CN103345580B - Based on the parallel CFD method of lattice Boltzmann method - Google Patents
Based on the parallel CFD method of lattice Boltzmann method Download PDFInfo
- Publication number
- CN103345580B CN103345580B CN201310274574.3A CN201310274574A CN103345580B CN 103345580 B CN103345580 B CN 103345580B CN 201310274574 A CN201310274574 A CN 201310274574A CN 103345580 B CN103345580 B CN 103345580B
- Authority
- CN
- China
- Prior art keywords
- parallel
- computing node
- flow field
- max
- represent
- 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.)
- Expired - Fee Related
Links
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a kind of parallel CFD method based on lattice Boltzmann method.The method can carry out hydromechanical calculating and analysis concurrently for the object of various complexity, its feature is to introduce the parallel distributed grid cutting algorithm being suitable for lattice Boltzmann method, the mesh generation time is shortened greatly, the distributed computer architecture of basis simultaneously, construct and reduce the traffic algorithm of lattice Boltzmann method in parallel iteration calculates, accelerate the speed of iteration, improve the extensibility of algorithm, make lattice Boltzmann method be applicable to large-scale calculations.A large amount of numerical experiments demonstrates this parallel calculating method and is with good expansibility, and is more suitable for ultra-large computing system.
Description
Technical field
The present invention relates to Fluid Mechanics Computation and computer realm, propose a kind of parallel CFD method based on lattice Boltzmann method.
Background technology
Through the development of decades, computation fluid dynamics is in Aero-Space, and large-scale energy source device (as nuclear power station), new traffic tool, numerous engineering department such as environmental protection and field are obtained for applies widely.At present, Fluid Mechanics Computation oneself become modern computing science and the most effectively one of promote.Make us advising object to be in progress although CFD has achieved, due to the complicacy of fluid motion and the restriction of computer resource, still there is many limitation in CFD.The calculating of some problem, Forecast and control are not sure yet.Such as, owing to lacking the restriction of effective computation model and computing machine condition, sometimes may reach 100% to the error of the calculating of spacecraft heat problem, much disastrous weather can't compare accurate forecast.The flow field problem that unit solves these complexity often needs long computing time, and some extensive problem even cannot solve, and this limits the application of CFD in scientific research and engineering problem to a certain extent.As can be seen here, efficient computation model and advanced Computer Architecture must be adopted to solve complicated CFD problem, so the needing of the development of engineering and problem in science has researched and proposed more urgent requirement to CFD, and fluid calculation becomes a challenging field.
The effective utilization of massively parallel computer needs new parallel algorithm.Even if because utilize classic method to solve some engineering and problem in science on supercomputer, as the direct Numerical etc. of turbulent flow, still need the time grown very much.Traditional method solving Navier-Stokes equation needs to solve large-scale Algebraic Equation set, and Navier-Stokes equation explicit scheme concurrency is better, but numerical stability is poor, and speed of convergence is slow; Implicit expression solution better numerical value stability, fast convergence rate, but its parallel scalability is poor.
Summary of the invention
The object of the invention is to, in order to solve the problem, a kind of parallel CFD method based on lattice Boltzmann method is provided, be to provide a kind of novel full parellel highly effective algorithm that time, the lattice Boltzmann method of space approximate shceme and parallel computation adapt, the method, based on Molecule Motion Theory, has physical background clearly.The method is being macroscopically discrete method, microcosmic is continuation method, is thus called as Mesoscopic simulation method.The advantages such as the method has the advantage of many uniquenesses, as high in counting yield, boundary condition treatment is simple, program is easy to implement.The method is a kind of discrete statistical model, in computation process, information transmission is local, therefore the concurrent operation program degree of association is less, need the quantity of information of transmission less between each computing node, be particularly suitable for carrying out parallel computation on the large-scale computer with distributed architecture, fairly large numerical simulation can be carried out when expense is smaller, particularly the arrival in cloud epoch, resource obtains and is more prone to, and more massive parallel computation more easily launches.
For achieving the above object, design of the present invention is: first according to the data of surface grids and the parameter request of actual computation, and exploitation carries out cartesian grid subdivision method in a distributed manner, accelerates mesh generation; Simultaneously in conjunction with certain data structure, the initialization of acceleration grid, then the parallel trellis Boltzmann method having an enhanced scalability of the less traffic according to lattice Boltzmann method exploitation carries out iterative computation, until export flow field and object plane information after meeting the accuracy requirement calculated, carry out visual analyzing afterwards.
According to foregoing invention design, for the needs of actual computation problem, Flow Field Calculation scope, object plane information, the present invention adopts following technical proposals:
A, computation requirement according to practical problems, read in the Parameter File of calculating;
B, path according to the surface grids stl file in Parameter File, reading face grid file;
C, data according to the flow field scope in Parameter File and surface grids, the distributed cartesian grid subdivision method of exploitation, the concurrently distributed subdivision carrying out cartesian grid;
D, according to the calculation requirement in Parameter File, the cartesian grid after subdivision is carried out initialization distributed parallel;
The parallel trellis Boltzmann method of the enhanced scalability of E, exploitation carries out parallel iteration calculating, until meet the accuracy requirement in Parameter File;
Information of flow in F, output result of calculation and object plane information;
The visual analyzing of G, result of calculation.
Parallel CFD method based on lattice Boltzmann method of the present invention, compared with prior art, has following outstanding substantive distinguishing features and remarkable advantage:
1. the method establishes one distributed grid subdivision method fast, accelerates the speed of mesh generation.Because grid distributedly carries out subdivision, be do not need to carry out communication between each node carrying out subdivision, be complete parallel pattern, have the speed-up ratio of near-linear.
2. due to after mesh generation be distributed storage on each computing node, so grid computing information initializing can be carried out in a distributed manner, do not need to communicate, accelerate initialization time, also there is the speed-up ratio of near-linear simultaneously.
3. the method is simultaneously according to lattice Boltzmann method, give a kind of parallel trellis Boltzmann method with Highly Scalable, carried out iterative computation post, shortened the former methodical parallel computation time, the method is also with good expansibility simultaneously.
Accompanying drawing explanation
Fig. 1 is the process flow diagram of the parallel CFD method based on lattice Boltzmann method of the present invention.
Fig. 2 is the particular flow sheet of distributed variable-frequencypump cartesian grid subdivision in step C in Fig. 1.
Fig. 3 is that the information of flow in Fig. 1 described in step D carries out initialized process flow diagram.
Fig. 4 is step e and the process flow diagram described in F in Fig. 1.
Fig. 5 is the model velocity structural drawing of D3Q19 in lattice Boltzmann method.
Fig. 6 is the communication concurrent type frog of two-dimensional lattice Boltzmann method.
Fig. 7 is the communication sequence formula of two-dimensional lattice Boltzmann method.
Fig. 8 is the test result figure of scale and efficiency.
Fig. 9 is the speed-up ratio figure of three kinds of scale test problems.
Figure 10 is the efficiency diagram of three kinds of scale test problems.
Embodiment
Below in conjunction with accompanying drawing, the preferred embodiments of the present invention are described in detail.
Embodiment one:
See Fig. 1, this is based on the parallel CFD method of lattice Boltzmann method, be to provide a kind of novel full parellel highly effective algorithm that time, the lattice Boltzmann method of space approximate shceme and parallel computation adapt, it is characterized in that: for given real fluid computational problem, the method can be adopted efficiently to simulate, and concrete operation step is as follows:
A, needs according to actual computation problem, given calculating parameter also saves as ini format parameter file, reads this Parameter File; Concrete steps comprise:
A1, host process read in Parameter File;
The Parameter File content of reading in is sent in all the other each processes by A2, host process;
B, according to the surface grids stl file storage location provided in steps A, read in this file; Concrete steps comprise:
B1, host process read in surface grids stl file;
The surface grids data of reading in are sent in all the other each processes by B2, host process;
C, according to the content in steps A and step C, distributed parallel carries out mesh generation;
D, according to the calculation requirement in Parameter File, the cartesian grid that each computing node is responsible for carries out initialization;
The parallel trellis Boltzmann method of the enhanced scalability of E, exploitation carries out parallel iteration calculating, until meet the accuracy requirement in Parameter File;
The result that F, output calculate; Concrete steps are as follows:
F1, master computing node export object plane result and flow field result, and the output format of result is Tecplot/VTK file layout;
Object plane and Flow Field Calculation result are input to after the result of host node output by F2, all the other computing nodes in order, form complete object plane and Flow Field Calculation result;
G, object plane and Flow Field Calculation result visual: mainly adopt Tecplot business software or ParaView open source software.
Embodiment two:
In the present embodiment, this experiment based on the parallel CFD method of lattice Boltzmann method is at the magic square 5000A of Shanghai supercomputer center, magic square 5000A is the large computer system based on cluster concept design, and its overall calculation capability theory peak value is 200Tflops (1Tflops is 1012 Floating-point Computation per second).
For the needs of actual computation problem, Flow Field Calculation scope, object plane information, the parallel CFD method based on lattice Boltzmann method of the present invention, as shown in Fig. 1-Fig. 7, comprises the following steps:
H, needs according to actual computation problem, given calculating parameter also saves as ini format parameter file, reads this Parameter File.Concrete steps comprise:
A1, host process read in Parameter File;
The Parameter File content of reading in is sent in all the other each processes by A2, host process.
I, according to the surface grids stl file storage location provided in steps A, read in this file.Concrete steps comprise:
B1, host process read in surface grids stl file;
The surface grids data of reading in are sent in all the other each processes by B2, host process.
J, according to the content in steps A and step C, distributed parallel carries out mesh generation.Concrete steps comprise:
C1, information according to the calculating parameter in Parameter File and surface grids, by each computing node the mesh generation region of being responsible for divide.Concrete steps comprise:
C11, according to computing node
nwith flow field regions size, by flow field regions (
x,
y,
zthree directions) be evenly divided into
nequal portions, each computing node portion wherein, the scope that namely each process is responsible for is:
x∈[
min_x i ,
max_x i ]、
y∈[
min_y i ,
max_y i ]、
z∈[
min_z i ,
max_z i ];
In formula
min_x i represent flow field regions
xthe minimum value in direction,
max_x i represent flow field regions
xthe maximal value in direction,
min_y i represent flow field regions
ythe minimum value in direction,
max_y i represent flow field regions
ythe maximal value in direction,
min_z i represent flow field regions
zthe minimum value in direction,
max_z i represent flow field regions
zthe maximal value in direction;
C12, according to the size of mesh opening in the Parameter File read in steps A
size, the portion be responsible in each computing node respectively to
x,
y,
zthree directions all increase and reduce one
size, the scope that namely each process is responsible for is:
x∈[
min_x i -
size,
max_x i +
size]、
y∈[
min_y i -
size,
max_y i +
size]、
z∈[
min_z i -
size,
max_z i +
size];
C2, each computing node carry out mesh generation to the scope be responsible for separately, judge the type of each net point simultaneously.Concrete steps are:
C21, generate initial cartesian grid;
C22, judge the situation that each net point is crossing with object plane;
In C3, each computing node, the net point of solid interior (namely object plane surrounds and formed) is deleted.
K, according to the calculation requirement in Parameter File, the cartesian grid that each computing node is responsible for carries out initialization.Concrete steps comprise:
D1, according to the calculating parameter in Parameter File, different trellis-types is carried out to the assignment of flow field macroscopic information;
D2, calculate microscopic information amount according to the equilibrium distribution function of the macroscopic information of assignment and the D3Q19 rate pattern of lattice Boltzmann method, specific formula for calculation is shown in as follows:
In formula
represent
the equilibrium distribution function value of individual discrete velocity reversal,
represent density,
representation speed,
represent
individual discrete speed (see figure 5) is:
Weight function
for:
The velocity of sound:
, wherein
cfor grid speed.
The parallel trellis Boltzmann method of the enhanced scalability of L, exploitation carries out parallel iteration calculating, until meet the accuracy requirement in Parameter File.Concrete steps are as follows:
E1, iterations variable Iteration, if reach the iterations in Parameter File, then exit iteration, otherwise proceed iteration;
The error of E2, calculating macroscopic quantity, judges according to the accuracy requirement in Parameter File, if met, exits, otherwise continue;
E21, according to microscopic quantity and following formula, the new macroscopic quantity on computing grid point:
In formula
represent
the distribution function value of individual discrete velocity reversal;
E22, original macroscopic quantity on the new macroscopic quantity calculated and net point being taken absolute value after subtracting each other obtains the error of macroscopic quantity;
Whether E23, error in judgement meet accuracy requirement, if met, exit, otherwise continue;
The Parallel Collision transition process of E3, lattice Boltzmann method, main computing formula is shown in as follows:
In formula
represent relaxation factor,
represent the time,
represent the time interval, this formula is divided into two parts: migration and collision, and concrete steps are as follows:
E31, mobile grid (transition process), in the transition process of the net point be responsible at each computing node, all need the gridding information using neighborhood calculation node, this just needs to carry out the traffic operation between computing node, there is the following two kinds operation steps:
E311, according to the communication concurrent type frog under two-dimensional case in Fig. 6, under three-dimensional situation can being generalized to easily.Each computing node first communicates according to similar burse mode
E312 or according to the communication sequence formula under two-dimensional case in Fig. 7, under also can being generalized to three-dimensional situation easily.Each computing node first communicates according to similar ordered mode
E32, communicated after, each computing node can carry out impact operations, and this step does not need to communicate, and is complete parallel, but collision is only meeting
x∈ [
min_x i ,
max_x i ],
y∈ [
min_y i ,
max_y i ],
z∈ [
min_z i ,
max_z i ] net point in scope carries out
E33, iteration variable Iteration=Iteration+1, continue to perform E1 step.
The result that M, output calculate.Concrete steps are as follows:
F1, master computing node export object plane result and flow field result, and the output format of result is Tecplot/VTK file layout;
Object plane and Flow Field Calculation result are input to after the result of host node output by F2, all the other computing nodes in order, form complete object plane and Flow Field Calculation result;
N, object plane and Flow Field Calculation result visual.Main employing Tecplot business software or ParaView open source software.
Fig. 8-10 below surpasses in Shanghai the calculating test result calculated on dawn 5000A mainly for the parallel CFD method based on lattice Boltzmann method of the present invention and is described.
With reference to Fig. 8, give the parallel CFD method based on lattice Boltzmann method of the present invention and calculating the efficiency diagram under scale problem for difference.Describe the process that the inventive method can be applicable to extensive problem, and can good efficiency be kept.
Fig. 9 gives the comparison diagram for the parallel speedup ratio under three kinds of different scales and ideal value, and as can be seen from the figure, method of the present invention is with good expansibility.
Figure 10 gives and the parallel efficiency figure under same three kinds of scales in Fig. 9, and as can be seen from the figure, method parallel efficiency of the present invention still maintains the efficiency of more than 80% in 4000 core magnitudes.
In sum, Fig. 8 ~ Figure 10 shows, parallel CFD method based on lattice Boltzmann method of the present invention, can calculate for extensive problem, and the extensibility of the method is good, efficiency can maintain higher level, and the method can play the effectiveness of ultra-large supercomputer well simultaneously, exchanges computing time for computational resource.This parallel CFD method is effectively supplementing of traditional C FD method, and implements more for convenience.
Carry out elaboration in conjunction with Figure of description and specific embodiment herein and just understand method of the present invention and core concept for helping.Method of the present invention is not limited to the embodiment described in embodiment, other embodiment that those skilled in the art draw according to method of the present invention and thought, belongs to technological innovation scope of the present invention equally.This description should not be construed as limitation of the present invention.
Claims (4)
1., based on the parallel CFD method of lattice Boltzmann method, be to provide a kind of full parellel highly effective algorithm that time, the lattice Boltzmann method of space approximate shceme and parallel computation adapt, implementation step is as follows:
A, needs according to actual computation problem, given calculating parameter also saves as ini format parameter file, reads this Parameter File; Concrete steps comprise:
A1, host process read in Parameter File;
The Parameter File content of reading in is sent in all the other each processes by A2, host process;
B, according to the surface grids stl file storage location provided in steps A, read in this file; Concrete steps comprise:
B1, host process read in surface grids stl file;
The surface grids data of reading in are sent in all the other each processes by B2, host process;
C, according to the content in steps A and step B, distributed parallel carries out mesh generation;
D, according to the calculation requirement in Parameter File, the cartesian grid that each computing node is responsible for carries out initialization;
The parallel trellis Boltzmann method of the enhanced scalability of E, exploitation carries out parallel iteration calculating, until meet the accuracy requirement in Parameter File;
The result that F, output calculate; Concrete steps are as follows:
F1, master computing node export object plane result and flow field result, and the output format of result is Tecplot/VTK file layout;
Object plane and Flow Field Calculation result are input to after the result of host node output by F2, all the other computing nodes in order, form complete object plane and Flow Field Calculation result;
G, object plane and Flow Field Calculation result visual: mainly adopt Tecplot business software or ParaView open source software;
Distributed must the walking abreast of described step C carries out mesh generation, and concrete steps are as follows:
C1, information according to the calculating parameter in Parameter File and surface grids, by each computing node the mesh generation region of being responsible for divide; Concrete steps comprise:
C11, according to computing node
nwith flow field regions size, by flow field regions
x,
y,
zthree directions are evenly divided into
nequal portions, each computing node portion wherein, the scope that namely each process is responsible for is:
x∈[
min_x i ,
max_x i ]、
y∈[
min_y i ,
max_y i ]、
z∈[
min_z i ,
max_z i ];
In formula
min_x i represent flow field regions
xthe minimum value in direction,
max_x i represent flow field regions
xthe maximal value in direction,
min_y i represent flow field regions
ythe minimum value in direction,
max_y i represent flow field regions
ythe maximal value in direction,
min_z i represent flow field regions
zthe minimum value in direction,
max_z i represent flow field regions
zthe maximal value in direction;
C12, according to the size of mesh opening in the Parameter File read in steps A
size, the portion be responsible in each computing node increases and reduces one respectively to x, y, z three directions
size, the scope that namely each process is responsible for is:
x∈[
min_x i -
size,
max_x i +
size]、
y∈[
min_y i -
size,
max_y i +
size]、
z∈[
min_z i -
size,
max_z i +
size];
C2, each computing node carry out mesh generation to the scope be responsible for separately, judge the type of each net point simultaneously; Concrete steps are:
C21, generate initial cartesian grid;
C22, judge the situation that each net point is crossing with object plane;
Surround to solid interior and object plane the net point formed in C3, each computing node to delete.
2. the parallel CFD method based on lattice Boltzmann method according to claim 1, it is characterized in that: the cartesian grid in described step D carries out initialization, its concrete grammar is:
D1, according to the calculating parameter in Parameter File, different trellis-types is carried out to the assignment of flow field macroscopic information;
D2, calculate microscopic information amount according to the equilibrium distribution function of the macroscopic information of assignment and the D3Q19 rate pattern of lattice Boltzmann method, specific formula for calculation is shown in as follows:
In formula
represent
the equilibrium distribution function value of individual discrete velocity reversal,
represent density,
representation speed,
represent
individual discrete speed is:
Weight function
for:
The velocity of sound:
, wherein
cfor grid speed.
3. the parallel CFD method based on lattice Boltzmann method according to claim 1, it is characterized in that: in described step e, utilize lattice Boltzmann method to carry out parallel iteration calculating, distributed parallel carries out the numerical evaluation of fluid, and concrete steps are as follows:
E1, iterations variable Iteration, if reach the iterations in Parameter File, then exit iteration, otherwise proceed iteration;
The error of E2, calculating macroscopic quantity, judges according to the accuracy requirement in Parameter File, if met, exits, otherwise continue;
E21, according to microscopic quantity and following formula, the new macroscopic quantity on computing grid point:
In formula
represent
the distribution function value of individual discrete velocity reversal;
E22, original macroscopic quantity on the new macroscopic quantity calculated and net point being taken absolute value after subtracting each other obtains the error of macroscopic quantity;
Whether E23, error in judgement meet accuracy requirement, if met, exit, otherwise continue;
The Parallel Collision transition process of E3, lattice Boltzmann method, main computing formula is shown in as follows:
In formula
represent relaxation factor,
represent the time,
represent the time interval, this formula is divided into two parts: migration and collision, and concrete steps are as follows:
E31, mobile grid, transition process, in the transition process of the net point be responsible at each computing node, all need the gridding information using neighborhood calculation node, this just needs to carry out the traffic operation between computing node, can adopt sequential and concurrent type frog communication pattern
E32, communicated after, each computing node can carry out impact operations, and this step does not need to communicate, and is complete parallel, but collision is only meeting
x∈ [
min_x i ,
max_x i ],
y∈ [
min_y i ,
max_y i ],
z∈ [
min_z i ,
max_z i ] net point in scope carries out
E33, iteration variable Iteration=Iteration+1, continue to perform E1 step.
4. the parallel CFD method based on lattice Boltzmann method according to claim 3, is characterized in that: in described step e 31, sequential and concurrent type frog communication pattern, and concrete grammar is as follows:
E311, communication concurrent type frog, the second layer grid in claim 1 is sent to the computing node needed separately by each computing node simultaneously, and this communication mode needs cushion space, so be applicable to the cluster of middle and small scale;
E312, communication sequence formula, each computing node is in a certain order, point-to-point, passes data to corresponding computing node, and this communication mode does not need cushion space, and applicability is good, is applicable to large-scale cluster.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201310274574.3A CN103345580B (en) | 2013-07-02 | 2013-07-02 | Based on the parallel CFD method of lattice Boltzmann method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201310274574.3A CN103345580B (en) | 2013-07-02 | 2013-07-02 | Based on the parallel CFD method of lattice Boltzmann method |
Publications (2)
Publication Number | Publication Date |
---|---|
CN103345580A CN103345580A (en) | 2013-10-09 |
CN103345580B true CN103345580B (en) | 2016-04-27 |
Family
ID=49280375
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201310274574.3A Expired - Fee Related CN103345580B (en) | 2013-07-02 | 2013-07-02 | Based on the parallel CFD method of lattice Boltzmann method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN103345580B (en) |
Families Citing this family (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104573151A (en) * | 2013-10-22 | 2015-04-29 | 北京临近空间飞行器***工程研究所 | Cyclic communication method for acquiring wall grid point data |
CN104573296B (en) * | 2013-10-22 | 2017-09-22 | 北京临近空间飞行器***工程研究所 | A kind of Hypersonic Flow Field initial method towards similar grid |
CN104462829B (en) * | 2014-12-11 | 2019-01-18 | 中国航天空气动力技术研究院 | The processing method of complicated multi-region grid in space propultion solution |
CN104933225B (en) * | 2015-05-25 | 2018-01-09 | 中国科学院过程工程研究所 | The method for realizing the extensive real-time Simulation of Fluid Mechanics Computation |
CN105760602A (en) * | 2015-12-30 | 2016-07-13 | 南京航空航天大学 | Total flow field numerical simulation method for finite volume weighted essentially non-oscillatory scheme |
CN107515987A (en) * | 2017-08-25 | 2017-12-26 | 中国地质大学(北京) | The simulation accelerated method of Groundwater Flow based on more relaxation Lattice Boltzmann models |
CN111105341B (en) * | 2019-12-16 | 2022-04-19 | 上海大学 | Framework method for solving computational fluid dynamics with low power consumption and high operational performance |
CN111489447B (en) * | 2020-04-14 | 2022-04-29 | 西北工业大学 | Right-angle grid adaptive modeling method suitable for lattice Boltzmann method |
CN111538487B (en) * | 2020-04-17 | 2022-06-14 | 中国空气动力研究与发展中心计算空气动力研究所 | Distributed parallel grid generation software framework |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102681972A (en) * | 2012-04-28 | 2012-09-19 | 浪潮电子信息产业股份有限公司 | Method for accelerating lattice-Boltzmann by utilizing graphic processing units (GPUs) |
CN102945295A (en) * | 2012-10-15 | 2013-02-27 | 浪潮(北京)电子信息产业有限公司 | Parallel acceleration method and system of lattice Boltzmann method |
CN103064819A (en) * | 2012-10-25 | 2013-04-24 | 浪潮电子信息产业股份有限公司 | Method for utilizing microwave integrated circuit (MIC) to rapidly achieve lattice Boltzmann parallel acceleration |
-
2013
- 2013-07-02 CN CN201310274574.3A patent/CN103345580B/en not_active Expired - Fee Related
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102681972A (en) * | 2012-04-28 | 2012-09-19 | 浪潮电子信息产业股份有限公司 | Method for accelerating lattice-Boltzmann by utilizing graphic processing units (GPUs) |
CN102945295A (en) * | 2012-10-15 | 2013-02-27 | 浪潮(北京)电子信息产业有限公司 | Parallel acceleration method and system of lattice Boltzmann method |
CN103064819A (en) * | 2012-10-25 | 2013-04-24 | 浪潮电子信息产业股份有限公司 | Method for utilizing microwave integrated circuit (MIC) to rapidly achieve lattice Boltzmann parallel acceleration |
Non-Patent Citations (1)
Title |
---|
《球面空间网格的分布式并行技术研究与实现》;宋能;《中国优秀硕士学位论文全文数据库-信息科技辑》;20110315;第2011年卷(第3期);第32页第2段,第35页最后一段,图3.5 * |
Also Published As
Publication number | Publication date |
---|---|
CN103345580A (en) | 2013-10-09 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN103345580B (en) | Based on the parallel CFD method of lattice Boltzmann method | |
CN103970960B (en) | The element-free Galerkin structural topological optimization method accelerated parallel based on GPU | |
Kampolis et al. | CFD-based analysis and two-level aerodynamic optimization on graphics processing units | |
Groth et al. | Parallel implicit adaptive mesh refinement scheme for body-fitted multi-block mesh | |
CN103617150A (en) | GPU (graphic processing unit) based parallel power flow calculation system and method for large-scale power system | |
CN103164190B (en) | A kind of fast parallelization method of full distributed river basin ecological hydrology model | |
CN103106301B (en) | A kind of method of the calculating radiation shield be coupled with characteristic line method based on Monte Carlo method | |
CN105302974A (en) | Real-time cutting simulation method of flexible object on the basis of finite element and time-variant modal analysis | |
Gan et al. | Accelerating solvers for global atmospheric equations through mixed-precision data flow engine | |
CN107609141A (en) | It is a kind of that quick modelling method of probabilistic is carried out to extensive renewable energy source data | |
CN104933225B (en) | The method for realizing the extensive real-time Simulation of Fluid Mechanics Computation | |
Liu et al. | Sunwaylb: Enabling extreme-scale lattice boltzmann method based computing fluid dynamics simulations on sunway taihulight | |
Käppeli et al. | FISH: a three-dimensional parallel magnetohydrodynamics code for astrophysical applications | |
CN107403466A (en) | Ultra-large unstrctured grid generation method based on overall situation encryption | |
Williamschen et al. | Parallel anisotropic block-based adaptive mesh refinement algorithm for three-dimensional flows | |
CN112632874A (en) | Optimization method and system for numerical simulation of helicopter flow field | |
Lesser et al. | Loki: a unified multiphysics simulation framework for production | |
CN102902590A (en) | Parallel digital terrain analysis-oriented massive DEM (Digital Elevation Model) deploying and scheduling method | |
CN106846457B (en) | Octree parallel construction method for CT slice data visual reconstruction | |
CN104463940B (en) | A kind of compound tree parallel constructing method based on GPU | |
CN104599318B (en) | A kind of method and system of the seamless fusion of plant three-dimensional model gridding | |
Wang et al. | Parallel Monte Carlo method with MapReduce for option pricing | |
CN107038028B (en) | Multithreading real-time simulation method of RTDS custom element | |
Liu et al. | SunwayLB: Enabling Extreme-Scale Lattice Boltzmann Method Based Computing Fluid Dynamics Simulations on Advanced Heterogeneous Supercomputers | |
Sun et al. | Improved sampling techniques for the direct simulation Monte Carlo method |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
C14 | Grant of patent or utility model | ||
GR01 | Patent grant | ||
CF01 | Termination of patent right due to non-payment of annual fee | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20160427 Termination date: 20190702 |