CN106548512B - The generation method of grid model data - Google Patents
The generation method of grid model data Download PDFInfo
- Publication number
- CN106548512B CN106548512B CN201510606499.5A CN201510606499A CN106548512B CN 106548512 B CN106548512 B CN 106548512B CN 201510606499 A CN201510606499 A CN 201510606499A CN 106548512 B CN106548512 B CN 106548512B
- Authority
- CN
- China
- Prior art keywords
- grid model
- mesh point
- point
- block
- geologic body
- 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.)
- Active
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T17/00—Three dimensional [3D] modelling, e.g. data description of 3D objects
- G06T17/05—Geographic models
Landscapes
- Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Geometry (AREA)
- Software Systems (AREA)
- Remote Sensing (AREA)
- Computer Graphics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a kind of generation methods of grid model data, belong to geophysical prospecting technology field, solve existing grid model data creation method and spend the too long time, the low technical problem of execution efficiency.This method comprises: determining its max calculation number of threads according to the physical resource of calculate node;The grid model is divided into multiple blocks in the max calculation number of threads;Computational threads are distributed for each block;The computational threads carry out assignment calculating to the mesh point in the grid model in respective block, generate the grid model data.
Description
Technical field
The present invention relates to geophysical prospecting technology fields, specifically, are related to a kind of generation method of grid model data.
Background technique
It needs to create earthquake model in physical prospecting field, in forward modeling procedure, and grid model data is carried out to it
It calculates.
Currently, grid model data calculating step is: 1, in the closing quadrangle assumed at one, being drawn with geometry broken line
Geologic body.2, using triangulation, the closed polygon of geologic body geometrically is found out, and gives the polygon of these geologic bodies
It is assigned to attribute value.3, the closing quadrangle of hypothesis, being divided into (direction x) step-length in the horizontal direction is dx, the vertical direction (side z
To) step-length is the small quadrilateral mesh of dz, and needs to judge the position (coordinate on the direction x and the direction z where each mesh point
Value) it falls in which geologic body, then assign the attribute value of the geologic body where this mesh point to the mesh point.With such
It pushes away, calculates all mesh point attribute values, also just complete the gridding of model data, obtained grid model data.
In traditional seismic prospecting, model data gridding is completed the above three steps using serial mode, but with
The high density explored at present, wide-azimuth acquisition become new trend, the scale of model is also increasing, still use serial mode
The problems such as generation that grid model data certainly will be will cause spends the too long time, and execution efficiency is low.
Therefore, a kind of grid model data generation time that can save model data gridding, improved efficiency is needed
Method.
Summary of the invention
The purpose of the present invention is to provide a kind of generation methods of grid model data, to solve existing grid model number
According to generation spend too long time, the low technical problem of execution efficiency.
The embodiment of the present invention provides a kind of generation method of grid model data, this method comprises:
Its max calculation number of threads is determined according to the physical resource of calculate node;
The grid model is divided into multiple blocks in the max calculation number of threads;
Computational threads are distributed for each block;
The computational threads carry out assignment calculating to the mesh point in the grid model in respective block, described in generation
Grid model data.
Include: in the step of grid model is divided into multiple blocks
The grid model is divided into multiple blocks, institute according to the independence between the mesh point in the grid model
The memory address stated between the mesh point in block is continuous, if carrying out assignment between two in the grid model mesh point
There is no dependence when calculating, then there is independence between described two mesh points.
Include: in the step of grid model is divided into multiple blocks
If there is independence between the adjacent rows mesh point in the grid model and row internal net point memory address connects
It is continuous, then the mesh point in the grid model is averagely divided into multiple blocks with behavior unit;
If there is independence between the adjacent two column mesh point in the grid model and column internal net point memory address connects
It is continuous, then the mesh point in the grid model is averagely divided into multiple blocks for unit to arrange.
Include: in the step of mesh point in the grid model is averagely divided into multiple blocks with behavior unit
The quotient a and remainder b of the line number of mesh point and the number of blocks M in the grid model are calculated, the block
Number is from top to bottom respectively 0 to M-1;
If b is 0, the mesh point line number in each block is a;
If b is not 0, the mesh point line number in block that number is 0 to b-1 is a+1, and number is in the block of b to M-1
Mesh point line number be a.
Including: to arrange in the step of being averagely divided into multiple blocks for unit by the mesh point in the grid model
The quotient a and remainder b of the columns of mesh point and the number of blocks M in the grid model are calculated, the block
Number is from top to bottom respectively 0 to N-1;
If b is 0, the mesh point columns in each block is a;
If b is not 0, the mesh point columns in block that number is 0 to b-1 is a+1, and number is in the block of b to N-1
Mesh point columns be a.
The number of the block is max calculation number of threads.
Include: in the step of determining max calculation number of threads
The processor nucleus number of the calculate node is obtained by application programming interface, so that it is determined that the calculate node
Max calculation number of threads.
Include: in the step of carrying out assignment calculating to the mesh point in the grid model
If the mesh point in the grid model is divided into multiple blocks with behavior unit, the computational threads are corresponding
Assignment calculating successively is carried out to mesh point according to sequence from left to right, from top to bottom in block.
Include: in the step of carrying out assignment calculating to the mesh point in the grid model
If the mesh point in the grid model is divided into multiple blocks to arrange for unit, the computational threads are corresponding
Assignment calculating successively is carried out to mesh point according to sequence from top to bottom, from left to right in block.
The grid model is the closing quadrangle for being divided into multiple grids, utilizes triangulation including multiple
The geologic body of drafting, in the grid model mesh point carry out assignment calculating the step of in include:
Step 1, the coordinate of target network lattice point is obtained;
Step 2, judge whether to have the triangle in geologic body and geologic body where the coordinate of other mesh points to be remembered
Record;If it is not, then carrying out step 4;
If it is, judging whether the coordinate of the target network lattice point is located at the triangle in the geologic body being recorded
In;If it is, carrying out step 5;
If it is not, then judging whether the coordinate of the target network lattice point is located at other triangles in the geologic body being recorded
In shape;If it is, carrying out step 5;
If it is not, then carrying out step 3;
Step 3, it is searched in other geologic bodies in geologic body and geologic body where the coordinate of the target network lattice point
Triangle, and carry out step 5;
Step 4, it is searched in all geologic bodies in geologic body and geologic body where the coordinate of the target network lattice point
Triangle, and carry out step 5;
Step 5, the triangle in the geologic body and geologic body where the coordinate of the target network lattice point is recorded, and will be described
The grid property value of target network lattice point is assigned a value of the attribute value of the geologic body where it.
The generation method of grid model data provided in an embodiment of the present invention is by using the parallel mode of multithreading to net
Mesh point in lattice model carries out assignment calculating, saves the plenty of time compared to serial mode, while to grid point location
When with assignment, efficiently solves invalid triangle and search number, significantly improve the formation efficiency of grid model data.
Other features and advantages of the present invention will be illustrated in the following description, also, partial becomes from specification
It obtains it is clear that understand through the implementation of the invention.The objectives and other advantages of the invention can be by specification, right
Specifically noted structure is achieved and obtained in claim and attached drawing.
Detailed description of the invention
It, below will be to required in embodiment description for the clearer technical solution illustrated in the embodiment of the present invention
Attached drawing does simple introduction:
Fig. 1 is the flow diagram of the generation method of grid model data provided in an embodiment of the present invention;
Fig. 2 is grid model schematic diagram provided in an embodiment of the present invention;
Fig. 3 is the flow diagram that mesh point provided in an embodiment of the present invention carries out assignment calculating.
Specific embodiment
Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings and examples, how to apply to the present invention whereby
Technological means solves technical problem, and the realization process for reaching technical effect can fully understand and implement.It needs to illustrate
As long as not constituting conflict, each feature in each embodiment and each embodiment in the present invention can be combined with each other,
It is within the scope of the present invention to be formed by technical solution.
The embodiment of the present invention provides a kind of generation method of grid model data, as shown in Figure 1, this method comprises:
In a step 101, its max calculation number of threads is determined according to the physical resource of calculate node.First by master control line
Journey obtains the physical resource of calculate node, and calculate node can be server or PC etc..Then according to the physics of calculate node
The given computational threads number that will be opened of resource.
In order to enable the generation method of grid model data can be cross-platform use, the embodiment of the invention provides across flat
The method that the acquisition calculate node current system of platform can use nucleus number, i.e., obtain calculate node by application programming interface API
Processor nucleus number, so that it is determined that the max calculation number of threads of calculate node.Specifically, first to current calculate node
Operating system does pre- judgement, if it is windows system, then returns to current calculate using GetsystemInfo () this API
The processor nucleus number of node;If it is linux system, then it is true current calculate node to be returned to using get_nprocs () this API
Just available processor nucleus number.
Optionally, in embodiments of the present invention, as soon as being calculated using the data that a processor cores carry out a thread,
It is to say that really available processor nucleus number is its max calculation number of threads to current calculate node.
In a step 102, according to the independence between the mesh point in grid model by grid model in max calculation line
Multiple blocks are divided into number of passes mesh, the memory address between mesh point in block is continuous, if two nets in grid model
There is no dependence when carrying out assignment calculating between lattice point, then there is independence between two mesh points.
Specifically, the concurrency demand of analytical calculation task first, as shown in Fig. 2, grid model be divided into it is multiple
The closing quadrangle of grid, including multiple geologic bodies drawn using triangulation, in the life of grid model data
At in the process, time-consuming more part is the calculating that assignment is carried out to the mesh point in grid model, i.e., will be in grid model
Each mesh point is assigned to the attribute value of geologic body where it, and the attribute value of geologic body includes velocity of longitudinal wave, shear wave velocity, in density
One or more.Carrying out assignment to mesh point one by one can take a substantial amount of time, and by multithreading operation simultaneously to multiple
Mesh point, which carries out assignment, will greatly improve efficiency.Therefore, the generation of grid model data this general assignment is resolved into multiple
Subtask opens up concurrency, determines that the parallel subtask of multiple computational threads falls for mesh point by the analysis to each subtask
Enter judgement and the attribute assignment operations of geology body position.
Then, in the case where guaranteeing the continuous situation of memory address between the mesh point in each block, according to mesh point it
Between independence grid model is divided into multiple blocks, the assignment that multithreadings are carried out to distribute multiple threads calculates.Block
Between with independence be premise that block independently carries out assignment calculating, the memory address company between the mesh point in guarantee block
It is continuous, frequently jumping when can be addressed to avoid memory address space.
Further, if between adjacent rows mesh point in grid model with independence and row internal net point memory
Location is continuous, then the mesh point in grid model is averagely divided into multiple blocks with behavior unit.That is each block packet
Containing multiple adjacent rows being made of mesh point, since the mesh point memory address in every row is continuous, so all in block
Mesh point memory address is continuous.And there is between the adjacent row being made of mesh point of any two independence in grid model,
So having independence between block.
Specifically, total line number of mesh point and the quotient a of number of blocks M and remainder b, the volume of block in grid model are calculated
It number is from top to bottom respectively 0 to M-1;
If b is 0, i.e., total line number divides exactly with block counts, then the mesh point line number in each block is a;
If b is not 0, i.e., total line number and block counts are aliquant, then the mesh point line number in block that number is 0 to b-1
For a+1, number is that the mesh point line number in the block of b to M-1 is a.So far, all mesh points in grid data are all distributed
It has arrived in corresponding block.
Grid model is divided into block is roughly the same with behavior unit from top to bottom with above-mentioned, if adjacent in grid model
There is independence between two column mesh points and column internal net point memory address is continuous, then be to arrange by the mesh point in grid model
Unit is averagely divided into multiple blocks.
Specifically, the columns of mesh point and the quotient a of number of blocks M and remainder b, the number of block in grid model are calculated
It is from top to bottom respectively 0 to N-1;
If b is 0, i.e., total line number divides exactly with block counts, then the mesh point columns in each block is a;
If b is not 0, i.e., total line number and block counts are aliquant, then the mesh point columns in block that number is 0 to b-1
For a+1, number is that the mesh point columns in the block of b to N-1 is a.
In step 103, computational threads are distributed for each block.Optionally, in embodiments of the present invention, using at one
The data calculating that device kernel carries out a thread is managed, that is, distributes to one computational threads of each block.Likewise, can also basis
Actual demand is calculated using the data that a processor cores carry out multiple threads, or carries out one using multiple processor cores
The data of a thread calculate, herein with no restrictions.
At step 104, computational threads carry out assignment calculating to the mesh point in grid model in respective block, generate
Grid model data.Wherein, if the mesh point in grid model is divided into multiple blocks with behavior unit, computational threads are in phase
It answers in block and assignment calculating successively is carried out to mesh point according to sequence from left to right, from top to bottom.Due in row internal net point
It is continuous to deposit address, mesh point memory address is continuous end to end for adjacent rows, therefore the behaviour of a line has from left to right been carried out in block
After work, next line is taken turns to from top to bottom, is from left to right operated again in next line, allows for the grid in block in this way
It when point carries out assignment calculating, keeps the addressing space of memory address continuous always, avoids frequency when memory address space addressing
It is numerous to jump.
Likewise, computational threads are corresponding if the mesh point in grid model is divided into multiple blocks to arrange for unit
Assignment calculating successively is carried out to mesh point according to sequence from top to bottom, from left to right in block.
Finally, discharging related resource after all computational threads execute.
Below by taking grid model shown in Fig. 2 as an example, to the generation side of grid model data provided in an embodiment of the present invention
Step 101 to step 103 in method is specifically described.
The grid model is the closing quadrangle for being divided into multiple grids, is drawn including multiple using triangulation
The geologic body in polygon of system, wherein the quantity of the mesh point on (direction x) is Nx in the horizontal direction, in depth direction (z
Direction) on mesh point quantity be Nz.Mesh point memory address addressing space in the horizontal direction is continuous.
The processor nucleus number for getting calculate node first is NCORES, and foundation processor nucleus number starts computational threads
Number NTHREADS=NCORES.One data block distributes to a computational threads.
Then it according to the data independence of model data, determines in grid model, master control thread is according to from top to bottom
The division of data block when sequence progress multi-threaded parallel.
The division mode of data block uses the principle of mean allocation as far as possible.In grid model, the scale of Z-direction is
NZ, then it is NZ that Z-direction, which is used to divide the line number of data block, and number is respectively 0 to NZ-1.It is first when data block divides
Obtain the basic number of lines BASE=NZ/NTHREADS that each computational threads are assigned to, and increment INCREMENT=NZ%
NTHREADS (/ indicate to ask quotient, % expression rems).Show that NZ/NTHREADS is divided exactly when INCREMENT is 0, all meters
Calculating the number of lines that thread obtains all is BASE, i.e., the mesh point row number that the computational threads that number is 0 are assigned to is 0 to BASE-
1, and so on, the mesh point row number that the computational threads that number is NTHREADS-1 obtain be BASE* (NTHREADS-1) extremely
NZ-1.Show that NZ/NTHREADS is not divided exactly when INCREMENT is not 0, the computational threads that number is 0 to INCREMENT-1 obtain
The mesh point number of lines arrived is BASE+1, and the mesh point number of lines that remaining computational threads obtains is BASE, that is, numbers the meter for being 0
Calculating the mesh point row number that thread is assigned to is 0 to BASE, and so on, the computational threads that number is NTHREADS-1 obtain
Mesh point row number is NZ-BASE to NZ-1.
The generation method of grid model data provided in an embodiment of the present invention is by using the parallel mode of multithreading to net
Mesh point in lattice model carries out assignment calculating, greatly improves the formation speed of grid model data, is a kind of efficient
Grid model data creation method.Using this method for 18000 × 15000 model, linearly accelerate when opening 2 threads
Than reaching 1.8, when opening 4 computational threads, linear speed-up ratio reaches 3.2.
Further, in embodiments of the present invention, as shown in figure 3, carrying out assignment calculating to the mesh point in grid model
The step of specifically include:
S1: the coordinate of target network lattice point is obtained.
In grid model, according to sequence from left to right, from top to bottom, the coordinate value of mesh point is obtained.In addition, may be used also
First to judge acquired coordinate value whether in the grid model, if not terminating process if;If carrying out step if
S2。
S2: judge whether to have the triangle in geologic body and geologic body where the coordinate of other mesh points and be recorded.
If it is not, carrying out step S4;It has been recorded if so, then judging whether the coordinate of target network lattice point is located at
In triangle in geologic body, it usually can first judge that the geologic body where the coordinate of previous mesh point (can be described as recently
Plastid) in triangle.
If it is, carrying out step S5;If it is not, then judging whether the coordinate of target network lattice point is located at the ground being recorded
In other triangles in plastid, i.e., then judge whether the coordinate of target network lattice point is located at other triangles in nearest geologic body
In.
If it is, carrying out step S5;If it is not, then carrying out step S3.
S3: the triangle in geologic body and geologic body in other geologic bodies where the coordinate of search target network lattice point,
And carry out step S5.
Specifically, when the coordinate of target network lattice point is not in nearest geologic body, then other other than nearest geologic body
The triangle in geologic body and geologic body in geologic body where the coordinate of search target network lattice point.If except nearest geologic body with
Outside, there are also other geologic bodies to be recorded, then the geologic body that first search has been recorded searches again for the geology not being recorded also
Body, because the coordinate of target network lattice point is relatively larger a possibility that being located at the geologic body being recorded.
If the triangle in geologic body and geologic body where having searched the coordinate of target network lattice point, carries out step
S5, the otherwise still each geologic body of cyclic search, until having searched for all geologic bodies.Certainly, all geologic bodies have been searched for
When, it must can search the triangle in the geologic body and geologic body where the coordinate of target network lattice point.
S4: the triangle in geologic body and geologic body in all geologic bodies where the coordinate of search target network lattice point,
And carry out step S5.
If the triangle in the geologic body and geologic body where the coordinate of other mesh points is recorded not yet, in institute
There is the triangle in the geologic body and geologic body in geologic body where the coordinate of search target network lattice point.
If the triangle in geologic body and geologic body where having searched the coordinate of target network lattice point, carries out step
S5, the otherwise still each geologic body of cyclic search, until having searched for all geologic bodies.Certainly, all geologic bodies have been searched for
When, it must can search the triangle in the geologic body and geologic body where the coordinate of target network lattice point.
S5: the triangle in geologic body and geologic body where the coordinate of record target network lattice point, and by target network lattice point
Grid property value be assigned a value of the attribute value of the geologic body where it.
In the present embodiment, the attribute value of geologic body may include velocity of longitudinal wave, shear wave velocity, one or more in density
It is a.After the completion of target network lattice point assignment, continue return step S1, assignment, Ji Kesheng successively are carried out to subsequent each mesh point
At grid model data.
The following are the C Plus Plus puppet generations that the mesh point provided in an embodiment of the present invention in grid model carries out assignment calculating
Code.Certainly, in other embodiments, pseudocode can also be defined by various other language.
Wherein, nx indicates the number of mesh point when model data gridding on the direction x, and nz indicates model data gridding
When the direction z on mesh point number;M indicates the position of the mesh point in the direction z when previous cycle is searched for;N expression currently follows
The position of the mesh point in the direction x when loops detection;IsPrev indicates whether the geologic body where the coordinate of a upper mesh point
And the flag bit that the triangle in geologic body is remembered, isPrev=1 indicate that coordinate value is remembered, isPrev=0 expression does not have
There is coordinate value to be remembered;I indicates geology body position when previous cycle search;J indicates triangle morpheme when previous cycle search
It sets;PrevTriPos indicates the triangle position in current geologic body;NPrevTriangles indicates three in current geologic body
Angular total number;PrevRegionPos indicates the position of current geologic body;NRegion indicates the total number of all geologic bodies.
In the generation method of grid model data provided in an embodiment of the present invention, in the coordinate for obtaining a target network lattice point
Afterwards, it is first determined whether the triangle in geologic body and geologic body where having the coordinate of other mesh points is recorded.If
Have, then preferentially judge whether the coordinate of target network lattice point is located in the triangle in the geologic body being recorded, and by target network
The grid property value of lattice point is assigned a value of the attribute value of the geologic body where it.
In the present embodiment, (target network lattice point and previous mesh point when target network lattice point is at the boundary line of geologic body
In different geologic bodies), the worst algorithm complexity is O (nx1×nz1×(R1+…+Rm)), wherein nx1For the side of geologic body
The Grid dimension on the direction x at boundary line, nz1For the Grid dimension on the direction z at the boundary line of geologic body, [R1,…,Rk]
For the 1st triangle number into k-th of geologic body.And (the target when target network lattice point is at the boundary line of non-geologic body
Mesh point and previous mesh point are located in the same geologic body), average algorithm complexity is O (nx2×nz2), wherein nx2
For the Grid dimension on the direction x at the boundary line of non-geologic body, nz2For the net on the direction z at the boundary line of non-geologic body
Lattice point number.
Because target network lattice point is sequentially obtained in grid model, the target network lattice point currently obtained with it is previous
The target network lattice point of secondary acquisition is to be located at the same geologic body in most cases, i.e., the boundary line of all geologic bodies is in x
Few part that Grid dimension Zhan on direction and the direction z always counts.If the target network lattice point currently obtained is once obtained with preceding
The target gridding point taken is located at the same geologic body, so that it may which the attribute value of the geologic body is quickly assigned to the mesh currently obtained
Mesh point is marked, is looped to determine without each triangle to each geologic body, so as to make the complexity of algorithm
It reduces, significantly improves the formation efficiency of grid model data.
The generation method of grid model data provided in an embodiment of the present invention is by using the parallel mode of multithreading to net
Mesh point in lattice model carries out assignment calculating, saves the plenty of time compared to serial mode, while to grid point location
When with assignment, efficiently solves invalid triangle and search number, significantly improve the formation efficiency of grid model data.
While it is disclosed that embodiment content as above but described only to facilitate understanding the present invention and adopting
Embodiment is not intended to limit the invention.Any those skilled in the art to which this invention pertains are not departing from this
Under the premise of the disclosed spirit and scope of invention, any modification and change can be made in the implementing form and in details,
But scope of patent protection of the invention, still should be subject to the scope of the claims as defined in the appended claims.
Claims (9)
1. a kind of generation method of grid model data characterized by comprising
Its max calculation number of threads is determined according to the physical resource of calculate node;
The grid model is divided into multiple blocks in the max calculation number of threads;
Computational threads are distributed for each block;
The computational threads carry out assignment calculating to the mesh point in the grid model in respective block, generate the grid
Model data,
Wherein, the grid model is the closing quadrangle for being divided into multiple grids, is calculated including multiple using triangulation
Method draw geologic body, in the grid model mesh point carry out assignment calculating the step of in include:
Step 1, the coordinate of target network lattice point is obtained;
Step 2, judge whether to have the triangle in geologic body and geologic body where the coordinate of other mesh points to be recorded;Such as
Fruit is no, then carries out step 4;
If it is, judging whether the coordinate of the target network lattice point is located in the triangle in the geologic body being recorded;Such as
Fruit is then to carry out step 5;
If it is not, then judging whether the coordinate of the target network lattice point is located at other triangles in the geologic body being recorded
In;If it is, carrying out step 5;
If it is not, then carrying out step 3;
Step 3, the triangle in geologic body and geologic body where the coordinate of the target network lattice point is searched in other geologic bodies
Shape, and carry out step 5;
Step 4, the triangle in geologic body and geologic body where the coordinate of the target network lattice point is searched in all geologic bodies
Shape, and carry out step 5;
Step 5, the triangle in the geologic body and geologic body where the coordinate of the target network lattice point is recorded, and by the target
The grid property value of mesh point is assigned a value of the attribute value of the geologic body where it.
2. the method according to claim 1, wherein the step of the grid model is divided into multiple blocks
In include:
The grid model is divided into multiple blocks, the area according to the independence between the mesh point in the grid model
The memory address between mesh point in block is continuous, if carrying out assignment calculating between two mesh points in the grid model
When there is no dependence, then between described two mesh points have independence.
3. according to the method described in claim 2, it is characterized in that, the step of the grid model is divided into multiple blocks
In include:
If there is independence between the adjacent rows mesh point in the grid model and row internal net point memory address is continuous,
Mesh point in the grid model is averagely divided into multiple blocks with behavior unit;
If there is independence between the adjacent two column mesh point in the grid model and column internal net point memory address is continuous,
Mesh point in the grid model is averagely divided into multiple blocks for unit to arrange.
4. according to the method described in claim 3, it is characterized in that, by the mesh point in the grid model with behavior unit
Include: in the step of being averagely divided into multiple blocks
Calculate the quotient a and remainder b of the line number of mesh point and the number of blocks M in the grid model, the number of the block
It is from top to bottom respectively 0 to M-1;
If b is 0, the mesh point line number in each block is a;
If b is not 0, the mesh point line number in block that number is 0 to b-1 is a+1, and number is the net in the block of b to M-1
Lattice point line number is a.
5. according to the method described in claim 3, it is characterized in that, by the mesh point in the grid model to arrange as unit
Include: in the step of being averagely divided into multiple blocks
Calculate the quotient a and remainder b of the columns of mesh point and the number of blocks M in the grid model, the number of the block
It is from top to bottom respectively 0 to N-1;
If b is 0, the mesh point columns in each block is a;
If b is not 0, the mesh point columns in block that number is 0 to b-1 is a+1, and number is the net in the block of b to N-1
Lattice point columns is a.
6. method according to any one of claims 1 to 5, which is characterized in that the number of the block is max calculation line
Number of passes mesh.
7. according to the method described in claim 6, it is characterized in that, including: in the step of determining max calculation number of threads
The processor nucleus number of the calculate node is obtained by application programming interface, so that it is determined that the calculate node is most
It is big to calculate number of threads.
8. the method according to the description of claim 7 is characterized in that carrying out assignment meter to the mesh point in the grid model
Include: in the step of calculation
If the mesh point in the grid model is divided into multiple blocks with behavior unit, the computational threads are in respective block
It is interior that assignment calculating successively is carried out to mesh point according to sequence from left to right, from top to bottom.
9. the method according to the description of claim 7 is characterized in that carrying out assignment meter to the mesh point in the grid model
Include: in the step of calculation
If the mesh point in the grid model is divided into multiple blocks to arrange for unit, the computational threads are in respective block
It is interior that assignment calculating successively is carried out to mesh point according to sequence from top to bottom, from left to right.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510606499.5A CN106548512B (en) | 2015-09-22 | 2015-09-22 | The generation method of grid model data |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510606499.5A CN106548512B (en) | 2015-09-22 | 2015-09-22 | The generation method of grid model data |
Publications (2)
Publication Number | Publication Date |
---|---|
CN106548512A CN106548512A (en) | 2017-03-29 |
CN106548512B true CN106548512B (en) | 2019-10-29 |
Family
ID=58365238
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201510606499.5A Active CN106548512B (en) | 2015-09-22 | 2015-09-22 | The generation method of grid model data |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106548512B (en) |
Families Citing this family (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110211234B (en) * | 2019-05-08 | 2023-09-22 | 上海索辰信息科技有限公司 | Grid model stitching system and method |
CN110503193B (en) * | 2019-07-25 | 2022-02-22 | 瑞芯微电子股份有限公司 | ROI-based pooling operation method and circuit |
CN117931419A (en) * | 2023-05-15 | 2024-04-26 | 苏州互微智速科技有限公司 | High-performance calculation method for implicit fluctuation rate |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102736947A (en) * | 2011-05-06 | 2012-10-17 | 新奥特(北京)视频技术有限公司 | Multithread realization method for rasterization stage in graphic rendering |
CN103543468A (en) * | 2013-10-28 | 2014-01-29 | 北京大学 | Method and system for earthquake forward modeling |
CN104318035A (en) * | 2014-11-07 | 2015-01-28 | 中铁第四勘察设计院集团有限公司 | General-to-part type triangular network multi-threading parallel generation method for massive terrain big data cloud |
CN104765589A (en) * | 2014-01-02 | 2015-07-08 | 广州中国科学院软件应用技术研究所 | Grid parallel preprocessing method based on MPI |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2014032008A2 (en) * | 2012-08-23 | 2014-02-27 | Old Dominion University Reasearch Foundation | Method and system for generating mesh from images |
-
2015
- 2015-09-22 CN CN201510606499.5A patent/CN106548512B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102736947A (en) * | 2011-05-06 | 2012-10-17 | 新奥特(北京)视频技术有限公司 | Multithread realization method for rasterization stage in graphic rendering |
CN103543468A (en) * | 2013-10-28 | 2014-01-29 | 北京大学 | Method and system for earthquake forward modeling |
CN104765589A (en) * | 2014-01-02 | 2015-07-08 | 广州中国科学院软件应用技术研究所 | Grid parallel preprocessing method based on MPI |
CN104318035A (en) * | 2014-11-07 | 2015-01-28 | 中铁第四勘察设计院集团有限公司 | General-to-part type triangular network multi-threading parallel generation method for massive terrain big data cloud |
Also Published As
Publication number | Publication date |
---|---|
CN106548512A (en) | 2017-03-29 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN103970960A (en) | Grid-free Galerkin method structural topology optimization method based on GPU parallel acceleration | |
CN104765589B (en) | Grid parallel computation preprocess method based on MPI | |
CN106548512B (en) | The generation method of grid model data | |
CN103279633A (en) | Brain fiber three-dimensional display method based on diffusion-weighted magnetic resonance data | |
CN105654552B (en) | A kind of quick Delaunay network construction methods towards Arbitrary distribution large-scale point cloud data | |
CN107908913B (en) | Earth power digital-analog method based on parallel computer | |
CN102681972A (en) | Method for accelerating lattice-Boltzmann by utilizing graphic processing units (GPUs) | |
Dhar et al. | GDP: GPU accelerated detailed placement | |
CN108828669A (en) | A kind of two-dimensional intersection survey line static corrections processing method, apparatus and system | |
Bisson et al. | Multiscale hemodynamics using GPU clusters | |
CN112989746B (en) | Integrated circuit simulation coarse grain parallel method and device for multi-thread management | |
CN103886129B (en) | By the discrete method and apparatus to reservoir grid model of log data | |
CN106485030A (en) | A kind of symmetrical border processing method for SPH algorithm | |
CN104614761B (en) | A kind of double-deck flood filling seismic horizon space method for automatic tracking and device | |
CN106373192B (en) | A kind of non-topological coherence three-dimensional grid block tracing algorithm | |
CN101354710B (en) | Method and apparatus of line segment intersection | |
CN109670001A (en) | Polygonal gird GPU parallel calculating method based on CUDA | |
de Gomensoro Malheiros et al. | Simple and efficient approximate nearest neighbor search using spatial sorting | |
CN105572730B (en) | 3 D complex structure sound wave forward modeling method | |
CN109388876A (en) | A kind of groundwater solute transfer numerical simulation parallel acceleration method | |
CN110968930A (en) | Geological variable attribute interpolation method and system | |
CN105573834B (en) | A kind of higher-dimension vocabulary tree constructing method based on heterogeneous platform | |
Garanzha | Efficient clustered BVH update algorithm for highly-dynamic models | |
de Gomensoro Malheiros et al. | Spatial sorting: an efficient strategy for approximate nearest neighbor searching | |
Gerritsen et al. | Parallel implementations of streamline simulators |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |