J. Shen and J. A. McCammon. Molecular dynamics simulation of Superoxide interacting with Superoxide Dismutase. Chemical Physics, 158:191--198, 1991.  Home/Search   Document Not in Database   Summary   Related Articles   Check

.... add of f(j loc(1:j cnt) after the executor (line 81) An optimization that is not implemented yet but at least conceptionally fairly straightforward is to use incremental schedules for pruning messages in case at least some of the data covered by a reference are already locally available [DPSM91, HKK 92] The information about what data are already available is stored for each node n 2 N in Read.GIVEN(n) 5.3.6 Reduction initialization An issue specific to reduction communications (such as Add and Mult) is the need for initializing buffer space for non local data (assigning 0 for ....

....6.14 CMFortran MPFortran version of unflattened nbf. 119 6.2.4 The input data We ran our test case for the bovine superoxide dismutase molecule (SOD ) which has N = 6968 atoms. SOD is a catalytic enzyme composed of two identical subunits, each with 151 amino acid residues and two metal atoms [SM91] Figure 6.15 shows maximal and average numbers of interaction partners, pCnt max and pCnt ave , which indicate the computational workloads for different cutoff radii. As expected, both values increase cubicly with the cutoff radius. As indicated in Equations 6:1 00 and 6:2 00 , the ....

J. Shen and J. A. McCammon. Molecular dynamics simulation of Superoxide interacting with Superoxide Dismutase. Chemical Physics, 158:191--198, 1991.

....However, using the Fortran D FORALL loop we achieve the desired semantics, see Figure 11. 7 Performance results Figure 12 gives performance results for the parallel implementation of GROMOS using IPfortran. The calculation on an Intel iPSC 860 uses a model for the enzyme Superoxide Dismutase [18], with a total of 6968 atoms, for 500 timesteps. Execution times for both the overall and principal sections of the calcula DECOMPOSITION CoordD(N) DISTRIBUTE CoordD(BLOCK) ALIGN X with CoordD much = n = P FORALL p = 0, P Gamma1 myslice = p much DO i = myslice 1, myslice much XI(i) 0.0 ....

Jian Shen and J. Andrew McCammon. Molecular dynamics simulation of superoxide interacting with superoxide dismutase. Chemical Physics, 158:191--198, 1991.

....reduces the effective parallelism. Performance of this problem is therefore not as good as others. The third one, a molecular dynamics program named GROMOS, is a real application problem [17, 16] The test data for GROMOS is the bovine superoxide dismutase molecule (SOD) which has 6968 atoms [12]. The cutoff radius is predefined to 8 A, 12 A, and 16 A. GROMOS has a more predictable structure. The number of tasks is known with the given input data, but the computation density in each task varies. The Modified Tree Walking Algorithm (MTWA) does not improve performance substantially ....

J. Shen and J. A. McCammon. Molecular dynamics simulation of superoxide interacting with superoxide dismutase. Chemical Physics, 158:191--198, 1991.

....data parallel programming language. We have tested the P kernel system using two sample programs: the N queen problem and the GROMOS Molecular Dynamics program [38, 37] GROMOS is a loosely synchronous problem. The test data is the bovine superoxide dismutase molecule (SOD) which has 6968 atoms [28]. The cutoff radius is predefined to 8 A, 12 A, and 16 A. The total execution time of a P kernel program consists of two parts, the time to execute the system program, T sys , and the time to execute the user program, T usr . The system efficiency is defined as follows: sys = T usr T = T ....

J. Shen and J. A. McCammon. Molecular dynamics simulation of superoxide interacting with superoxide dismutase. Chemical Physics, 158:191--198, 1991.

....reduces the effective parallelism. Performance of this problem is therefore not as good as others. The third one, a molecular dynamics program named GROMOS, is a real application problem [25] The test data for GROMOS is the bovine superoxide dismutase molecule (SOD) which has 6968 atoms [18]. The cutoff radius is predefined to 16 A. GROMOS has a more predictable structure. The number of uProcesses is known with the given input data, but the computation density in each uThread varies. Thus a load balancing mechanism is necessary. We use these problems to test the system behavior from ....

J. Shen and J. A. McCammon. Molecular dynamics simulation of superoxide interacting with superoxide dismutase. Chemical Physics, 158:191--198, 1991.

....structure. The number of tasks generated and the computation amount in each task are unpredictable. The second one, a molecular dynamics program named GROMOS, is a real application problem [14] The test data for GROMOS is the bovine superoxide dismutase molecule (SOD) which has 6968 atoms [11]. The cutoff radius is predefined to 8 A and 16 A. GROMOS has a more predictable structure. The number of tasks is known with the given input data, but the grain size of each task varies. We first compare four combinations of the transfer policies: ALL Eager, ALL Lazy, ANY Eager, and ANY Lazy. ....

J. Shen and J. A. McCammon. Molecular dynamics simulation of superoxide interacting with superoxide dismutase. Chemical Physics, 158:191-- 198, 1991.

....typically accounts for about 90 of the overall simulation cost. For atom i, the atoms close enough to i are precomputed into an array partners(i; 1:pCnt(i) This precomputation can be quite expensive in itself and is usually done only every k simulation steps, where k = 10 is one common value [20]. Figure 13 shows a F77 version NBFORCE for calculating the nonbonded forces between N atoms. This code can be parallelized by partitioning the set of all atoms into P disjoint subsets and assigning one subset to each processor p. To achieve load balancing, the sum over the number of the partners ....

....is active [5] This overhead is saved in the L 2 u version. 5.4 The input data We ran our test case for the bovine superoxide dismutase molecule (SOD) which has N = 6968 atoms. SOD is a catalytic enzyme composed of two identical subunits, each with 151 amino acid residues and two metal atoms [20]. Figure 18 shows maximal and average numbers of interaction partners, pCnt max and pCnt ave , which indicate the computational workloads for different cutoff radii. As expected, both values increase cubicly with the cutoff radius. As indicated in Equations 1 00 and 2 00 , the difference ....

J. Shen and J. A. McCammon. Molecular dynamics simulation of Superoxide interacting with Superoxide Dismutase. Chemical Physics, 158:191--198, 1991.

....provide a fairly good performance. However, it is not able to balance the load as well as RIPS can do. The third one, a molecular dynamics program named GROMOS, is a real application problem [37, 38] The test data for GROMOS is the bovine superoxide dismutase molecule (SOD) which has 6968 atoms [39]. The kernel of the GROMOS code is a calculation of the forces between pairs of atoms. Since the forces decrease as the distances between the pairs of atoms increase, they are approximated by considering only pairs of atoms which are closer together than a predefined cufoff radius. In our test, ....

J. Shen and J. A. McCammon, "Molecular dynamics simulation of superoxide interacting with superoxide dismutase," Chemical Physics, vol. 158, pp. 191--198, 1991.

....Problem CM 2 CM 5 11 queen 12 queen 13 queen 11 queen 12 queen 13 queen 4. 8 11.7 14.2 40.4 45.4 48.4 The second example is a loosely synchronous problem, the GROMOS Molecular Dynamics program [21, 20] The test data is the bovine superoxide dismutase molecule (SOD) which has 6968 atoms [16]. The cutoff radius is predefined to 8 A, 12 A, and 16 A. The overall efficiencies on CM 2 and CM 5 are shown in Table 8. Because of the small test data set, only 1K processors in CM 2 are used. On 1K processors in CM 2, this program achieved 102 MFLOPS for the cutoff radius of 16 A. It is ....

J. Shen and J. A. McCammon. Molecular dynamics simulation of superoxide interacting with superoxide dismutase. Chemical Physics, 158:191--198, 1991.

....reduces the effective parallelism. Therefore, performance of this problem is not as good as others. The third one, a molecular dynamics program named GROMOS, is a real application problem [29, 28] The test data for GROMOS is the bovine superoxide dismutase molecule (SOD) which has 6968 atoms [22]. The cutoff radius is predefined to 8 A, 12 A, and 16 A. GROMOS has a more predictable structure. The number of processes is known with the given input data, but the computation density in each process varies. Thus, a load balancing mechanism is necessary. In Table I, we compare RIPS to three ....

J. Shen and J. A. McCammon. Molecular dynamics simulation of superoxide interacting with superoxide dismutase. Chemical Physics, 158:191--198, 1991.

....structure. The number of tasks generated and the computation amount in each task are unpredictable. The second one, a molecular dynamics program named GROMOS, is a real application problem [23, 22] The test data for GROMOS is the bovine superoxide dismutase molecule (SOD) which has 6968 atoms [20]. The cutoff radius is predefined to 8 A, 12 A, and 16 A. GROMOS has a more predictable structure. The number of tasks is known with the given input data, but the computation amount in each task varies. Thus a load balancing mechanism is necessary. We use these problems to test the system ....

J. Shen and J. A. McCammon. Molecular dynamics simulation of superoxide interacting with superoxide dismutase. Chemical Physics, 158:191--198, 1991.

....to our parallel implementation. 5 UHGROMOS simulation results UHGROMOS has been used for performing computer simulations of the bovine superoxide dismutase molecule (SOD) SOD is a catalytic enzyme composed of two identical subunits, each with 151 amino acid residues and two metal atoms [27]. Superoxide dismutase converts the free radical O Gamma 2 to the neutral molecules O 2 and H 2 O 2 , playing an important role in removing O Gamma 2 , which is toxic, from living organisms. For the simulation discussed here, a 1 picosecond trajectory requires 500 timesteps. Upon completion ....

....removing O Gamma 2 , which is toxic, from living organisms. For the simulation discussed here, a 1 picosecond trajectory requires 500 timesteps. Upon completion of many more trajectories, an improved kinetic model for the movement of O Gamma 2 down the active site channel will be developed [27]. 5.1 Performance results Figure 9 shows simulation times for 1 picosecond calculations of SOD and O Gamma 2 in water, a total of 6968 atoms. The dominating parts of the sequential algorithm, the nonbonded force and pair list calculations, have been parallelized with nearly perfect speedup. ....

Jian Shen and J. A. McCammon. Molecular dynamics simulation of Superoxide interacting with Superoxide Dismutase. Chemical Physics, 158:191--198, 1991.

..... Number of processors Time in minutes Parallel GROMOS: 500 steps, 6968 atoms o o o o o o pairlist o o o o o o nbforce o o o o o global sum o o o o o o bonded force o o o o o o shake o o o o o load balancing o o o o o o total iPSC 860 Figure 10: Performance results. Superoxide Dismutase [22], with a total of 6968 atoms, for 500 timesteps. Both the overall execution times and the breakdowns into the principal sections of the calculation are given. The dominating parts of the sequential algorithm, the nonbonded forces and pair list, have been parallelized with nearly perfect speedup. ....

J. Shen and J. A. McCammon. Molecular dynamics simulation of Superoxide interacting with Superoxide Dismutase. Chemical Physics, 158:191--198, 1991.

....iteration reduces the effective parallelism. Performance of this problem is therefore not as good as others. The third one, a molecular dynamics program GROMOS, is a real application problem [30, 31] The test data for GROMOS is the bovine superoxide dismutase molecule (SOD) which has 6968 atoms [32]. The cutoff radius is predefined to 16 A. GROMOS has a more predictable structure. The number of processes is known with the given input data, but the computation density in each process varies. Thus a load balancing mechanism is necessary. Table II shows the exhaustive search (16 queen) ....

J. Shen and J. A. McCammon, "Molecular dynamics simulation of superoxide interacting with superoxide dismutase," Chemical Physics, vol. 158, pp. 191--198, 1991.

No context found.

J. Shen and J. A. McCammon. Molecular dynamics simulation of Superoxide interacting with Superoxide Dismutase. Chemical Physics, 158:191--198, 1991.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC