've found that the explicit calculation is running
nearly twice as fast as the generalized Born version! Have you guys really
improved explicit calculations this much??? Could it be a crossover of timing
profiles as the number of atoms gets really large?

Below are input files and the timing section from the output for gB and
explicit calcs. Please let me know if any more information is needed to help
explain this.

Thanks!

Kristina

*NMR restraint refers to a virtual bond between the protein and the heme.

===========================================================
50ps generalized Born on 16 processors SGI Origin
===========================================================
Generalized Born model on Sander 7--equilibration
 &cntrl
   imin=0, nstlim=50000,
   ntx=5, irest=1,
   ntpr=100, ntwx=100, ntwe=100,
   ntf=1, ntc=2, ntb=0,
   tempi = 298.0, temp0 = 298.0,
   ntt=1, tautp=1.0,
   dielc=1.0, cut=12.0, scee=1.2,
   igb=1, saltcon=0.2, gbsa=1,
   ntr = 1, ntrx = 1, nmropt=1,
   nrespa=4
 &end
 &wt type='END' &end
DISANG=Heme_tetherCW.rst
CONSTRAINED RESIDUES
1.0
FIND
O * * *
SEARCH
RES 1119 1319
END
END
------------------------------------------------------------------------------

--
   5.  TIMINGS
------------------------------------------------------------------------------
--

|>>>>>>>>PROFILE of Average TIMES>>>>>>>>> | Calc gb radii 26406.29 (23.19% of Gen B) | Communicate gb radii 3279.71 ( 2.88% of Gen B) | Calc gb diag 38332.51 (33.67% of Gen B) | Calc gb off-diag 41634.81 (36.57% of Gen B) | Surface area energy 4209.52 ( 3.70% of Gen B) | Gen Born time 113862.85 (100.0% of Nonbo) | Nonbond force 113862.85 (97.35% of Force) | Bond energy 22.54 ( 0.02% of Force) | Angle energy 93.08 ( 0.08% of Force) | Dihedral energy 523.89 ( 0.45% of Force) | FRC Collect time 1502.79 ( 1.28% of Force) | Other 960.63 ( 0.82% of Force) | Force time 116966.01 (99.07% of Runmd) | Shake time 252.40 ( 0.21% of Runmd) | Verlet update time 102.87 ( 0.09% of Runmd) | CRD distribute time 523.87 ( 0.44% of Runmd) | Other 215.06 ( 0.18% of Runmd) | Runmd Time 118060.20 (100.0% of Total) | Total time 117777.66 (100.0% of ALL )

| Highest rstack allocated: 209275 | Highest istack allocated: 76100

| Setup wallclock 4 seconds | Nonsetup wallclock 117775 seconds

=========================================================== 50ps truncatedOct on 16 processors SGI Origin =========================================================== Equilibration of cox dimer, explicit water, 9.0 cut &cntrl

nmropt = 1, ntx = 5, irest = 1, ntrx = 1, ntxo = 1, ntpr = 100, ntwx = 100, ntwv = 0, ntwe = 100,

ntf = 2, ntb = 2, cut = 9.0,

ibelly = 0, ntr = 0,

imin = 0, nstlim = 50000, nscm = 0, t = 0.0, dt = 0.001,

temp0 = 298.0, tempi = 298.0, ig = 71277, heat = 0.0, ntt = 1, tautp = 0.2, vlimit = 20.0,

ntp = 1, pres0 = 1.0, comp = 44.6, taup = 0.2,

ntc = 2, tol = 0.00001, watnam = 'SPC ',

&end &ewald

nbflag = 1, skinnb = 2.0, use_pme = 1,

&end &wt type='END' &end DISANG=Heme_tetherEXP.rst

------------------------------------------------------------------------------ -- 5. TIMINGS ------------------------------------------------------------------------------ --

|>>>>>>>>PROFILE of Average TIMES>>>>>>>>> | Ewald setup time 0.24 ( 0.01% of List ) | Check list validity 73.41 ( 2.22% of List ) | Map frac coords 106.55 ( 3.22% of List ) | Grid unit cell 33.30 ( 1.01% of List ) | Grid image cell 16.39 ( 0.50% of List ) | Build the list 2985.96 (90.33% of List ) | Other 89.79 ( 2.72% of List ) | List time 3305.74 ( 7.58% of Nonbo) | Direct Ewald time 22365.02 (55.52% of Ewald) | Adjust Ewald time 297.84 ( 0.74% of Ewald) | Finish NB virial 1377.08 ( 3.42% of Ewald) | Fill Bspline coeffs 1471.12 (10.31% of Recip) | Fill charge grid 1949.59 (13.66% of Recip) | Scalar sum 2145.70 (15.03% of Recip) | Grad sum 2664.36 (18.66% of Recip) | FFT communication ti 2690.10 (46.17% of FFT t) | Other 3136.66 (53.83% of FFT t) | FFT time 5826.76 (40.82% of Recip) | Other 218.14 ( 1.53% of Recip) | Recip Ewald time 14275.68 (35.44% of Ewald) | Ewald MPI wait 39.78 ( 0.10% of Ewald) | Other 1925.95 ( 4.78% of Ewald) | Ewald time 40282.25 (92.41% of Nonbo) | Nonbond force 43588.75 (85.93% of Force) | Bond energy 20.79 ( 0.04% of Force) | Angle energy 95.94 ( 0.19% of Force) | Dihedral energy 413.51 ( 0.82% of Force) | FRC Collect time 4650.71 ( 9.17% of Force) | Other 1957.02 ( 3.86% of Force) | Force time 50726.97 (78.74% of Runmd) | Shake time 629.52 ( 0.98% of Runmd) | Verlet update time 1993.48 ( 3.09% of Runmd) | Ekcmr time 2910.31 ( 4.52% of Runmd) | CRD distribute time 7198.36 (11.17% of Runmd) | Other 962.48 ( 1.49% of Runmd) | Runmd Time 64421.12 (100.0% of Total) | Other 13.88 ( 0.02% of Total) | Total time 64435.09 (100.0% of ALL )

| Highest rstack allocated: 625408 | Highest istack allocated: 3935320

| Setup wallclock 14 seconds | Nonsetup wallclock 64424 seconds

=========================================================== For comparison, same input file, but LINUX: 50ps truncatedOct on 16 processors LINUX cluster ===========================================================

------------------------------------------------------------------------------ -- 5. TIMINGS ------------------------------------------------------------------------------ --

|>>>>>>>>PROFILE of Average TIMES>>>>>>>>> | Check list validity 1797.84 (46.11% of List ) | Map frac coords 201.29 ( 5.16% of List ) | Grid unit cell 72.72 ( 1.86% of List ) | Grid image cell 23.69 ( 0.61% of List ) | Build the list 1736.75 (44.54% of List ) | Other 66.86 ( 1.71% of List ) | List time 3899.40 ( 7.02% of Nonbo) | Direct Ewald time 30638.63 (59.31% of Ewald) | Adjust Ewald time 248.71 ( 0.48% of Ewald) | Finish NB virial 1850.59 ( 3.58% of Ewald) | Fill Bspline coeffs 1580.40 ( 9.68% of Recip) | Fill charge grid 2333.22 (14.30% of Recip) | Scalar sum 1821.88 (11.16% of Recip) | Grad sum 3129.43 (19.17% of Recip) | FFT communication ti 2800.28 (38.89% of FFT t) | Other 4400.58 (61.11% of FFT t) | FFT time 7200.85 (44.12% of Recip) | Other 255.29 ( 1.56% of Recip) | Recip Ewald time 16321.08 (31.59% of Ewald) | Ewald MPI wait 43.95 ( 0.09% of Ewald) | Other 2558.72 ( 4.95% of Ewald) | Ewald time 51662.14 (92.98% of Nonbo) | Nonbond force 55561.87 (93.49% of Force) | Bond energy 13.93 ( 0.02% of Force) | Angle energy 83.54 ( 0.14% of Force) | Dihedral energy 274.20 ( 0.46% of Force) | FRC Collect time 2605.01 ( 4.38% of Force) | Other 891.54 ( 1.50% of Force) | Force time 59430.15 (86.25% of Runmd) | Shake time 940.36 ( 1.36% of Runmd) | Verlet update time 2082.98 ( 3.02% of Runmd) | Ekcmr time 1617.09 ( 2.35% of Runmd) | CRD distribute time 3733.09 ( 5.42% of Runmd) | Other 1099.76 ( 1.60% of Runmd) | Runmd Time 68903.43 (100.0% of Total) | Other 23.63 ( 0.03% of Total) | Total time 68927.14 (100.0% of ALL )

| Highest rstack allocated: 599503 | Highest istack allocated: 3935320

| Setup wallclock 25 seconds | Nonsetup wallclock 68905 seconds

**************************************************** Kristina E. Furse Department of Chemistry Center for Structural Biology Vanderbilt University email: kfurse_at_structbio.vanderbilt.edu