CN103150463A - Method and device for realizing molecular dynamics calculation - Google Patents
Method and device for realizing molecular dynamics calculation Download PDFInfo
- Publication number
- CN103150463A CN103150463A CN2011104035792A CN201110403579A CN103150463A CN 103150463 A CN103150463 A CN 103150463A CN 2011104035792 A CN2011104035792 A CN 2011104035792A CN 201110403579 A CN201110403579 A CN 201110403579A CN 103150463 A CN103150463 A CN 103150463A
- Authority
- CN
- China
- Prior art keywords
- gpu
- cpu
- molecular dynamics
- calculation
- calculative
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a method and device for realizing molecular dynamics calculation. The method comprises the following steps of: receiving a calculation task inputted by a user through a graphic interface; distributing data required to be calculated in the calculation task to a GPU (Graphics Processing Unit) and a CPU (Central Processing Unit); and carrying out molecular dynamics calculation on the respectively-distributed data required to be calculated according to a calculation mode of molecular dynamics by the GPU and the CPU. According to the method and the device disclosed by the invention, the convenience for task input by users is provided through a visible interface, the distributed task is commonly finished by the CPU and the GPU and the heterogeneity of a calculation system is realized; and in addition, the calculation speed can be effectively increased and a calculation scheme of the molecular dynamics is effectively improved.
Description
Technical field
The present invention relates to computer realm, and especially, relate to a kind of implementation method and device of Molecular Dynamics Calculation.
Background technology
At present, molecular dynamics (referred to as MD) computation requirement is very huge, and the high-performance computer in the TOP500 seniority among brothers and sisters much all is engaged in the relevant calculating of molecular dynamics.Molecular dynamics simulation calculates can solve the variety of issue that the basic subjects such as biology, physics, chemistry run into when theory research, can replace traditional experimental technique, makes the researchist understand more efficiently, accurately the essential information of research object.Along with improving constantly of computational complexity, people have sought much can promote the method for calculated performance, wherein just comprises the method that adopts various accelerators to accelerate.
In present molecular dynamics application system, some systems arithmetic speed is fast not, and some systems operability is relatively poor, and therefore, the present Molecular Dynamics Calculation scheme that provides is also perfect not, still, to this, not yet proposes at present effective solution.
Summary of the invention
For the perfect not problem of Molecular Dynamics Calculation scheme in correlation technique, the present invention proposes a kind of implementation method and device of Molecular Dynamics Calculation, can user friendlyly operate, and improves the speed of calculating.
Technical scheme of the present invention is achieved in that
A kind of implementation method of Molecular Dynamics Calculation is provided according to an aspect of the present invention.
The method comprises: receive the calculation task that the user inputs by graphic interface; With calculative data allocations in calculation task to graphic process unit GPU and central processor CPU; GPU and CPU carry out Molecular Dynamics Calculation according to the computing method of molecular dynamics to the calculative data that are assigned to separately.
And, calculative data allocations is comprised to GPU and CPU: according to the processing power of GPU and CPU, GPU and CPU are distributed respectively need data to be processed.
And, calculative data allocations is comprised to GPU and CPU:
Calculative data are distributed to GPU and CPU in batches.
The method can further comprise: in GPU and CPU carry out the process of Molecular Dynamics Calculation, according to the situation of change of the processing power of GPU and CPU, dynamically adjust every batch of calculative data volume that GPU and CPU are distributed.
And in the situation that distribute first calculative data to GPU and CPU, the method further comprises: the nominal processing power of obtaining GPU by application programming interface API.
In addition, the quantity of CPU is at least one; The quantity of GPU is at least one.
A kind of implement device of Molecular Dynamics Calculation is provided according to a further aspect in the invention.
This device comprises: at least one CPU; At least one GPU; The interface provides module, is used for providing graphic interface, and receives the calculation task that the user inputs by graphic interface; Distribution module is used for the calculative data allocations of calculation task to GPU and CPU; And GPU and CPU are used for the computing method according to molecular dynamics, and the calculative data that are assigned to are separately carried out Molecular Dynamics Calculation.
And distribution module is further used for the processing power according to GPU and CPU, and GPU and CPU are distributed respectively need data to be processed.
And distribution module is further used for calculative data are distributed to GPU and CPU in batches.
Further, distribution module is further used in GPU and CPU carry out the process of Molecular Dynamics Calculation, according to the situation of change of the processing power of GPU and CPU, dynamically adjusts every batch of calculative data volume that GPU and CPU are distributed.
The present invention is by providing visualization interface to facilitate user's incoming task, and the task of distributing is transferred to CPU and GPU completes jointly, realized the isomery of computing system, and can effectively improve the speed of calculating, made the numerical procedure of molecular dynamics obtain effectively perfect.
Description of drawings
Fig. 1 is the process flow diagram according to the implementation method of the Molecular Dynamics Calculation of the embodiment of the present invention;
Fig. 2 is the block diagram according to the implement device of the Molecular Dynamics Calculation of the embodiment of the present invention.
Embodiment
According to embodiments of the invention, provide a kind of implementation method of Molecular Dynamics Calculation.
As shown in Figure 1, the implementation method according to the Molecular Dynamics Calculation of the embodiment of the present invention comprises:
Step S101 receives the calculation task that the user inputs by graphic interface;
Step S103, with calculative data allocations in calculation task to graphic process unit (Graphic Processing Unit is referred to as GPU) and central processing unit (referred to as CPU);
Step S105, GPU and CPU carry out Molecular Dynamics Calculation according to the computing method of molecular dynamics to the calculative data that are assigned to separately.
Wherein, with calculative data allocations during to GPU and CPU, can be according to the processing power of GPU and CPU, distribute respectively GPU and CPU needs data to be processed.Because common calculative data volume in Molecular Dynamics Calculation is larger, therefore calculative data can be distributed to GPU and CPU in batches.
Preferably, due to processor in operational process, certain fluctuation can appear in its performance, therefore, in GPU and CPU carry out the process of Molecular Dynamics Calculation, can be according to the situation of change of the processing power of GPU and CPU, dynamically adjust every batch of calculative data volume that GPU and CPU are distributed.So just can guarantee all GPU and CPU to be distributed rational task amount in each time.
In addition, because the processing power of GPU can't can be shown as the processing power of CPU intuitively, so, in the situation that distribute first calculative data to GPU and CPU, can obtain by application programming interface API the nominal processing power of GPU, thereby understand the processing power of GPU, for the distribution of data to be calculated provides reference.
When calculating, the quantity of the CPU that adopts can be at least one, and the quantity of the GPU of employing is similarly at least one, and the CPU and the GPU quantity separately that participate in calculating can be adjusted according to actual needs, in case of necessity, also can stop whole CPU or GPU and carry out Molecular Dynamics Calculation.In addition, in the situation that the quantity of CPU is a plurality of, can carry out the dynamic adjustment of data allocations and distribute data amount according to the processing power of each CPU, also these CPU can be considered as a whole, can adopt equally similar mode to carry out data allocations and sendout adjustment for GPU.
The present invention considers that GPU is a kind of accelerator, and than CPU, the performance of GPU is calculated, and finds that the GPU calculated performance has promoted 10～100 times in the part molecular dynamics is used.But in some molecular dynamics is used, can't use GPU to accelerate, can only adopt CPU to calculate.Therefore, reasonable manner is to adopt the mode of CPU+GPU Heterogeneous Computing, allows different processors process the task of being good at separately.Therefore, such scheme can effectively be accelerated the speed calculated, raises the efficiency.
In addition, for most of researchist, cluster operation interface in the face of order line, usually need the study of a period of time could skillfully grasp use-pattern, can use high performance computing machine afterwards, and the application has visual use interface by providing, and can greatly promote user's experience, facilitates the user that computing machine is operated.
For example, in actual applications, the solution of the present invention can be supported NAMD/Gromacs/Lammps equimolecular dynamics calculation software, supports the CPU/GPU Heterogeneous Computing, supports the GPU of the CPU of Intel/AMD and Nvidia (or ATI etc.) to carry out simultaneously Heterogeneous Computing.
In addition, the present invention can provide based on the visualized graphs of B/S framework and process the interface, and the user of service only need to pass through browser, just calculation task can be submitted on the high-performance computer of isomery.
In concrete operation process, the solution of the present invention can be supported the job scheduling of molecular dynamics under isomerous environment, realize following functions: isomery device resource automatic sensing, many isomeries of support equipment calculate, support the scheduling strategies such as FIFO, resource reservation, Backfill, preemptive type, support MPI/PVM, batch processing/homework types such as interactive mode, support the customization of many queue managements, support program applying template and support abundant form book keeping operation function etc., this paper will not enumerate.The present invention can also realize the portal website of Molecular Dynamics Calculation, completes the post-processing functions such as the pretreatment function such as application data preparation and data result collection, arrangement by graphical interfaces.
According to embodiments of the invention, also provide a kind of implement device of Molecular Dynamics Calculation.
As shown in Figure 2, the implement device according to the Molecular Dynamics Calculation of the embodiment of the present invention comprises:
At least one CPU 21;
At least one GPU 22 (for purpose clearly, only show a CPU and a GPU in Fig. 2, in fact, the quantity of CPU and GPU can be for a plurality of, and this paper illustrates no longer one by one);
The interface provides module 23, is used for providing graphic interface, and receives the calculation task that the user inputs by graphic interface;
And GPU and CPU are used for the computing method according to molecular dynamics, and the calculative data that are assigned to are separately carried out Molecular Dynamics Calculation.
To sum up, by means of technique scheme of the present invention, by providing visualization interface to facilitate user's incoming task, and the task of distributing is transferred to CPU and GPU completes jointly, realized the isomery of computing system, and can effectively improve the speed of calculating, make the numerical procedure of molecular dynamics obtain effectively perfect.
These are only preferred embodiment of the present invention, in order to limit the present invention, within the spirit and principles in the present invention not all, any modification of doing, be equal to replacement, improvement etc., within all should being included in protection scope of the present invention.
Claims (10)
1. the implementation method of a Molecular Dynamics Calculation, is characterized in that, comprising:
Receive the calculation task that the user inputs by graphic interface;
With calculative data allocations in described calculation task to graphic process unit GPU and central processor CPU;
Described GPU and described CPU carry out Molecular Dynamics Calculation according to the computing method of molecular dynamics to the described calculative data that are assigned to separately.
2. implementation method according to claim 1, is characterized in that, calculative data allocations is comprised to GPU and CPU:
According to the processing power of described GPU and described CPU, described GPU and described CPU are distributed respectively need data to be processed.
3. implementation method according to claim 2, is characterized in that, calculative data allocations is comprised to GPU and CPU:
Calculative data are distributed to described GPU and described CPU in batches.
4. implementation method according to claim 3, is characterized in that, further comprises:
In described GPU and described CPU carry out the process of Molecular Dynamics Calculation, according to the situation of change of the processing power of described GPU and described CPU, dynamically adjust every batch of calculative data volume that described GPU and described CPU are distributed.
5. implementation method according to claim 4, is characterized in that, in the situation that distribute first calculative data to described GPU and described CPU, described method further comprises:
Obtain the nominal processing power of described GPU by application programming interface API.
6. the described implementation method of any one according to claim 1 to 5, is characterized in that, the quantity of described CPU is at least one; The quantity of described GPU is at least one.
7. the implement device of a Molecular Dynamics Calculation, is characterized in that, comprising:
At least one CPU;
At least one GPU;
The interface provides module, is used for providing graphic interface, and receives the user by the calculation task of described graphic interface input;
Distribution module is used for the calculative data allocations of described calculation task to described GPU and described CPU;
And described GPU and described CPU are used for the computing method according to molecular dynamics, and the described calculative data that are assigned to are separately carried out Molecular Dynamics Calculation.
8. implement device according to claim 7, is characterized in that, described distribution module is further used for the processing power according to described GPU and described CPU, and described GPU and described CPU are distributed respectively need data to be processed.
9. implement device according to claim 8, is characterized in that, described distribution module is further used for calculative data are distributed to described GPU and described CPU in batches.
10. implement device according to claim 9, it is characterized in that, described distribution module is further used in described GPU and described CPU carry out the process of Molecular Dynamics Calculation, according to the situation of change of the processing power of described GPU and described CPU, dynamically adjust every batch of calculative data volume that described GPU and described CPU are distributed.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2011104035792A CN103150463A (en) | 2011-12-06 | 2011-12-06 | Method and device for realizing molecular dynamics calculation |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2011104035792A CN103150463A (en) | 2011-12-06 | 2011-12-06 | Method and device for realizing molecular dynamics calculation |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN103150463A true CN103150463A (en) | 2013-06-12 |
Family
ID=48548539
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN2011104035792A Pending CN103150463A (en) | 2011-12-06 | 2011-12-06 | Method and device for realizing molecular dynamics calculation |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103150463A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2023123288A1 (en) * | 2021-12-30 | 2023-07-06 | 深圳晶泰科技有限公司 | Method and apparatus for determining contribution to relative binding free energy, and storage medium |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1753031A (en) * | 2005-11-10 | 2006-03-29 | 北京航空航天大学 | Particle system based on GPU |
| CN101685530A (en) * | 2008-09-23 | 2010-03-31 | 中国科学院过程工程研究所 | Method for calculating particles on GPU by utilizing multi-body interaction model |
| CN101802789A (en) * | 2007-04-11 | 2010-08-11 | 苹果公司 | Parallel runtime execution on multiple processors |
-
2011
- 2011-12-06 CN CN2011104035792A patent/CN103150463A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1753031A (en) * | 2005-11-10 | 2006-03-29 | 北京航空航天大学 | Particle system based on GPU |
| CN101802789A (en) * | 2007-04-11 | 2010-08-11 | 苹果公司 | Parallel runtime execution on multiple processors |
| CN101685530A (en) * | 2008-09-23 | 2010-03-31 | 中国科学院过程工程研究所 | Method for calculating particles on GPU by utilizing multi-body interaction model |
Non-Patent Citations (2)
| Title |
|---|
| 纪剑雄等: "一种异构计算系统动态任务分配模型", 《南通大学学报（自然科学版）》 * |
| 费辉等: "基于GPU的分子动力学模拟并行化及实现", 《计算机科学》 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2023123288A1 (en) * | 2021-12-30 | 2023-07-06 | 深圳晶泰科技有限公司 | Method and apparatus for determining contribution to relative binding free energy, and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Zhang et al. | Energy-aware virtual machine allocation for cloud with resource reservation | |
| CN103049241B (en) | A kind of method improving CPU+GPU isomery device calculated performance | |
| CN102521055B (en) | Virtual machine resource allocating method and virtual machine resource allocating system | |
| Zidenberg et al. | Multiamdahl: How should i divide my heterogenous chip? | |
| CN102591709B (en) | Shapefile master-slave type parallel writing method based on OGR (open geospatial rule) | |
| CN104572290A (en) | Method and device for controlling message processing threads | |
| CN105808347A (en) | Cluster server deployment calculation method and device | |
| CN105488134A (en) | Big data processing method and big data processing device | |
| CN106293947A (en) | GPU CPU mixing resource allocation system and method under virtualization cloud environment | |
| US20170371713A1 (en) | Intelligent resource management system | |
| CN105229608A (en) | Based on the database processing towards array of coprocessor | |
| Wu et al. | Using hybrid MPI and OpenMP programming to optimize communications in parallel loop self-scheduling schemes for multicore PC clusters | |
| CN112667368A (en) | Task data processing method and device | |
| CN103150212B (en) | The implementation method of Quantum mechanical calculation and device | |
| CN110457124A (en) | For the processing method and its device of business thread, electronic equipment and medium | |
| CN103150463A (en) | Method and device for realizing molecular dynamics calculation | |
| CN106354552A (en) | Parallel Computer Task Distribution Method and Device | |
| CN105975434B (en) | Data transfer optimization method towards heterogeneous system | |
| Liu et al. | BSPCloud: A hybrid distributed-memory and shared-memory programming model | |
| Jiang et al. | PB-FCFS-a task scheduling algorithm based on FCFS and backfilling strategy for grid computing | |
| CN103034747A (en) | Method and system for processing large-scale CST (computer simulation technology) simulation tasks | |
| CN104699520B (en) | A kind of power-economizing method based on virtual machine (vm) migration scheduling | |
| CN102968388B (en) | Data layout's method and device thereof | |
| CN102339274A (en) | Fast Fourier transform processor | |
| CN109710395A (en) | Parameter optimization control method, device and distributed computing system |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20130612 |