Stratum imaging system, stratum imaging method, equipment and medium

Abstract

The embodiment of the invention provides a formation imaging system, a formation imaging method, formation imaging equipment and a medium. The system performs formation imaging based on a reverse time migration formation imaging algorithm. The system comprises at least one group of nodes and a CXL2.0 memory pool communicating with the at least one group of nodes, the at least one group of nodes are used for inputting calculation process data into the CXL2.0 memory pool in a forward propagation simulation stage of the reverse time migration stratum imaging algorithm; and reading calculation process data from the CXL2.0 memory pool in a backward propagation simulation stage of the reverse time migration stratigraphic imaging algorithm, and the CXL2.0 memory pool is used for dynamically allocating memory to at least one group of nodes. According to the embodiment of the invention, the dynamic requirement of a reverse time migration (RTM) stratigraphic imaging algorithm in the cluster for the memory can be met, and the effective management of the public memory pool in the cluster is realized, so that the consumption of hardware resources is reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of formation imaging, and in particular to a formation imaging system, a formation imaging method, an electronic device and a computer-readable storage medium. Background Art

[0002] In the field of oil and gas exploration, RTM (Reverse Time Migration) technology has become the mainstream deep formation imaging technology for global oil companies. RTM is used to depict the earth's subsurface with high resolution. The algorithm is based on the two-way wave equation and imaging conditions.

[0003] In the process of implementing the algorithm, the storage of the source wave field needs to be constantly exchanged between the main memory and the high-speed memory of the GPU (Graphics Processing Unit) to update the forward wave field and the reverse wave field in space and time and calculate the imaging conditions. In this process, the demand for memory is huge. How to meet the memory demand of the RTM algorithm in the cluster has become one of the current problems that need to be solved urgently. Summary of the invention

[0004] In view of the above problems, a formation imaging system, a formation imaging method, an electronic device and a computer-readable storage medium are proposed to overcome the above problems or at least partially solve the above problems, including:

[0005] A formation imaging system, the system performs formation imaging based on a reverse time migration formation imaging algorithm; the system comprises: at least one group of nodes, and a CXL2.0 memory pool communicating with the at least one group of nodes;

[0006] The at least one group of nodes is used to input the calculation process data into the CXL2.0 memory pool during the forward propagation simulation phase of the reverse time migration formation imaging algorithm; and to read the calculation process data from the CXL2.0 memory pool during the backward propagation simulation phase of the reverse time migration formation imaging algorithm;

[0007] The CXL2.0 memory pool is used to dynamically allocate memory to the at least one group of nodes.

[0008] Optionally, for any set of nodes, including master nodes and slave nodes;

[0009] The master node is used to reserve CXL memory pool resources in the CXL2.0 memory pool; when the reservation is successful, the master node and the slave node enter the computing process data into the CXL2.0 memory pool.

[0010] Optionally, any node includes a graphics processing unit and a central processing unit, the graphics processing unit is reserved with at least one first data slot, and the central processing unit is reserved with at least one second data slot;

[0011] The master node is used to transfer the calculation process data from the first data slot to the second data slot according to a preset time step in the forward propagation simulation phase; and, in the backward propagation simulation phase, when the first data slot is idle, push the calculation process data in the second data slot from the second data slot into the first data slot.

[0012] Optionally, the at least one group of nodes is further configured to destroy the corresponding computing process data in the CXL2.0 memory pool after reading the computing process data from the CXL2.0 memory pool.

[0013] Optionally, the at least one group of nodes includes at least a first node group and a second node group;

[0014] The first node group and the second node group interleave calls to the CXL2.0 memory pool.

[0015] Optionally, the first node group performs the step of recording the computing process data into the CXL2.0 memory pool between the first time point and the second time point; and performs the step of reading the computing process data from the CXL2.0 memory pool between the second time point and the third time point;

[0016] The second node group, between the second time point and the third time point, executes the step of recording the computing process data into the CXL2.0 memory pool.

[0017] An embodiment of the present invention further provides a formation imaging method, which is applied to the system as described above, wherein the system comprises: at least one group of nodes, and a CXL2.0 memory pool communicating with the at least one group of nodes; the method comprises:

[0018] In the forward propagation simulation stage of the reverse time migration formation imaging algorithm, the first calculation process data is recorded into the CXL2.0 memory pool;

[0019] In the back propagation simulation stage of the reverse time migration formation imaging algorithm, the first calculation process data is read from the CXL2.0 memory pool, and a target formation image is generated according to the first calculation process data.

[0020] Optionally, the at least one group of nodes includes at least a first node group and a second node group; and in the forward propagation simulation phase of the reverse time migration formation imaging algorithm, recording the first calculation process data into the CXL2.0 memory pool includes:

[0021] Between the first time point and the second time point, recording the first computing process data into the CXL2.0 memory pool;

[0022] In the back propagation simulation phase of the reverse time migration formation imaging algorithm, reading the first calculation process data from the CXL2.0 memory pool includes:

[0023] Between the second time point and the third time point, reading the first computing process data from the CXL2.0 memory pool;

[0024] The second node group is used to enter the second computing process data into the CXL2.0 memory pool between the second time point and the third time point.

[0025] An embodiment of the present invention further provides an electronic device, comprising a processor, a memory, and a computer program stored in the memory and capable of running on the processor, wherein the computer program implements the above formation imaging method when executed by the processor.

[0026] An embodiment of the present invention further provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the above-mentioned formation imaging method is implemented.

[0027] The embodiments of the present invention have the following advantages:

[0028] In an embodiment of the present invention, the system performs formation imaging based on a reverse time migration formation imaging algorithm; the system includes: at least one group of nodes, and a CXL2.0 memory pool communicating with the at least one group of nodes; the at least one group of nodes is used to enter calculation process data into the CXL2.0 memory pool during the forward propagation simulation phase of the reverse time migration formation imaging algorithm; and, during the backward propagation simulation phase of the reverse time migration formation imaging algorithm, read the calculation process data from the CXL2.0 memory pool; the CXL2.0 memory pool is used to dynamically allocate memory to the at least one group of nodes.

[0029] Through the embodiments of the present invention, the dynamic demand for memory of the reverse time migration (RTM) algorithm in the cluster can be met, and the effective management of the public memory pool in the cluster is achieved, thereby reducing the consumption of hardware resources. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] In order to more clearly illustrate the technical solution of the present invention, the accompanying drawings required for use in the description of the present invention will be briefly introduced below. Obviously, the accompanying drawings in the following description are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative labor.

[0031] Figure 1a It is a schematic diagram of a data scheduling process in the prior art;

[0032] Figure 1b It is a flow chart of steps of a data scheduling process in the prior art;

[0033] Figure 2 is a structural schematic diagram of a formation imaging system according to an embodiment of the present invention;

[0034] Figure 3 is a flow chart of steps of a formation imaging method according to an embodiment of the present invention;

[0035] Figure 4 is a flowchart of another method for formation imaging according to an embodiment of the present invention;

[0036] Figure 5 is a schematic diagram of a data scheduling process according to an embodiment of the present invention;

[0037] Figure 6 is a schematic diagram of another data scheduling process according to an embodiment of the present invention;

[0038] Figure 7 is a schematic diagram of a staggered scheduling according to an embodiment of the present invention;

[0039] Figure 8 It is a flow chart of the steps of a data scheduling process according to an embodiment of the present invention. DETAILED DESCRIPTION

[0040] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0041] RTM is considered a time-consuming and data-intensive technique. The computational complexity of solving the governing wave equations for RTM using numerical methods is currently quite high, and when discretized using the finite difference method, more accurate simulations also directly lead to higher computational complexity.

[0042] Fortunately, RTM-based imaging algorithms are embarrassingly parallel, and the simulation of each seismic cannon is completely independent, usually only requiring the coordination of multiple GPU servers to calculate one seismic cannon.

[0043] Specifically, Figure 1aAs shown, the user (Users 1) can use the HPC cabinet (HPC cabinet) 1 in the HPC (High-Performance Computing) cluster (HPC cluster) 1 through the front end (Frontal 1) to complete the task of formation imaging. The HPC cabinet 1 can include multiple nodes (Nodes).

[0044] However, there are several problems to be solved in the existing technical solutions:

[0045] 1. GPU computing speed is much faster than CPU (Central Processing Unit), and node memory is limited by procurement cost and host slot restrictions, so it cannot be expanded in the same proportion.

[0046] 2. When the RTM algorithm calculates the forward propagation wave field, the memory requirement increases linearly, reaches a peak before the reverse propagation begins, and then decreases linearly as the reverse propagation proceeds. The actual limit of the algorithm is actually the maximum memory value of the node.

[0047] For the first problem, the common technical solution is to use lossy compression algorithms (such as ZFP, WPC, etc.) on the source wave field. Such algorithms can achieve a compression ratio of nearly 10 times, which can greatly alleviate storage requirements. For the specific algorithm structure, see Figure 1b :

[0048] First, the RTM (reverse time migration) shot record may be initialized; specifically, a reverse time migration (RTM) calculation task may be started first, and related data and parameters may be initialized.

[0049] The data can then be moved to the GPU; specifically, the data required for the computation can be transferred from the host memory (CPU) to the GPU memory for computation on the GPU.

[0050] Next, the forward wave field may be propagated; specifically, the propagation calculation of the forward wave field may be performed on the GPU to simulate the propagation process of the seismic wave in the underground medium.

[0051] When propagating the forward wavefield, the forward wavefield may be compressed; specifically, the calculated forward wavefield data may be compressed to reduce storage space and transmission time.

[0052] Then, the forward wavefield may be moved to the host; specifically, the compressed forward wavefield data may be transferred from the GPU memory back to the host memory.

[0053] After completing a seismic cannon, it can be detected whether the time reaches Tmax from Tmin; if not, the step of propagating the forward wavefield is re-executed; if Tmax is reached, the forward wavefield can be moved to the GPU; specifically, the forward wavefield data can be transferred to the GPU.

[0054] Decompress the forward wave field; specifically, the forward wave field data transmitted to the GPU may be decompressed to restore the original data format.

[0055] Propagate reverse wavefield: Perform reverse wavefield propagation calculations on the GPU to simulate the process of seismic waves returning from the receiver to the source.

[0056] Calculate imaging conditions; Based on the propagation results of the forward and reverse wave fields, calculate imaging conditions to generate images of underground structures.

[0057] Next, it can be detected whether the time reaches Tmin from Tmax; if not, the step of moving the forward wave field to the host is re-executed; if Tmin is reached, the host image can be updated and the result can be saved; specifically, the calculated imaging result can be updated to the image data in the host memory, and the final result can be saved to the storage device.

[0058] Even so, since the RTM algorithm needs to store the forward propagation wave field, this solution still cannot fundamentally reduce the RTM algorithm's dependence on the total node memory. In fact, the node memory size directly determines the number of GPU servers required for a single shot, which indirectly affects the parallel efficiency.

[0059] In order to completely solve the above problems, the embodiment of the present invention successfully meets the dynamic memory requirements of the reverse time migration formation imaging (RTM) algorithm within the cluster by introducing memory pooling technology, and realizes effective management of the public memory pool within the cluster, thereby reducing the consumption of hardware resources.

[0060] Specifically, based on the above theory, an embodiment of the present invention provides a formation imaging system; Figure 2 The formation imaging system 10 performs formation imaging based on a reverse time migration formation imaging algorithm; the formation imaging system 10 may include:

[0061] At least one group of nodes 210 , and a CXL 2.0 memory pool 220 in communication with the at least one group of nodes 210 .

[0062] At least one set of nodes 210 for recording computational process data into a CXL2.0 memory pool 220 during a forward propagation simulation phase of a reverse time migration formation imaging algorithm; and for reading computational process data from the CXL2.0 memory pool 220 during a backward propagation simulation phase of a reverse time migration formation imaging algorithm;

[0063] The CXL2.0 memory pool 220 is used to dynamically allocate memory to at least one group of nodes 210 .

[0064] In practical applications, reverse time migration (RTM) is an advanced imaging technology used to generate high-resolution underground structure images. RTM uses seismic data to reconstruct images of underground structures by simulating the propagation of seismic waves in underground media. The following are the basic steps and principles of the reverse time migration stratigraphic imaging algorithm.

[0065] The basic principle of reverse time migration is to use seismic data to reconstruct the image of underground structure by simulating the forward and backward propagation of seismic waves. The specific steps are as follows:

[0066] Forward propagation simulation: simulates the process of seismic waves propagating from the seismic source in the underground medium.

[0067] Data acquisition: Recording the data received by the detector during the propagation of seismic waves underground.

[0068] Backpropagation simulation: Using the received seismic data, simulate the process of seismic waves propagating backward from the detector location back into the underground.

[0069] Imaging: Generate images of subsurface structures by cross-correlating the forward and backward propagated wave fields.

[0070] The steps of reverse time migration include data preparation, forward propagation simulation, backward propagation simulation and imaging; among them:

[0071] 1. Data preparation includes:

[0072] Seismic data: includes the excitation signal of the seismic cannon and the seismic wave data received by the detector.

[0073] Velocity model: A model that describes the velocity distribution of underground media.

[0074] Other parameters: such as time step, spatial grid size, etc.

[0075] 2. Forward propagation simulation includes:

[0076] In the forward propagation simulation stage, the algorithm simulates the process of seismic waves propagating from the seismic cannon in the underground medium.

[0077] Stimulating seismic waves: Seismic cannons stimulate seismic waves, which propagate underground.

[0078] Wave propagation: Seismic waves are reflected, refracted, and scattered when they encounter interfaces between different media underground.

[0079] Data acquisition: The reflected waves are received by seismic detectors on the surface or in the well, which record the arrival time and amplitude of the seismic waves.

[0080] 3. Back propagation simulation includes:

[0081] In the back-propagation simulation phase, the algorithm simulates the process of seismic waves propagating back into the underground from the received seismic data.

[0082] Reverse propagation: Using the received seismic data, simulate the reverse propagation of seismic waves from the detector location back into the underground.

[0083] Wavefield reconstruction: Reconstruct the wavefield of underground structures through reverse propagation.

[0084] 4. Imaging includes:

[0085] During the imaging phase, the algorithm cross-correlates the forward- and backward-propagated wave fields to generate an image of the subsurface structure.

[0086] Cross-correlation: The forward-propagated and backward-propagated wave fields are cross-correlated to generate an image of the subsurface structure.

[0087] Image enhancement: Enhance and filter the generated images to improve the resolution and clarity of the images.

[0088] The formation imaging system 20 may include at least one group of nodes 210 and a CXL (Compute Express Link) 2.0 memory pool communicating with the at least one group of nodes 210 .

[0089] in:

[0090] In at least one group of nodes 210 , any group of nodes 210 may include multiple nodes 210 , and the multiple nodes 210 may communicate with the CXL2.0 memory pool 220 respectively.

[0091] As a key new technology in the development of information technology, CXL is becoming more and more mature. By supporting high-performance, low-latency connections, CXL enables flexible access to components such as processors, accelerators, and memory, improves overall system performance, and has advanced error handling and reliability functions.

[0092] In terms of standard formulation and promotion, it has received widespread support from many technology companies and has gradually become the mainstream unified standard in the industry.

[0093] In addition, the continuous improvement of CXL at the software level, including the improvement of drivers, tools and libraries, provides developers with a more convenient way to take advantage of its performance advantages and promotes the development of more efficient applications and solutions.

[0094] The CXL 2.0 standard provides a memory pooling capability (memory pool), which binds multiple CXL memory blocks on a server platform together to form a shared memory pool, which can provide significant cost-effectiveness and resource utilization for each group of nodes 210 .

[0095] In practical applications, each group of nodes 210 in at least one group of nodes 210 may enter the calculation process data into the CXL2.0 memory pool 220 in the forward propagation simulation phase of the reverse time migration formation imaging algorithm; wherein the calculation process data may refer to the data of the intermediate process generated in the forward propagation simulation phase of the reverse time migration formation imaging algorithm, such as: forward wave field, reverse wave field, etc., which is not limited in the embodiment of the present invention.

[0096] On the other hand, each group of nodes 210 can also read previously entered calculation process data from the CXL2.0 memory pool 220 for data processing during the back propagation simulation phase of the reverse time migration formation imaging algorithm. The embodiment of the present invention does not limit this.

[0097] In practical applications, the CXL2.0 memory pool 220 can be used to dynamically allocate memory to at least one group of nodes 210. The CXL2.0 memory pool 220 can meet the memory requirements of the reverse time migration stratigraphic imaging algorithm of each node 210 and meet the effective management of the shared memory pool within the cluster, thereby reducing the consumption of hardware resources by the reverse time migration stratigraphic imaging algorithm.

[0098] In an embodiment of the present invention, the system performs formation imaging based on a reverse time migration formation imaging algorithm; the system includes: at least one group of nodes 210, and a CXL2.0 memory pool 220 communicating with the at least one group of nodes 210; the at least one group of nodes 210 is used to enter calculation process data into the CXL2.0 memory pool 220 during the forward propagation simulation phase of the reverse time migration formation imaging algorithm; and, during the backward propagation simulation phase of the reverse time migration formation imaging algorithm, read the calculation process data from the CXL2.0 memory pool 220; the CXL2.0 memory pool 220 is used to dynamically allocate memory to the at least one group of nodes 210.

[0099] Through the embodiments of the present invention, the dynamic demand for memory of the reverse time migration (RTM) algorithm in the cluster can be met, and the effective management of the public memory pool in the cluster is achieved, thereby reducing the consumption of hardware resources.

[0100] In one embodiment of the present invention, for any group of nodes 210, including a master node and a slave node;

[0101] The master node is used to reserve CXL memory pool resources in the CXL2.0 memory pool 220 ; when the reservation is successful, the master node and the slave node enter the computing process data into the CXL2.0 memory pool 220 .

[0102] In some feasible embodiments, any group of nodes 210 may include a master node and other nodes 210 (hereinafter referred to as slave nodes) except the master node.

[0103] For a master node of any group of nodes 210 , the master node can communicate with master nodes of other groups and initialize the CXL2.0 memory pool 220 .

[0104] When preparing to process the next seismic cannon, the forward propagation simulation can be initialized first; during the forward propagation simulation stage, the master node can reserve resources first.

[0105] Specifically, the master node may first reserve CXL memory pool resources in CXL 2.0 memory pool 220. If the reservation fails, the master node continues to wait.

[0106] If the reservation is successful, the master node and the slave nodes of the same group to which the master node belongs may record the computing process data into the CXL2.0 memory pool 220 .

[0107] In one embodiment of the present invention, any node 210 includes a graphics processing unit and a central processing unit, the graphics processing unit is reserved with at least one first data slot, and the central processing unit is reserved with at least one second data slot;

[0108] The master node is used to transfer the calculation process data from the first data slot to the second data slot according to the preset time step in the forward propagation simulation stage; and, in the backward propagation simulation stage, when the first data slot is idle, push the calculation process data in the second data slot from the second data slot to the first data slot.

[0109] In some feasible embodiments, for any node 210, the node 210 may include a graphics processing unit and a central processing unit; wherein:

[0110] At least one source wave source data slot, ie, a first data slot, may be reserved in the graphics processing unit; and at least one source wave source data slot, ie, a second data slot, may be reserved in the central processing unit.

[0111] For a master node of any group of nodes 210 , during the forward propagation simulation phase, the calculation process data can be transmitted from the first data slot of the graphics processing unit to the second data slot of the central processing unit according to a preset time step.

[0112] In addition, for the master node, during the backward propagation simulation phase, when the first data slot of the graphics processing unit is idle, the calculation process data in the second data slot of the central processing unit can be pushed from the second data slot of the graphics processing unit into the first data slot. This embodiment of the present invention does not impose any restrictions on this.

[0113] Exemplarily, for a slave node, the above steps may also be performed; specifically:

[0114] The slave node can transmit the calculation process data from the first data slot of the graphics processing unit to the second data slot of the central processing unit according to the preset time step during the forward propagation simulation stage.

[0115] In addition, for the slave node, during the backward propagation simulation phase, when the first data slot of the graphics processing unit is idle, the calculation process data in the second data slot of the central processing unit can be pushed from the second data slot of the graphics processing unit into the first data slot. This embodiment of the present invention does not impose any restrictions on this.

[0116] In one embodiment of the present invention, at least one group of nodes 210 is further configured to destroy corresponding computing process data in CXL2.0 memory pool 220 after reading the computing process data from CXL2.0 memory pool 220 .

[0117] In some feasible embodiments, for any group of nodes 210, after reading the computing process data from the CXL2.0 memory pool 220, the corresponding computing process data in the CXL2.0 memory pool 220 may be destroyed so that other groups of nodes 210 can enter data.

[0118] For example, for the master node, after reading the computing process data from the CXL2.0 memory pool 220 , the corresponding computing process data in the CXL2.0 memory pool 220 may be destroyed.

[0119] For the slave node, after reading the computing process data from the CXL2.0 memory pool 220 , the corresponding computing process data in the CXL2.0 memory pool 220 may also be destroyed, and this embodiment of the present invention does not limit this.

[0120] In one embodiment of the present invention, at least one group of nodes 210 includes at least a first node group and a second node group;

[0121] The first node group and the second node group interleave calls to CXL 2.0 memory pool 220 .

[0122] In some feasible embodiments, when the total amount in the CXL2.0 memory pool 220 is much larger than the total memory required for single-shot calculations of all node 210 groups, the RTM single-shot resource (node ​​210 group) scheduling strategy returns to normal mode, that is, it is called when there is demand.

[0123] When the total amount in the CXL2.0 memory pool 220 is less than the total memory required for single-shot calculations of all node groups 210, it is necessary to control the call of resources; specifically, since the forward propagation simulation stage and the backward propagation simulation stage have completely different requirements for memory, the embodiment of the present invention can control the first node group and the second node group to call the CXL2.0 memory pool 220 in an interlaced manner.

[0124] For example, when the first node group is performing the forward propagation simulation phase and recording the computational process data into the CXL2.0 memory pool 220, the second node group can be performing the backward propagation simulation phase to read data from the CXL2.0 memory pool 220; conversely, when the first node group is performing the backward propagation simulation phase and reading data from the CXL2.0 memory pool 220, the second node group can be performing the forward propagation simulation phase to record the computational process data into the CXL2.0 memory pool 220. By interleaving, the utilization of resources can be maximized.

[0125] In one embodiment of the present invention, the first node group performs the step of recording the computing process data into the CXL2.0 memory pool 220 between the first time point and the second time point; and performs the step of reading the computing process data from the CXL2.0 memory pool 220 between the second time point and the third time point;

[0126] The second node group, between the second time point and the third time point, executes the step of recording the computing process data into the CXL2.0 memory pool 220 .

[0127] In some feasible embodiments, the step of interleaving the calling of CXL2.0 memory pool 220 may be implemented as follows:

[0128] Specifically, each node 210 in the first node group can execute the step of entering the computing process data into the CXL2.0 memory pool 220 between the first time point and the second time point; and each node 210 in the first node group can execute the step of reading the computing process data from the CXL2.0 memory pool 220 between the second time point and the third time point.

[0129] In addition, the second node group performs the step of recording the computing process data into CXL2.0 memory pool 220 between the second time point and the third time point.

[0130] The order of the first time point, the second time point and the third time point may be: first time point→second time point→third time point, which is not limited in the embodiment of the present invention.

[0131] Based on the formation imaging system mentioned in the above embodiment, the embodiment of the present invention also provides a formation imaging method, which can be referred to Figure 3 , Figure 3 A flow chart showing the steps of a formation imaging method according to an embodiment of the present invention is shown.

[0132] like Figure 3 As shown, the formation imaging method may include the following steps:

[0133] Step 301: In the forward propagation simulation phase of the reverse time migration formation imaging algorithm, the first calculation process data is recorded into the CXL2.0 memory pool.

[0134] In practical applications, any one of the at least one group of nodes may enter the first calculation process data into the CXL2.0 memory pool during the forward propagation simulation phase of the reverse time migration formation imaging algorithm.

[0135] Step 302: In the back propagation simulation phase of the reverse time migration formation imaging algorithm, first calculation process data is read from the CXL2.0 memory pool, and a target formation image is generated according to the first calculation process data.

[0136] On the other hand, the group of nodes can also read the first calculation process data previously entered from the CXL2.0 memory pool for data processing during the back propagation simulation phase of the reverse time migration formation imaging algorithm. The embodiment of the present invention does not limit this.

[0137] Exemplarily, after obtaining the first computing process data, the nodes in the group of nodes may generate a formation image according to the first computing process data, thereby obtaining a target formation image.

[0138] In an embodiment of the present invention, in the forward propagation simulation stage of the reverse time migration formation imaging algorithm, the first calculation process data is entered into the CXL2.0 memory pool; in the backward propagation simulation stage of the reverse time migration formation imaging algorithm, the first calculation process data is read from the CXL2.0 memory pool, and the target formation image is generated according to the first calculation process data.

[0139] Through the embodiments of the present invention, the dynamic demand for memory of the reverse time migration (RTM) algorithm in the cluster can be met, and the effective management of the public memory pool in the cluster is achieved, thereby reducing the consumption of hardware resources.

[0140] Reference Figure 4, shows a flow chart of another formation imaging method according to an embodiment of the present invention, which may include the following steps:

[0141] Step 401: at least one group of nodes includes at least a first node group and a second node group; between a first time point and a second time point, first computing process data is entered into a CXL2.0 memory pool.

[0142] In some feasible embodiments, the step of interleaving the calling of the CXL2.0 memory pool may be implemented as follows:

[0143] Specifically, each node in the first node group may execute the step of recording the computing process data into the CXL2.0 memory pool between the first time point and the second time point.

[0144] Step 402: read the first computing process data from the CXL2.0 memory pool between the second time point and the third time point; the second node group is used to enter the second computing process data into the CXL2.0 memory pool between the second time point and the third time point.

[0145] Between the second time point and the third time point, each node in the first node group may execute the step of reading the computing process data from the CXL2.0 memory pool.

[0146] In addition, the second node group, between the second time point and the third time point, executes the step of recording the computing process data into the CXL2.0 memory pool.

[0147] The order of the first time point, the second time point and the third time point may be: first time point→second time point→third time point, which is not limited in the embodiment of the present invention.

[0148] For example, Figure 5 As shown, users (Users2) can use HPC cabinet2 in HPCcluster2 to complete the formation imaging task through the front end (Frontal2). HPC cabinet2 can include multiple nodes (Nodes). These multiple nodes can interact with the nodes (CXL2.0 nodes) of the CXL2.0 memory pool to use the memory resources of the CXL2.0 memory pool.

[0149] like Figure 6As shown, for each group of nodes, a read-write control valve (control) can be set; for each group of seismic cannons, the GPU memory (GPU Memory) can exchange data with the CPU memory (CPU Memory), and the CPU Memory can be controlled by the control to interact with the CXL2.0 memory pool (CXL Memory Pool). Controls of different groups can communicate data through MPI (Message Passing Interface) to control the resource scheduling of the CXL2.0 memory pool.

[0150] For different groups of nodes, the resources of the CXL2.0 memory pool (CXL2.0nodes) can be called alternately; specifically, Figure 7 As shown, the forward propagation calculation and the back propagation calculation of multiple nodes can be staggered in the time dimension.

[0151] like Figure 8 As shown, first, the RTM (reverse time migration) shot record may be initialized; specifically, a reverse time migration (RTM) calculation task may be started first, and related data and parameters may be initialized.

[0152] The data can then be moved to the GPU; specifically, the data required for the computation can be transferred from the host memory (CPU) to the GPU memory for computation on the GPU.

[0153] Next, the forward wave field may be propagated; specifically, the propagation calculation of the forward wave field may be performed on the GPU to simulate the propagation process of the seismic wave in the underground medium.

[0154] When propagating the forward wavefield, the forward wavefield may be compressed; specifically, the calculated forward wavefield data may be compressed to reduce storage space and transmission time.

[0155] Then, the forward wavefield can be moved to the host; specifically, the compressed forward wavefield data can be transferred from the GPU memory back to the host memory and recorded into the CXL2.0 memory pool.

[0156] After completing a seismic cannon, it is possible to detect whether the time has reached Tmax from Tmin; if not, the step of propagating the forward wavefield is re-executed; if Tmax is reached, the forward wavefield can be moved to the GPU; specifically, the forward wavefield data can be read from the CXL2.0 memory pool first, and then transferred to the GPU.

[0157] Decompress the forward wave field; specifically, the forward wave field data transmitted to the GPU may be decompressed to restore the original data format.

[0158] Propagate reverse wavefield: Perform reverse wavefield propagation calculations on the GPU to simulate the process of seismic waves returning from the receiver to the source.

[0159] Calculate imaging conditions; Based on the propagation results of the forward and reverse wave fields, calculate imaging conditions to generate images of underground structures.

[0160] Next, it can be detected whether the time reaches Tmin from Tmax; if not, the step of moving the forward wave field to the host is re-executed; if Tmin is reached, the host image can be updated and the result can be saved; specifically, the calculated imaging result can be updated to the image data in the host memory, and the final result can be saved to the storage device.

[0161] In the embodiment of the present invention, at least one group of nodes includes at least a first node group and a second node group; between a first time point and a second time point, the first computing process data is recorded into the CXL2.0 memory pool; between a second time point and a third time point, the first computing process data is read from the CXL2.0 memory pool; and the second node group is used to record the second computing process data into the CXL2.0 memory pool between the second time point and the third time point. Through the embodiment of the present invention, resource utilization can be improved. And when the total amount of memory is the same, more complex and more accurate simulations can be processed.

[0162] It should be noted that, for the sake of simplicity, the method embodiments are described as a series of action combinations, but those skilled in the art should be aware that the embodiments of the present invention are not limited by the order of the actions described, because according to the embodiments of the present invention, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions involved are not necessarily required by the embodiments of the present invention.

[0163] An embodiment of the present invention further provides an electronic device, including a processor, a memory, and a computer program stored in the memory and capable of running on the processor, and when the computer program is executed by the processor, the above formation imaging method is implemented.

[0164] An embodiment of the present invention further provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the above-mentioned formation imaging method is implemented.

[0165] As for the device embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0166] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.

[0167] Those skilled in the art will appreciate that the embodiments of the present invention may be provided as methods, devices, or computer program products. Therefore, the embodiments of the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the embodiments of the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.

[0168] The embodiments of the present invention are described with reference to the flowcharts and / or block diagrams of the methods, terminal devices (systems), and computer program products according to the embodiments of the present invention. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing terminal device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing terminal device generate a device for implementing the functions specified in one process or multiple processes in the flowchart and / or one box or multiple boxes in the block diagram.

[0169] These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing terminal device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce a manufactured product including an instruction device that implements the functions specified in one or more processes in the flowchart and / or one or more boxes in the block diagram.

[0170] These computer program instructions can also be loaded onto a computer or other programmable data processing terminal device so that a series of operating steps are executed on the computer or other programmable terminal device to produce computer-implemented processing, so that the instructions executed on the computer or other programmable terminal device provide steps for implementing the functions specified in one or more processes in the flowchart and / or one or more boxes in the block diagram.

[0171] Although the preferred embodiments of the present invention have been described, those skilled in the art may make additional changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the embodiments of the present invention.

[0172] Finally, it should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or terminal device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or terminal device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the existence of other identical elements in the process, method, article or terminal device including the elements.

[0173] The above provides a detailed introduction to a formation imaging system, a formation imaging method, an electronic device and a computer-readable storage medium. Specific examples are used herein to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core idea. At the same time, for a person skilled in the art, according to the idea of ​​the present invention, there will be changes in the specific implementation method and application scope. In summary, the content of this specification should not be understood as a limitation on the present invention.

Claims

1. A formation imaging system, characterized in that: The system performs formation imaging based on a reverse time migration formation imaging algorithm; the system comprises: at least one group of nodes, and a CXL2.0 memory pool communicating with the at least one group of nodes; The at least one group of nodes is used to input the calculation process data into the CXL2.0 memory pool during the forward propagation simulation phase of the reverse time migration formation imaging algorithm; and to read the calculation process data from the CXL2.0 memory pool during the backward propagation simulation phase of the reverse time migration formation imaging algorithm; The CXL2.0 memory pool is used to dynamically allocate memory to the at least one group of nodes.

2. The system according to claim 1, characterized in that For any set of nodes, including master nodes and slave nodes; The master node is used to reserve CXL memory pool resources in the CXL2.0 memory pool; when the reservation is successful, the master node and the slave node enter the computing process data into the CXL2.0 memory pool.

3. The system according to claim 2, characterized in that Any node includes a graphics processing unit and a central processing unit, the graphics processing unit is reserved with at least one first data slot, and the central processing unit is reserved with at least one second data slot; The master node is used to transfer the calculation process data from the first data slot to the second data slot according to a preset time step in the forward propagation simulation phase; and, in the backward propagation simulation phase, when the first data slot is idle, push the calculation process data in the second data slot from the second data slot into the first data slot.

4. The system according to claim 1, characterized in that The at least one group of nodes is further configured to destroy the corresponding computing process data in the CXL2.0 memory pool after reading the computing process data from the CXL2.0 memory pool.

5. The system according to claim 1, characterized in that The at least one group of nodes includes at least a first node group and a second node group; The first node group and the second node group interleave calls to the CXL2.0 memory pool.

6. The system according to claim 5, characterized in that The first node group, between the first time point and the second time point, executes the step of recording the computing process data into the CXL2.0 memory pool; between the second time point and the third time point, executes the step of reading the computing process data from the CXL2.0 memory pool; The second node group, between the second time point and the third time point, executes the step of recording the computing process data into the CXL2.0 memory pool.

7. A method for formation imaging, characterized in that: Applied to the system according to any one of claims 1 to 6, the system comprising: at least one group of nodes, and a CXL2.0 memory pool communicating with the at least one group of nodes; the method comprising: In the forward propagation simulation stage of the reverse time migration formation imaging algorithm, the first calculation process data is recorded into the CXL2.0 memory pool; In the back propagation simulation stage of the reverse time migration formation imaging algorithm, the first calculation process data is read from the CXL2.0 memory pool, and a target formation image is generated according to the first calculation process data.

8. The method according to claim 7, characterized in that The at least one group of nodes includes at least a first node group and a second node group; in the forward propagation simulation phase of the reverse time migration formation imaging algorithm, recording the first calculation process data into the CXL2.0 memory pool includes: Between the first time point and the second time point, recording the first computing process data into the CXL2.0 memory pool; In the back propagation simulation phase of the reverse time migration formation imaging algorithm, reading the first calculation process data from the CXL2.0 memory pool includes: Between the second time point and the third time point, reading the first computing process data from the CXL2.0 memory pool; The second node group is used to enter the second computing process data into the CXL2.0 memory pool between the second time point and the third time point.

9. An electronic device, characterized in that: The method comprises a processor, a memory, and a computer program stored in the memory and capable of running on the processor, wherein when the computer program is executed by the processor, the method for formation imaging according to any one of claims 7 to 8 is implemented.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method for formation imaging according to any one of claims 7 to 6 is implemented.

Images