KR20210044405A - Visualization pipeline apparatus for cluster based scientific visualization tools and its visualization method - Google Patents

Abstract

The present invention relates to a visualization pipeline device for a cluster-based scientific visualization tool and to a visualization method thereof. The device includes: a data patch pipe converting input data input from other components within the visualization tool into a processable form and delivers the data; a plurality of visualization pipelines each using a plurality of visualization pipes to perform a visualization operation on the input data; and a result collection pipe collecting the results generated through the plurality of visualization pipes, converting the results into a desired form in the other component, and returning the converted results. The implementation and modification of the visualization pipe can be facilitated, as well as improve portability.

Description

A visualization pipeline device for cluster-based scientific visualization tools and its visualization method {VISUALIZATION PIPELINE APPARATUS FOR CLUSTER BASED SCIENTIFIC VISUALIZATION TOOLS AND ITS VISUALIZATION METHOD}

The present invention includes a standard pipe connected to each other in a random order, and includes a special pipe dedicated to communication with other components in the visualization tool, and the visualization pipeline is formed by combining the pipes according to the user's request. The present invention relates to a visualization pipeline apparatus and a visualization method for a cluster-based scientific visualization tool capable of improving portability as well as easy implementation and modification of a visualization pipe by being generated in real time and performing a visualization operation.

As is well known, scientists and engineers collect observational or experimental data for natural phenomena analysis, product development, etc., and scientific visualization refers to a set of methods for transforming these scientific data into visual information.

Such scientific visualization is widely used in fields such as medical, meteorology, aviation, space, heavy industry, etc., as it helps to obtain not only insights from vast and complex scientific data, but also meaningful information, and thanks to its usefulness, scientific visualization techniques Have been actively researched, and excellent open source libraries (eg, VTK, etc.) and visualization tools have been developed.

On the other hand, the visualization pipeline is one of the methods of defining the flow of data and visualization operations from the input data to the final visualization result.Since it is concise and powerful, it is a key processing technique for visualization libraries and visualization tools. It is widely used, and in particular, in the case of a field in which several visualization techniques are used in combination, the design and implementation of a visualization pipeline capable of flexible configuration is indispensable.

For example, FIG. 1 is a diagram showing a case in which a conventional visualization technique is used in combination, and a visualization technique (e.g., glyph) is applied to the result of clipping input data for flow analysis inside a reactor. Is shown.

In recent years, as the precision of measurement devices and the scale of simulations increase, the size of scientific data has increased significantly. However, it is difficult to process large scientific data using limited computational resources (e.g., single node PC, workstation, etc.). Due to its limitations, institutions handling large scientific data use open source visualization tools that support cluster-based visualization operations, or develop and utilize cluster-based visualization tools themselves.

For example, GLOVE, a distributed shared memory-based parallel visualization tool that defines the components of a large scientific data visualization tool and utilizes open software, was developed.This GLOVE is an existing open source-based visualization tool (for example, Paraview , VisIt, etc.) and proved its effectiveness, but there is a limit to supporting only a single visualization technique due to the absence of a visualization pipeline that supports cluster-based distributed processing.

1. Korean Patent Publication No. 10-2006-0135104 (published on December 29, 2006) 2. Korean Patent Publication No. 10-2017-0015789 (published on Feb. 9, 2017)

The present invention includes a standard pipe connected to each other in a random order, and includes a special pipe dedicated to communication with other components in the visualization tool, and the visualization pipeline is formed by combining the pipes according to the user's request. It is intended to provide a visualization pipeline apparatus and a visualization method for a cluster-based scientific visualization tool capable of improving portability as well as easy implementation and modification of a visualization pipe by being generated in real time and performing a visualization operation.

In addition, the present invention converts input data input from another component in a visualization tool into a form that can be processed and transmits it, and performs a visualization operation on the input data using a plurality of visualization pipes. By collecting the results generated through a plurality of visualization pipes, converting them to the desired form in the corresponding component, and returning them, it is possible to ensure flexible interconnection using a standardized pipe structure, and can be implemented and managed independently of the visualization tool. An attempt is made to provide a visualization pipeline device and a visualization method for a cluster-based scientific visualization tool.

The objects of the embodiments of the present invention are not limited to the above-mentioned objects, and other objects not mentioned will be clearly understood by those of ordinary skill in the technical field to which the present invention pertains from the following description. .

According to an aspect of the present invention, a data patch pipe that converts input data input from another component in a visualization tool into a form that can be processed and transmits it, and a visualization operation on the input data using a plurality of visualization pipes, respectively. A visualization pipeline device for a cluster-based scientific visualization tool is provided that includes a plurality of visualization pipelines and a result collection pipe that collects the results generated through the plurality of visualization pipes, converts the result into a desired form in the other component, and returns it. Can be.

In addition, according to an aspect of the present invention, the plurality of visualization pipes each initializes the visualization pipes to prepare the visualization operation, and requests data to be processed from the preceding visualization pipe to provide the data to be processed. A visualization pipeline apparatus for a cluster-based scientific visualization tool may be provided, including a pipe body part that performs the visualization operation on the image, and a post-processor part that generates result data by completing the visualization operation.

In addition, according to an aspect of the present invention, the plurality of visualization pipes are connected in reverse order from the last visualization pipe, and when the visualization pipes are connected, operation information necessary for the visualization operation is accumulated and transmitted to the preceding visualization pipe. A visualization pipeline device for a cluster-based scientific visualization tool may be provided.

In addition, according to an aspect of the present invention, the data patch pipe may be provided with a visualization pipeline apparatus for a cluster-based scientific visualization tool that is subordinately initialized to each first visualization pipe of the plurality of visualization pipelines.

Further, according to an aspect of the present invention, the data patch pipe is a cluster that determines the input data to be input from the other component and a conversion method based on the operation information and an input data format input to the first visualization pipe. A visualization pipeline device may be provided for the underlying scientific visualization tool.

In addition, according to an aspect of the present invention, the result collection pipe provides a cluster-based scientific visualization tool that requests next processed data to the visualization pipe at the front end, and is propagated to the visualization pipe at the front end and transmitted to the data patch pipe. A visualization pipeline device for can be provided.

In addition, according to an aspect of the present invention, the result collection pipe is initialized dependently on each of the last visualization pipes provided in the plurality of visualization pipelines, and is used for a cluster-based scientific visualization tool that determines a final result generation method. A visualization device of a visualization pipeline device may be provided.

According to another aspect of the present invention, the step of requesting the next processing data from the result collection pipe to the visualization pipe at the front end, the step of transmitting the data request to the data patch pipe by propagating each to the visualization pipe at the front end, and the data Transforming input data input from other components in the visualization tool in a patch pipe into a form that can be processed, and performing a visualization operation on the input data using a plurality of visualization pipes, respectively, in a plurality of visualization pipelines. And, a visualization method of a visualization pipeline device for a cluster-based scientific visualization tool comprising the step of collecting the result generated through the plurality of visualization pipes in the result collection pipe, converting the result into a desired form in the other component, and returning it. Can be provided.

In addition, according to another aspect of the present invention, the plurality of visualization pipes are connected in reverse order from the last visualization pipe, and when the visualization pipes are connected, operation information required for the visualization operation is accumulated and transmitted to the preceding visualization pipe. A visualization method of a visualization pipeline device for a cluster-based scientific visualization tool may be provided.

In addition, according to another aspect of the present invention, the data patch pipe is a visualization method of a visualization pipeline device for a cluster-based scientific visualization tool that is subordinately initialized to each of the first visualization pipes of the plurality of visualization pipelines. I can.

Further, according to another aspect of the present invention, the data patch pipe is a cluster that determines the input data to be input from the other component and a conversion method based on the operation information and an input data format input to the first visualization pipe. A visualization method of a visualization pipeline device for an underlying scientific visualization tool may be provided.

In addition, according to another aspect of the present invention, the result collection pipe is initialized dependently on each of the last visualization pipes provided in the plurality of visualization pipelines, and is used for a cluster-based scientific visualization tool that determines a final result generation method. A visualization method of a visualization pipeline device may be provided.

The present invention includes a standard pipe connected to each other in a random order, and includes a special pipe dedicated to communication with other components in the visualization tool, and the visualization pipeline is formed by combining the pipes according to the user's request. By being generated in real time and performing a visualization operation, not only can the visualization pipe be easily implemented and modified, but also portability can be improved.

In addition, the present invention converts input data input from another component in a visualization tool into a form that can be processed and transmits it, and performs a visualization operation on the input data using a plurality of visualization pipes. By collecting the results generated through a plurality of visualization pipes, converting them into a desired form in the corresponding component, and returning them, it is possible to ensure flexible interconnection using a standardized pipe structure, and can be implemented and managed independently of the visualization tools. .

1 is a diagram showing an example of using a conventional visualization technique in combination,
2 is a block diagram showing a visualization pipeline device for a cluster-based scientific visualization tool according to an embodiment of the present invention.
3 is a block diagram illustrating a structure of a standardized visualization pipe according to an embodiment of the present invention,
4 is a flowchart showing a process of visualization using a visualization pipeline device for a cluster-based scientific visualization tool according to another embodiment of the present invention.
5 to 12 are diagrams for explaining visualization results and implementation results using a visualization pipeline device for a cluster-based scientific visualization tool according to an embodiment of the present invention.

Advantages and features of the embodiments of the present invention, and a method of achieving them will be apparent with reference to the embodiments described later in detail together with the accompanying drawings. However, the present invention is not limited to the embodiments disclosed below, but may be implemented in a variety of different forms, and only these embodiments make the disclosure of the present invention complete, and are common knowledge in the technical field to which the present invention pertains. It is provided to completely inform the scope of the invention to those who have, and the invention is only defined by the scope of the claims. The same reference numerals refer to the same elements throughout the specification.

In describing embodiments of the present invention, if it is determined that a detailed description of a known function or configuration may unnecessarily obscure the subject matter of the present invention, a detailed description thereof will be omitted. In addition, terms to be described later are terms defined in consideration of functions in an embodiment of the present invention, which may vary according to the intention or custom of users or operators. Therefore, the definition should be made based on the contents throughout this specification.

Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings.

FIG. 2 is a block diagram showing a visualization pipeline device for a cluster-based scientific visualization tool according to an embodiment of the present invention, and FIG. 3 is a block diagram illustrating a standardized visualization pipe structure according to an embodiment of the present invention. It is a degree.

2 and 3, a visualization pipeline apparatus for a cluster-based scientific visualization tool according to an embodiment of the present invention includes a data patch pipe 100, a plurality of visualization pipelines 200, and a result collection pipe 300. ) And the like.

The data patch pipe 100 is one of the interface pipes, and can convert input data input from other components in the visualization tool into a form that can be processed and transmit it. A plurality of visualization pipes 210 included in the plurality of visualization pipelines 200 /1-210/n) can be initialized dependently on each of the first visualization pipes.

In addition, a plurality of visualization pipes (210/1-210/n) are connected in reverse order from the last visualization pipe, and when the visualization pipe is connected, operation information necessary for visualization operation is accumulated and transmitted to the preceding visualization pipe. The 100 may determine input data to be input from an external component and a conversion method based on the operation information and the input data format input to the first visualization pipe.

A plurality of visualization pipelines (200/1-200/m, a plurality of visualization pipe blocks) each include a plurality of visualization pipes 210/1-210/n to perform visualization operations on input data. The visualization pipes 210/1 to 210/n may each include a pre-treatment unit 211, a pipe body 212, a post-treatment unit 213, and the like.

Here, the preprocessing unit 211 may initialize each visualization pipe to prepare a visualization operation, and the pipe body 212 requests data to be processed from the preceding visualization pipe to perform a visualization operation on the data to be processed. In addition, the post-processing unit 213 may generate result data by completing the visualization operation.

The plurality of visualization pipes 210/1 to 210/n may be connected in reverse order from the last visualization pipe, and when the visualization pipe is connected, operation information necessary for a visualization operation may be accumulated and transmitted to the preceding visualization pipe.

The result collection pipe 300 is one of the interface pipes, and may collect the result generated through the plurality of visualization pipes 210/1-210/n, convert it into a desired shape in another component in the visualization tool, and return it.

The result collection pipe 300 is initialized dependently on each of the last visualization pipes provided in the plurality of visualization pipes 210/1 to 210/n, so that a final result generation method may be determined.

When the visualization pipeline device for a cluster-based scientific visualization tool according to an embodiment of the present invention as described above is described in detail, a basic unit constituting each of the plurality of visualization pipelines 200/1-200/m Is a pipe, and each pipe can express a single visualization technique such as iso-surface, surface extraction, etc., and pipes are arbitrary They can be connected to each other in order to create a pipeline.

Interface pipes (i.e., data patch pipes 100, result collection pipes 300, etc.) are connected to both ends of the plurality of visualization pipelines (200/1-200/m) created as described above, and from other components in the visualization tool. It receives input data and can return visualization results.

The role of the pipe as described above is to receive input data and perform a defined visualization operation, and then transfer the result of the visualization operation to the next pipe.In order to ensure a flexible connection between pipes, all pipes are shown in FIG. It may have a standardized structure as described above.

That is, one pipe may include a pre-processing unit 211, a pipe body 212, a post-processing unit 213, and the like. In the pre-processing unit 211, the pipe is When executed, the pipe is initialized as a pre-processing step, preparing to perform a visualization operation, requesting data (input) from a preceding pipe, and the pipe body 212 can perform visualization on the transmitted data. After the visualization operation is completed, the post-processing unit 213 may finish the work of the pipe and generate result data (output).

Meanwhile, as shown in FIG. 3, each visualization pipe constituting a plurality of visualization pipes 210/1-210/n provided in a plurality of visualization pipelines 200/1-200/m Is a pipe that performs a visualization operation, and the visualization pipe can be divided into a basic visualization pipe (Viz. pipe) and a barrier-visualization pipe (Viz.barrier pipe), and some visualization operations are several blocks to obtain the final result. There are cases in which the results of the results of the or multiple time-steps must be considered at the same time (e.g., point probing operation, etc.), and by reflecting these characteristics, the barrier-visible pipe has more input data to be received from the preceding pipe. After the visualization operation is repeatedly performed until there is no abnormality, it can be post-processed to produce the final result.

Interface pipes may be provided at the front and rear ends of the plurality of visualization pipelines 200/1-200/m as described above, and the plurality of visualization pipelines 200/1-200/m are visualized. A component that composes a library or visualization tool, and the interface pipe is a special pipe dedicated to communication between the pipeline and other components in the visualization tool, and is a data fetch pipe (100) and a result collector. pipe.300).

Here, the data patch pipe 100 receives input data from an external component (ie, another component in the visualization tool), converts it into a form that can be processed by a plurality of visualization pipelines (200/1-200/m), and delivers it. Can be performed, and the result collection pipe 300 can perform a role of collecting the results generated from a plurality of visualization pipelines (200/1-200/m), converting the result into a desired form by an external component, and returning the result. have.

By using these interface pipes, not only can the visualization pipes be implemented and managed independently from external components, but also the advantage of using the existing visualization pipe as it is by implementing only the interface pipe according to the visualization tool to which the pipeline will be used. have.

In addition, as shown in Fig. 2, a pipeline can be created in real time by connecting necessary visualization pipes according to a user's request. The visualization pipes can be connected in reverse order from the last pipe, and each pipe is required for a visualization operation. Information (for example, time-step, targer block, target element, etc.) can be accumulated in the process of connecting the sms pipe and transmitted to the preceding pipe.

In addition, the data patch pipe 100 may be connected to the front (front) of the plurality of visualization pipelines 200/1-200/m, and may be initialized dependently on the first pipe. Data to be read from an external component and a conversion method may be determined based on information required by the visualization pipeline 200/1-200/m and an input data format of the first pipe.

On the other hand, the result collection pipe 300 may be connected to the last (rear end) of the plurality of visualization pipelines 200/1-200/m, which may be initialized dependently on the last visualization pipe, and in the initialization process, the last visualization You can determine how the final result is created by checking the shape of the result generated by the pipe (for example, polygons, points, messages, etc.).

The basic processing unit of the plurality of visualization pipelines 200/1-200/m as described above is a data block, and the operation of the pipeline is the last pipe, while the result collection pipe 300 requests the next data. It can be initiated, and the data request can be propagated to the pipes at the front end, and can be propagated up to the data patch pipe 100.

In such a data patch pipe 100, one data block can be read and transmitted to a connected visualization pipe (Viz. pipe 1 in FIG. 2), and the visualization pipes perform a visualization operation on the input data, and the result is then Can be delivered by pipe.

After that, the result generated in the last visualization pipe (Viz. pipe n in Fig. 2) can be transferred to and collected by the result collection pipe 300, and the above-described process is repeated until there are no more blocks to be processed. When processing of all blocks is finished, the result collection pipe 300 generates a final visualization result, so that the operation of the plurality of visualization pipelines 200/1-200/m may be terminated.

As described above, in the visualization pipeline device for a cluster-based scientific visualization tool according to an embodiment of the present invention, the visualization pipeline can be easily extended to a distributed processing algorithm that utilizes a cluster through a multiple pipeline configuration. Configuration can begin by creating the same sub-pipeline on each compute node in the cluster. Here, the sub-pipeline is Viz. pipe 1-Viz. Like pipe n, it means a pipeline composed of only visualization pipes.

Here, all sub-pipelines share one and the same interface pipe, and the first visualization pipes 210/1 of all sub-pipelines may be connected to one data patch pipe 100, and the last visualization pipes (210/n) may be connected to the same result collection pipe 300. Finally, a cluster-based visualization pipeline as shown in FIG. 2 may be configured.

In addition, the data patch pipe 100 reads and transmits the next data block according to the request of each sub-pipeline, and the result collection pipe 300 requests the next data from the idle sub-pipeline, so that the pseudo round robin ( A round-robin) type of load distribution may be realized, and results of all sub-pipelines may be collected by one result collection pipe 300 to generate a final visualization result.

Accordingly, the present invention includes pipes in a standardized form connected to each other in a random order, includes a special pipe dedicated to communication with other components in the visualization tool, and combines the pipes according to the user's request. By generating a line in real time and performing a visualization operation, it is easy to implement and modify a visualization pipe, and improve portability.

Next, in the visualization pipeline device for the cluster-based scientific visualization tool having the configuration as described above, the input data input from other components in the visualization tool is converted into a form that can be processed and transmitted, and input using a plurality of visualization pipes. It performs visualization operations on data. The visualization process of collecting the results generated through a plurality of visualization pipes, converting them into a desired form in the corresponding component, and returning them will be described.

4 is a flowchart showing a process of visualization using a visualization pipeline device for a cluster-based scientific visualization tool according to another embodiment of the present invention. Since this process has been described in detail with reference to FIGS. 2 and 3, it will be briefly described below.

Referring to FIG. 4, the result collection pipe 300 may request next processed data from the preceding visualization pipe (step 410).

In addition, in the plurality of visualization pipes 210/1-210/n included in the plurality of visualization pipelines 200/1-200/m, the next processing data is requested from the preceding visualization pipe (i.e., the preceding visualization pipe). Each of these may be propagated and a data request may be transmitted to the data patch pipe 100 (step 420).

Here, the data patch pipe 100 is initialized dependently on each of the first visualization pipes of the plurality of visualization pipes 210/1-210/n included in the plurality of visualization pipelines 200/1-200/m. Can be.

In addition, in the data patch pipe 100, input to be input from an external component is based on operation information required for visualization operation in a plurality of visualization pipes 210/1-210/n and an input data format input to the first visualization pipe. You can decide on the data and how it is transformed.

Next, the data patch pipe 100 may convert input data input from other components in the visualization tool into a form that can be processed and transmit it to the plurality of visualization pipelines 200/1-200/m (step 430). That is, the data patch pipe 100 transmits input data to the first visualization pipes of the plurality of visualization pipes 210/1-210/n included in the plurality of visualization pipelines 200/1-200/m. I can.

In addition, in the plurality of visualization pipelines 200/1-200/m, a visualization operation on the input data may be performed using the plurality of visualization pipes 210/1-210/n, respectively (step 440).

The plurality of visualization pipes 210/1 to 210/n are connected in reverse order from the last visualization pipe, and when the visualization pipe is connected, operation information necessary for a visualization operation may be accumulated and transmitted to the preceding visualization pipe.

Subsequently, the result generated through the plurality of visualization pipes 210/1-210/n in the result collection pipe 300 may be collected, converted into a desired form in another component in the visualization tool, and returned (step 450).

The result collection pipe 300 is initialized dependently on each of the last visualization pipes provided in the plurality of visualization pipes 210/1 to 210/n, so that a final result generation method may be determined.

Accordingly, the present invention converts input data input from another component in a visualization tool into a form that can be processed and transfers it, and performs a visualization operation on the input data using a plurality of visualization pipes. By collecting the results generated through a plurality of visualization pipes, converting them into a desired form in the corresponding component, and returning them, it is possible to ensure flexible interconnection using a standardized pipe structure, and can be implemented and managed independently of the visualization tools. .

Meanwhile, FIGS. 5 to 12 are diagrams for explaining visualization results and implementation results using a visualization pipeline device for a cluster-based scientific visualization tool according to an embodiment of the present invention.

5 to 12, the visualization pipeline enables various visualization tasks by combining basic visualization techniques, which means that a more flexible and creative visualization result can be generated. As shown, ① is the visualization of the surface of the wing in the Delta-wing data, and ②, ③ and ④ of Fig. 5 are the results of applying the contour technique, glyph technique, and surface flow technique respectively on the result. Various features above can be visualized and compared.

Finally, ⑤ of FIG. 5 shows an image created by combining all the results of ②, ③ and ④, and the utility will be described later through a real world ship flow analysis case study.

As described above, the visualization pipeline also allows the visualization operation performed in the previous step to be reused without repeating, thereby improving the overall visualization operation speed. As shown in FIG. 5, ②, ③, and ④ are all By receiving the result of ① and applying visualization, the visualization result of ① can be reused and the speed of visualization work can be shortened.

Next, an experiment was conducted by applying a visualization pipeline device for a cluster-based scientific visualization tool according to an embodiment of the present invention to GLOVE and implementing it in a computing cluster having 32 nodes (Table 1 shown in FIG. 6). Hereinafter, an implementation case will be described, the effectiveness and scalability of the visualization pipeline proposed in the present invention will be confirmed through the experimental results, and the advantages of using the visualization pipeline proposed in the present invention will be described through comparison with the existing GLOVE. .

First, to explain the implementation case, the pipeBase class was created to ensure a standard pipe structure, and the basic behavior of the pipes was defined. All pipes are implemented as one class, and the pipeBase class must be inherited. VizPipeBase, which defines basic operations and common routines, was created, and all visualization pipes were inherited to facilitate implementation and management of each visualization pipe.

In addition, the VizBarrier-PipeBase class, which defines the basic operation of the barrier-visible pipes, is also implemented by inheriting the VizPipeBase class, and all barrier-visible pipes are made to inherit VizBarrier-PipeBase.

In addition, the VTK library was used as the visualization engine, and 6 basic visualization pipes (i.e. Clipping, Cutting slice, Extract surface, Iso-surface, Surface display, Surface Flow) and 5 barrier-visible pipes (i.e., Glyph, Glyph over plane, XYZ/IJK line plot, Point probing) was implemented, and each pipe could be implemented simply in approximately 130 to 170 lines, which can confirm the conciseness of the pipeline design according to the embodiment of the present invention.

Meanwhile, a data patch pipe and a result collection pipe are implemented as interface pipes to take charge of communication with the GLOVE framework.The two interface pipes operate on a master node that manages the entire pipeline in the cluster. Slave nodes can have respective lower pipelines, and interface pipes and lower pipelines are implemented to exchange input/output data through MPI (Message Passing Interface), and interfaces that implement actual functions with upper classes. And the relationship of the visualization pipes is as shown in FIG. 7 (pipe implementation class relationship diagram).

For example, the pipeline manager is responsible for the creation and management of the visualization pipeline, and the pipeline manager can interpret user requests and create sub-pipelines in each computational node in real time, and sub-visualization. By connecting the pipeline to the interface pipe of the master node, the cluster-based visualization pipeline can be completed, and the visualization pipeline that has been computed can be destroyed after work is done by the pipeline manager.

Here, the implementation of the interface pipes and the pipeline manager has a dependency on GLOVE, but the visualization pipes can be independently implemented and have a structure that can be easily ported to other visualization frameworks.

Second, when explaining the results and analysis of the implementation case, two types of experiments were largely performed to verify the performance and effectiveness of the visualization pipeline device for a cluster-based scientific visualization tool according to an embodiment of the present invention. Is an experiment to confirm the stability and scalability of the cluster-based visualization pipeline, and was conducted by applying various combinations of visualization operations to Reactor and Ship (KCS) data (Table 2 The information of benchmarks in FIG. 8). .

For example, FIG. 1 is a result of applying a glyph pipe after cutting the reactor data in half with a clippping pipe in order to check the flow inside the reactor (reactor data). Ship (KCS) data) shows the result of applying the iso-surface pipe to check the flow generated by the propeller and applying the clipping pipe to remove the noise generated in the ship part.

As shown from these results, it can be seen that the visualization pipeline of the present invention can successfully perform a single visualization operation and various combinations of visualization operations.In the present invention, the visualization result is increased while increasing the number of operation nodes from 1 to 32. And as a result of checking the processing time, it was confirmed that the processing time continuously decreased as the number of nodes increased, as shown in FIG. 10. That is, it can be seen that the stability and scalability of the visualization pipeline proposed in the present invention have been remarkably improved.

On the other hand, another experiment is a performance comparison between the existing GLOVE and the GLOVE to which the visualization pipeline function according to the embodiment of the present invention is added.In the existing GLOVE, a new visualization is performed on the previously generated visualization result like the visualization pipeline. Since the function to perform the task was not provided directly, the required visualization was performed by outputting the visualized result to a file, reading the file, and then applying the visualization operation.

In this experiment, Deltawing and Rotor data (Table 2 The information of benchmarks in FIG. 8) were used. For Deltawing, the result of performing contour, glyph, surface flow, etc. as shown in FIG. 5 on a surface display pipe is displayed. As for the rotor data, as shown in FIG. 11, the result of performing contour and glyph on the cutting slice pipe and the result of performing glyph on the clipping pipe were measured.

12 shows the visualization operation time taken for the visualization pipeline of the present invention and the modified GLOVE to process the same combination of visualization operations.It can be seen that the visualization pipeline of the present invention exhibits an average of 3.84 times higher performance than the conventional GLOVE. have.

Specifically, in the case of the Clipping+Glyph operation of Rotor data where the size of the intermediate data transmitted through the pipe is relatively large, the performance difference is approximately 7.72 times, which is, by using the pipeline, the intermediate result is stored and read again. As a result obtained because time could be reduced, the advantages of using the visualization pipeline of the present invention can be clearly confirmed.

Meanwhile, referring to FIG. 9, the utility of the visualization pipeline device for a cluster-based scientific visualization tool according to an embodiment of the present invention will be described with reference to FIG. 9, in order to additionally verify the effectiveness of the visualization pipeline of the present invention, actual large-capacity vessel flow analysis data A case study was conducted for (Table 2 The information of benchmarks in Fig. 8), and Ship (KCS) data is multi-block alignment grid data composed of a total of 73 blocks, for a total of 654 time steps, and a total size of 5.6 TB ( It is a huge amount of data with 8.64GB/timestep).

In this case study, we tried to visualize the direction of the flow on the surface of the ship and the vortex of the wake generated by the propeller at the tail of the ship. After visualizing the surface of the ship, as shown in (b) and (c), the flow of the flow on the surface was visualized using the LIC (Line Integral Convolution)-based surface flow and vector glyph. This visualization function is a function that cannot be performed in the conventional GLOVE, which does not support pipelines.

Through this, as shown in Fig. 9(d), it was confirmed that the flow of the flow on the surface of the ship was uniformly flowing from front to back, but the flow near the propeller behind the ship was very unstable.

In addition, as a result of plotting the colors of the glyphs as a helicity, red and blue are mixed and arranged, which means that the flows rotating in opposite directions are mixed with each other.

In order to more clearly observe the flow of the flow in the relevant region, it was attempted to visualize the flow of the flow in the vicinity of the propeller. The iso-surface was extracted. Here, Q-criteria is one of the representative vortex extraction techniques and can extract the eddy current region very effectively.

As a result, it was confirmed that the flow of the eddy current was well captured between the Q value 100-1000 as in (e) of FIG. 9, but there was a problem that the result of extracting the equivalent plane for all data included a lot of noise. In order to improve this, an additional box clip pipe was connected to the iso-surface pipe to remove noise, so that a better quality image could be secured as shown in (f) of FIG. 9.

In addition, in order to more clearly visualize the direction of the flow in the corresponding vortex region, the direction of the glyph in the corresponding region was visualized. First, as a result of glyph visualization for the entire flow region, as shown in (g) of FIG. It was difficult to check the direction of flow due to many glyphs, and in order to solve this problem, a box clip pipe was connected to the glyph pipe around the vortex region identified in (e) of FIG. 9 to visualize it. As a result, (h) of FIG. ), a more organized glyph visualization result was obtained.

However, it is still confirmed that the glyph objects overlap each other and cover each other. In order to solve the problem, a Slice pipe is added to the result as shown in (h) of FIG. 9, and the direction of the glyph for each section can be checked. An improved result image as shown in (i) of FIG. 9 was obtained.

As described above, through the visualization operation using the visualization pipeline of the present invention, which is difficult to perform in the conventional GLOVE without the visualization pipeline function, it was possible to visualize and analyze the vortex characteristics generated by the propeller in the ship flow analysis data.

As described above, in the present invention, the pipeline is composed of visualization pipes representing respective visualization techniques, and the visualization pipes use a standardized pipe structure, thereby ensuring mutually flexible connection.

In addition, in the present invention, by using a visualization tool using a pipeline and a special pipe (ie, an interface pipe) that transmits communication, there is an advantage that the visualization tool can be independently implemented and managed.

In addition, in the present invention, the visualization pipeline can be easily extended to a cluster-based visualization pipeline by arranging sub-pipelines excluding interface pipes in each computational node and connecting them to the interface pipes in the master node.

In addition, in the present invention, a pipeline may be created in real time as a combination of pipes capable of processing a user's visualization request.

In the above description, various embodiments of the present invention have been presented and described, but the present invention is not necessarily limited thereto, and those of ordinary skill in the art to which the present invention pertains, within the scope of the technical spirit of the present invention. It will be easy to see that branch substitutions, modifications and changes are possible.

100: data patch pipe
200/1-200/m: Multiple visualization pipelines
210/1-210/n: Multiple visualization pipes
211: pre-treatment unit
212: pipe body
213: post-processing unit
300: result collection pipe

Claims (12)

A data patch pipe that converts input data input from other components in the visualization tool into a form that can be processed and delivers it;
A plurality of visualization pipelines each performing a visualization operation on the input data using a plurality of visualization pipes,
A result collection pipe that collects the result generated through the plurality of visualization pipes, converts the result into a desired form in the other component, and returns it
A visualization pipeline device for a cluster-based scientific visualization tool comprising a.
The method of claim 1,
Each of the plurality of visualization pipes
A preprocessor for initializing each of the visualization pipes to prepare the visualization operation,
A pipe body that requests data to be processed from a preceding visualization pipe to perform the visualization operation on the data to be processed;
Post-processing unit for generating result data by completing the visualization operation
A visualization pipeline device for a cluster-based scientific visualization tool comprising a.
The method of claim 2,
The plurality of visualization pipes are connected in reverse order from the last visualization pipe, and when the visualization pipe is connected, a visualization pipeline for a cluster-based scientific visualization tool that accumulates computational information necessary for the visualization operation and transfers it to the preceding visualization pipe. Device.
The method of claim 3,
The data patch pipe is a visualization pipeline device for a cluster-based scientific visualization tool that is initialized dependently on each first visualization pipe of the plurality of visualization pipelines.
The method of claim 4,
The data patch pipe is a visualization pipeline device for a cluster-based scientific visualization tool that determines the input data to be input from the other component and a conversion method based on the operation information and an input data format input to the first visualization pipe. .
The method of claim 5,
The result collection pipe is a visualization pipeline device for a cluster-based scientific visualization tool that requests next processed data to a visualization pipe at a front end, is propagated to the visualization pipe at the front end, and is transmitted to the data patch pipe.
The method of claim 6,
The result collection pipe is a visualization pipeline device for a cluster-based scientific visualization tool that is subordinately initialized to each of the last visualization pipes provided in the plurality of visualization pipelines to determine a final result generation method.
The step of requesting the next processing data from the result collection pipe to the preceding visualization pipe, and
Transmitting a data request to the data patch pipe by being propagated to each of the visualization pipes at the front end;
Converting input data input from another component in a visualization tool in the data patch pipe into a form that can be processed and transmitting it,
Performing a visualization operation on the input data using a plurality of visualization pipes, respectively, in a plurality of visualization pipelines; and
Collecting the results generated through the plurality of visualization pipes in the result collection pipe, converting the result into a desired form in the other component, and returning it.
Visualization method of a visualization pipeline device for a cluster-based scientific visualization tool comprising a.
The method of claim 8,
The plurality of visualization pipes are connected in reverse order from the last visualization pipe, and when the visualization pipe is connected, a visualization pipeline for a cluster-based scientific visualization tool that accumulates computational information necessary for the visualization operation and transfers it to the preceding visualization pipe. How to visualize the device.
The method of claim 9,
The visualization method of a visualization pipeline apparatus for a cluster-based scientific visualization tool in which the data patch pipe is initialized dependently on each first visualization pipe of the plurality of visualization pipelines.
The method of claim 10,
The data patch pipe is a visualization pipeline device for a cluster-based scientific visualization tool that determines the input data to be input from the other component and a conversion method based on the operation information and an input data format input to the first visualization pipe. Visualization method.
The method of claim 11,
The result collection pipe is initialized dependently on each of the last visualization pipes provided in the plurality of visualization pipelines, and a visualization method of a visualization pipeline device for a cluster-based scientific visualization tool to determine a final result generation method.

Images