CN112988114A - GPU-based large number computing system - Google Patents
GPU-based large number computing system Download PDFInfo
- Publication number
- CN112988114A CN112988114A CN202110270744.5A CN202110270744A CN112988114A CN 112988114 A CN112988114 A CN 112988114A CN 202110270744 A CN202110270744 A CN 202110270744A CN 112988114 A CN112988114 A CN 112988114A
- Authority
- CN
- China
- Prior art keywords
- gpu
- calculation
- software
- majority
- computing
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000004364 calculation method Methods 0.000 claims abstract description 129
- 239000011159 matrix material Substances 0.000 claims abstract description 43
- 238000004891 communication Methods 0.000 claims description 23
- 238000012545 processing Methods 0.000 claims description 22
- 238000000034 method Methods 0.000 claims description 9
- 230000008569 process Effects 0.000 claims description 8
- 238000005516 engineering process Methods 0.000 abstract description 3
- 238000007726 management method Methods 0.000 description 9
- 238000010586 diagram Methods 0.000 description 5
- 238000011161 development Methods 0.000 description 3
- 238000007667 floating Methods 0.000 description 3
- 230000006870 function Effects 0.000 description 3
- 238000004088 simulation Methods 0.000 description 3
- HPTJABJPZMULFH-UHFFFAOYSA-N 12-[(Cyclohexylcarbamoyl)amino]dodecanoic acid Chemical compound OC(=O)CCCCCCCCCCCNC(=O)NC1CCCCC1 HPTJABJPZMULFH-UHFFFAOYSA-N 0.000 description 2
- 238000006467 substitution reaction Methods 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000007796 conventional method Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000007620 mathematical function Methods 0.000 description 1
- 230000008520 organization Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000004148 unit process Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F7/00—Methods or arrangements for processing data by operating upon the order or content of the data handled
- G06F7/38—Methods or arrangements for performing computations using exclusively denominational number representation, e.g. using binary, ternary, decimal representation
- G06F7/48—Methods or arrangements for performing computations using exclusively denominational number representation, e.g. using binary, ternary, decimal representation using non-contact-making devices, e.g. tube, solid state device; using unspecified devices
- G06F7/57—Arithmetic logic units [ALU], i.e. arrangements or devices for performing two or more of the operations covered by groups G06F7/483 – G06F7/556 or for performing logical operations
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/16—Matrix or vector computation, e.g. matrix-matrix or matrix-vector multiplication, matrix factorization
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Pure & Applied Mathematics (AREA)
- Mathematical Analysis (AREA)
- Mathematical Optimization (AREA)
- Mathematical Physics (AREA)
- Computational Mathematics (AREA)
- Computing Systems (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- Algebra (AREA)
- Databases & Information Systems (AREA)
- Software Systems (AREA)
- Hardware Redundancy (AREA)
Abstract
The invention relates to the field of scientific calculation, in particular to a GPU-based large number calculation system, and aims to solve the problems of large time consumption and high use technology threshold of large number calculation. The GPU-based majority computing system of the invention comprises: the system comprises a large-number computing hardware system and a large-number computing software system, wherein the large-number computing software system runs on the large-number computing hardware system. Wherein, the majority calculation software system includes: client software and server software; the client software is used for sending calculation request information to the server software; and the server software is used for performing majority calculation and returning operation information to the client software according to the calculation request information. The server software is a multi-layer software, and realizes basic large number calculation and large number matrix calculation by using a plurality of parallel processors of the GPU. The invention can greatly shorten the time of the large number of calculation tasks, and a user does not need to be familiar with or master the related content of the large number of calculation tasks so as to conveniently finish the large number of calculation tasks.
Description
Technical Field
The invention relates to the field of scientific computing, in particular to a GPU-based large number computing system.
Background
In current computer programming languages, the range of values that can be expressed by the data type is limited by the number of bits of the data type. Basic data types for computation include both the reshape and the floating point types. In a common programming language, the largest range of values in the reshaped data type is 64-bit reshaped data, and the largest range in the floating point data type is 64-bit double-precision floating point data. Currently, there are also individual programming languages that support 128-bit data types. However, whether 64-bit or 128-bit data types, the range of values that can be expressed is limited, and the number of significant digits and the precision with which they can be expressed are limited.
In some fields of physical simulation and astronomical calculation, conventional calculation cannot meet the requirements of numerical range and calculation precision, and most of calculations can process calculations with any precision and range. Currently, most computations rely primarily on certain open source libraries (e.g., GMP) and rely on the CPU to perform the computation. With the increasingly deep scientific research, the timeliness requirement of a large number of calculations in some simulation calculation fields is higher and higher. However, in these fields, there are several problems with large number calculations as follows: 1) the principle of big number calculation is different from that of conventional calculation, the big number calculation needs more CPU calculation time, and the time consumed for completing some simulation calculations is far beyond expectation; 2) the existing large-number calculation library has higher use threshold and is not convenient for quick use of personnel in a specific field.
In view of the above, the present invention is particularly proposed.
Disclosure of Invention
In order to solve the problems in the prior art, the invention provides a big number computing system based on a GPU, which effectively improves the speed of big number computing.
The invention provides a GPU-based majority computing system, which comprises: a majority computing hardware system and a majority computing software system;
the majority computing software system runs on the majority computing hardware system;
wherein,
the majority computing software system comprises: client software and server software;
the client software is used for sending calculation request information to the server software;
and the server software is used for carrying out majority calculation and returning operation information to the client software according to the calculation request information.
Preferably, the calculation request information includes: program code, and debug control instructions or run control instructions.
Preferably, the operation information includes: the results of the calculations, and debug information or run information.
Preferably, the server software includes: the system comprises a communication layer, a task management layer, a calculation unit layer and an interpreter layer;
wherein,
the communication layer comprises a communication manager used for executing the communication task between the server software and the client software;
the task management layer comprises a task manager used for distributing and scheduling the calculation tasks according to the received calculation request information and the resource occupation condition of the hardware;
the calculation unit layer comprises: a matrix calculation unit array and a majority calculation unit array;
the matrix computing unit array comprises a plurality of matrix computing units, and the matrix computing units are used for processing logic related to matrix computing;
the large number calculation unit array comprises a plurality of large number calculation units; the big number calculation unit is a bottom layer processing unit for big number calculation and is used for processing basic operation of the big number calculation;
the interpreter layer comprises at least 1 interpreter used for interpreting the program codes submitted by the client software and calling the matrix calculation unit and/or the large number calculation unit of the calculation unit layer according to the logic flow of the program codes.
Preferably, the hardware system comprises a first computer and a GPU server;
the client software runs on the first computer, the server software runs on the GPU server, and the client software and the server software communicate through a network; the network comprises: ethernet, 4G or 5G.
Preferably, the communication layer, the task management layer and the interpreter layer run on a CPU and a memory in the GPU server, and the computation unit layer runs on a GPU processor and a GPU memory in the GPU server.
Preferably, the hardware system comprises a second computer configured with a GPU;
the client software and the server software are both operated on the second computer, and the client software and the server software communicate through a shared memory.
Preferably, the communication layer, the task management layer and the interpreter layer run on a CPU and a memory in the second computer, and the computing unit layer runs on a GPU processor and a GPU memory in the second computer.
Preferably, all of said interpreters run in parallel; all the matrix computing units and the large number computing units are computing units which are executed in parallel and are realized by a parallel processor of a GPU.
Preferably, the matrix calculation unit processes only the logic of the matrix calculation, and the basic operation of the large number calculation is processed by the large number calculation unit.
Preferably, the first computer comprises: desktop, computer workstation, laptop or industrial control computer.
Preferably, the client software includes: the system comprises a program code editing module, a calculation request submitting module, an operation and debugging control module, an operation and debugging information display module and a calculation result display module.
Compared with the closest prior art, the invention has the following beneficial effects:
the GPU-based big number computing system provided by the invention adopts the GPU to complete the basic operation and the matrix computation of big number computing. Compared with the CPU, the GPU includes many parallel processors, and has absolute advantages in parallel processing capability and overall computing power. The calculation unit layer of the present invention includes: a matrix calculation cell array and a majority calculation cell array. The matrix calculation unit array comprises a plurality of matrix calculation units and processing logic related to matrix calculation; the large number calculation unit array comprises a plurality of large number calculation units, and the large number calculation units are bottom layer processing units of large number calculation and are used for processing basic operation of large number calculation. The large number computing unit and the matrix computing unit are parallel execution units of a parallel processor utilizing the GPU, so that the large number computing speed can be greatly improved, and particularly, when large number computing tasks related to large-scale matrix computing are processed, the time consumption of computing can be greatly shortened.
Meanwhile, the invention can also reduce the technical threshold of using the majority of calculation. The server software has the related functions of majority calculation packaged, when a user writes a calculation task program code in the client software, the problem of majority calculation does not need to be considered, the calculation task program code only needs to be written according to a conventional mode, and then a calculation request is submitted to the server software through the client software.
Drawings
FIG. 1 is a schematic diagram of the principal components of a GPU-based majority computing system embodiment of the present invention;
FIG. 2 is a schematic structural diagram of server software according to an embodiment of the present invention;
FIG. 3 is a flow chart of an interpreter processing a piece of program code;
FIG. 4 is a diagram of one form of hardware system in an embodiment of the invention;
FIG. 5 is another form of the hardware system of the embodiment of the present invention;
FIG. 6 is a flowchart illustrating a process of a GPU-based majority computing system performing a computing task according to an embodiment of the present invention.
Detailed Description
Preferred embodiments of the present invention are described below with reference to the accompanying drawings. It should be understood by those skilled in the art that these embodiments are only for explaining the technical principle of the present invention, and are not intended to limit the scope of the present invention.
It should be noted that the terms "first" and "second" in the description of the present invention are used for convenience of description only and do not indicate or imply relative importance of the devices, elements or parameters, and therefore should not be construed as limiting the present invention.
FIG. 1 is a schematic diagram of the main components of a GPU-based majority computing system embodiment of the present invention. As shown in fig. 1, the system 100 of the present embodiment includes: a majority computing hardware system 110 and a majority computing software system 120, and the majority computing software system 120 runs on the majority computing hardware system 110.
Among other things, the majority calculation software system 120 includes: client software 121 and server software 122; the client software 121 is configured to send calculation request information to the server software 122; the server software 122 is configured to perform majority calculation according to the calculation request information and return operation information to the client software 121.
In this embodiment, the calculating the request information includes: program code, and debug control instructions or run control instructions. The operation information includes: the results of the calculations, and debug information or run information.
The server software may be based on the existing GPU development technology, such as CUDA and OpenCL, but is not limited to these GPU programming development tools.
Fig. 2 is a schematic structural diagram of server software in the embodiment of the present invention. As shown in fig. 2, the server software in this embodiment includes: the system comprises a communication layer, a task management layer, a computing unit layer and an interpreter layer.
In this embodiment, the server software includes: the system comprises a communication layer, a task management layer, a computing unit layer and an interpreter layer.
In this embodiment, the communication layer includes a communication manager for executing a communication task between the server software and the client software. The communication content includes but is not limited to program codes, debugging control instructions, operation information, debugging information, calculation results and the like; for example, in some embodiments, in order to facilitate the user to grasp the processing situation of the majority calculation, it is necessary to feed back the processing status of the majority calculation request. In some embodiments, one-to-one communication is provided, i.e., 1 client software communicates with 1 server software. In some embodiments, one-to-many communication is also provided, i.e., communication between multiple client software and 1 server software.
In this embodiment, the task management layer includes a task manager, configured to allocate and schedule the computation task according to the received computation request information and the resource occupation condition of the hardware.
In this embodiment, the calculation unit layer includes: a matrix calculation unit array and a majority calculation unit array; the matrix computing unit array comprises a plurality of matrix computing units (p are included in fig. 2), and the matrix computing units are used for processing logic related to matrix computing; the large number calculation unit array comprises a plurality of large number calculation units (q are contained in figure 2); the large number calculation unit is a bottom layer processing unit of large number calculation and is used for processing basic operation of large number calculation. All the matrix computing units and the large number computing units are computing units which are executed in parallel and are realized by a parallel processor of a GPU. In some preferred embodiments, more sophisticated programming tools may be used to implement the majority and matrix computing units, such as CUDA being selected as the GPU programming tool. Utilizing a GPU stream processor to perform basic large number computing operations such as addition, subtraction, multiplication, division, mathematical functions, and the like; matrix operation logic, such as addition, subtraction, multiplication, division, inversion, etc., of a matrix is implemented using a GPU stream processor.
In this embodiment, the interpreter layer includes at least 1 interpreter, and all the interpreters operate in parallel to interpret a program code submitted by the client software, and call the matrix calculation unit and/or the large number calculation unit of the calculation unit layer according to a logic flow of the program code. The interpreter runs on the CPU, and the processing performance is related to the CPU main frequency and the number of cores (or CPU threads).
FIG. 3 is a flow chart of an interpreter processing a piece of program code. As shown in fig. 3, it includes 5 steps: 1) inputting program code to an interpreter; 2) lexical analysis; 3) analyzing the grammar; 4) AST traversal (AST-abstract syntax tree); 5) and (5) post-calculation processing. When the AST traverses, the interpreter can carry out majority calculation call, namely the interpreter calls a majority calculation unit or a matrix calculation unit; it is also possible to call the majority calculation unit indirectly when the interpreter calls the matrix calculation unit. In a preferred embodiment, the matrix calculation unit only processes the logic of the matrix calculation, and the basic operation for processing the large number calculation is realized by the large number calculation unit; the interpreter can in any case call for a large number of calculation units.
FIGS. 4 and 5 show two hardware components of the GPU-based majority computing system of the present invention:
(1) fig. 4 is a component of a hardware system according to an embodiment of the present invention. The hardware system in fig. 4 includes a first computer and a GPU server. The client software runs on the first computer, the server software runs on the GPU server, and the client software and the server software are communicated through a network; the network comprises: ethernet, 4G or 5G.
The term "first computer" is understood here to mean a conventional computer, which can be a conventional desktop computer, a computer workstation, a laptop or an industrial control computer, etc.
In the hardware composition form of the type (1), a communication layer, a task management layer and an interpreter layer run on a CPU and a memory in a GPU server, and a computing unit layer runs on a GPU processor and a GPU memory in the GPU server. The composition form has strong computing power, is suitable for large-scale large number computing requirements and multi-user large number computing requirements, and the computing power is only limited by the performance of the GPU server. With the development of GPU technology, this hardware organization can provide very powerful computing power.
(2) Fig. 5 shows another configuration of the hardware system according to the embodiment of the present invention. The hardware system of FIG. 5 includes a second computer configured with a GPU; the client software and the server software are both operated on the second computer, and the client software and the server software communicate through the shared memory.
In the hardware composition form in the step (2), the communication layer, the task management layer and the interpreter layer run on a CPU and a memory in the second computer, and the computing unit layer runs on a GPU processor and a GPU memory in the second computer. This form is suitable for small-scale computation, and the performance of the form is determined by the performance of the ordinary computer and the GPU configured by the ordinary computer. Its advantages are high economy and adaptability, and suitable for personal user.
In this embodiment, all interpreters run in parallel; all the matrix computing units and the large number computing units are computing units which are executed in parallel and are realized by a parallel processor of a GPU.
In a preferred embodiment, the matrix calculation unit only processes the logic of the matrix calculation, and the basic operations of the large number calculation are processed by the large number calculation unit.
In this embodiment, the client software includes: the system comprises a program code editing module, a calculation request submitting module, an operation and debugging control module, an operation and debugging information display module and a calculation result display module. Client software provides mainly human-computer interaction interface functions and does not provide a large number of computing functions. In the aspect of program code writing form, a user only needs to write the program code of the calculation task according to a conventional method, and the user is not required to know and master the related content of the calculation of a large number. The realization of the large number calculation is realized by server software, so that the user of the large number calculation system based on the GPU is separated from the concrete realization of the large number calculation.
FIG. 6 is a flowchart illustrating a process of a GPU-based majority computing system performing a computing task according to an embodiment of the present invention. As shown in fig. 6, the process flow includes the following steps: a) a user writes a calculation task code; b) the user submits a program code to the server software; c) the task manager distributes calculation tasks; d) starting an interpreter; e) the interpreter processes the program code and feeds back the processing result. Wherein, the step a) and the step b) are the operation of the user at the client software; steps c) to e) are processing tasks of the server software.
Those of skill in the art will appreciate that the method steps of the examples described in connection with the embodiments disclosed herein may be embodied in electronic hardware, computer software, or combinations of both, and that the components and steps of the examples have been described above generally in terms of their functionality in order to clearly illustrate the interchangeability of electronic hardware and software. Whether such functionality is implemented as electronic hardware or software depends upon the particular application and design constraints imposed on the solution. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present invention.
So far, the technical solutions of the present invention have been described in connection with the preferred embodiments shown in the drawings, but it is easily understood by those skilled in the art that the scope of the present invention is obviously not limited to these specific embodiments. Equivalent changes or substitutions of related technical features can be made by those skilled in the art without departing from the principle of the invention, and the technical scheme after the changes or substitutions can fall into the protection scope of the invention.
Claims (12)
1. A GPU-based majority computing system, the system comprising: a majority computing hardware system and a majority computing software system;
the majority computing software system runs on the majority computing hardware system;
wherein,
the majority computing software system comprises: client software and server software;
the client software is used for sending calculation request information to the server software;
and the server software is used for carrying out majority calculation and returning operation information to the client software according to the calculation request information.
2. A GPU-based majority computation system according to claim 1, wherein the computation request information comprises: program code, and debug control instructions or run control instructions.
3. A GPU-based majority computation system according to claim 1, wherein the operation information comprises: the results of the calculations, and debug information or run information.
4. A GPU-based majority computation system according to claim 1, wherein the server software comprises: the system comprises a communication layer, a task management layer, a calculation unit layer and an interpreter layer;
wherein,
the communication layer comprises a communication manager used for executing the communication task between the server software and the client software;
the task management layer comprises a task manager used for distributing and scheduling the calculation tasks according to the received calculation request information and the resource occupation condition of the hardware;
the calculation unit layer comprises: a matrix calculation unit array and a majority calculation unit array;
the matrix computing unit array comprises a plurality of matrix computing units, and the matrix computing units are used for processing logic related to matrix computing;
the large number calculation unit array comprises a plurality of large number calculation units; the big number calculation unit is a bottom layer processing unit for big number calculation and is used for processing basic operation of the big number calculation;
the interpreter layer comprises at least 1 interpreter used for interpreting the program codes submitted by the client software and calling the matrix calculation unit and/or the large number calculation unit of the calculation unit layer according to the logic flow of the program codes.
5. A GPU-based majority computing system according to claim 4, wherein the hardware system comprises a first computer and a GPU server;
the client software runs on the first computer, the server software runs on the GPU server, and the client software and the server software communicate through a network; the network comprises: ethernet, 4G or 5G.
6. A GPU-based majority computing system according to claim 5, wherein the communication layer, the task management layer and the interpreter layer run on a CPU and a memory in the GPU server, and the compute unit layer runs on a GPU processor and a GPU memory in the GPU server.
7. A GPU-based majority computing system according to claim 4, wherein the hardware system comprises a second computer configured with a GPU;
the client software and the server software are both operated on the second computer, and the client software and the server software communicate through a shared memory.
8. A GPU-based majority computing system according to claim 7, wherein said communication layer, said task management layer, and said interpreter layer run on a CPU and memory in said second computer, and said compute unit layer runs on a GPU processor and a GPU memory in said second computer.
9. A GPU-based majority computation system according to any of claims 4-8,
all of the interpreters run in parallel;
all the matrix computing units and the large number computing units are computing units which are executed in parallel and are realized by a parallel processor of a GPU.
10. A GPU-based majority computation system according to claim 9, wherein the matrix computation unit only processes logic for matrix computations, the basic operations of which are processed by the majority computation unit.
11. A GPU-based majority computation system according to claim 5, wherein the first computer comprises: desktop, computer workstation, laptop or industrial control computer.
12. A GPU-based majority computing system according to any of claims 1-8, wherein the client software comprises: the system comprises a program code editing module, a calculation request submitting module, an operation and debugging control module, an operation and debugging information display module and a calculation result display module.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110270744.5A CN112988114B (en) | 2021-03-12 | 2021-03-12 | GPU-based large number computing system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110270744.5A CN112988114B (en) | 2021-03-12 | 2021-03-12 | GPU-based large number computing system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112988114A true CN112988114A (en) | 2021-06-18 |
| CN112988114B CN112988114B (en) | 2022-04-12 |
Family
ID=76336442
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110270744.5A Active CN112988114B (en) | 2021-03-12 | 2021-03-12 | GPU-based large number computing system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112988114B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113485802A (en) * | 2021-06-29 | 2021-10-08 | 中国科学院自动化研究所 | Easy-to-use large number computing system |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN2821682Y (en) * | 2005-03-10 | 2006-09-27 | 中国科学院自动化研究所 | Distributed type video processor |
| CN103049245A (en) * | 2012-10-25 | 2013-04-17 | 浪潮电子信息产业股份有限公司 | Software performance optimization method based on central processing unit (CPU) multi-core platform |
| CN104461469A (en) * | 2014-11-14 | 2015-03-25 | 成都卫士通信息产业股份有限公司 | Method for achieving SM2 algorithm through GPU in parallelization mode |
| CN105404889A (en) * | 2014-08-21 | 2016-03-16 | 英特尔公司 | Method and apparatus for implementing a nearest neighbor search on a graphics processing unit (gpu) |
| CN108369513A (en) * | 2015-12-22 | 2018-08-03 | 英特尔公司 | For loading-indexing-and-collect instruction and the logic of operation |
| US20190026250A1 (en) * | 2017-07-24 | 2019-01-24 | Tesla, Inc. | Vector computational unit |
| CN110633276A (en) * | 2019-08-07 | 2019-12-31 | 合肥保安集团有限公司 | Armed escort safety early warning system and method based on big data and image recognition |
| KR102194513B1 (en) * | 2019-06-20 | 2020-12-23 | 배재대학교 산학협력단 | Web service system and method using gpgpu based task queue |
| CN112148437A (en) * | 2020-10-21 | 2020-12-29 | 深圳致星科技有限公司 | Calculation task acceleration processing method, device and equipment for federal learning |
-
2021
- 2021-03-12 CN CN202110270744.5A patent/CN112988114B/en active Active
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN2821682Y (en) * | 2005-03-10 | 2006-09-27 | 中国科学院自动化研究所 | Distributed type video processor |
| CN103049245A (en) * | 2012-10-25 | 2013-04-17 | 浪潮电子信息产业股份有限公司 | Software performance optimization method based on central processing unit (CPU) multi-core platform |
| CN105404889A (en) * | 2014-08-21 | 2016-03-16 | 英特尔公司 | Method and apparatus for implementing a nearest neighbor search on a graphics processing unit (gpu) |
| CN104461469A (en) * | 2014-11-14 | 2015-03-25 | 成都卫士通信息产业股份有限公司 | Method for achieving SM2 algorithm through GPU in parallelization mode |
| CN108369513A (en) * | 2015-12-22 | 2018-08-03 | 英特尔公司 | For loading-indexing-and-collect instruction and the logic of operation |
| US20190026250A1 (en) * | 2017-07-24 | 2019-01-24 | Tesla, Inc. | Vector computational unit |
| KR102194513B1 (en) * | 2019-06-20 | 2020-12-23 | 배재대학교 산학협력단 | Web service system and method using gpgpu based task queue |
| CN110633276A (en) * | 2019-08-07 | 2019-12-31 | 合肥保安集团有限公司 | Armed escort safety early warning system and method based on big data and image recognition |
| CN112148437A (en) * | 2020-10-21 | 2020-12-29 | 深圳致星科技有限公司 | Calculation task acceleration processing method, device and equipment for federal learning |
Non-Patent Citations (2)
| Title |
|---|
| 张向响: "基于GPU的综合孔径微波辐射计仿真平台设计", 《中国优秀博硕士学位论文全文数据库(硕士)工程科技Ⅱ辑》 * |
| 邹伟: "基于ARM处理器的单目视觉测距定位系统", 《控制工程》 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113485802A (en) * | 2021-06-29 | 2021-10-08 | 中国科学院自动化研究所 | Easy-to-use large number computing system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112988114B (en) | 2022-04-12 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US9983857B2 (en) | Dynamic computational acceleration using a heterogeneous hardware infrastructure | |
| Zhao et al. | Scientific workflow systems for 21st century, new bottle or new wine? | |
| CN102576314B (en) | The mapping with the data parallel thread across multiple processors processes logic | |
| RU2597556C2 (en) | Computer cluster arrangement for executing computation tasks and method for operation thereof | |
| US9753726B2 (en) | Computer for amdahl-compliant algorithms like matrix inversion | |
| Kelly | GPU computing for atmospheric modeling | |
| Maroosi et al. | Parallel and distributed computing models on a graphics processing unit to accelerate simulation of membrane systems | |
| CN112988114B (en) | GPU-based large number computing system | |
| Zhong et al. | Model-based parallelizer for embedded control systems on single-isa heterogeneous multicore processors | |
| CN112631986A (en) | Large-scale DSP parallel computing device | |
| CN111666077A (en) | Operator processing method and device, electronic equipment and storage medium | |
| US7831803B2 (en) | Executing multiple instructions multiple date (‘MIMD’) programs on a single instruction multiple data (‘SIMD’) machine | |
| Agullo et al. | Dynamically scheduled Cholesky factorization on multicore architectures with GPU accelerators. | |
| Liu et al. | BSPCloud: A hybrid distributed-memory and shared-memory programming model | |
| CN110209631A (en) | Big data processing method and its processing system | |
| CN115904328A (en) | LLVM intermediate language-based parallel computing framework conversion method for cross-GPU architecture | |
| CN113076191A (en) | Cluster GPU resource scheduling system | |
| Krömer et al. | An implementation of differential evolution for independent tasks scheduling on GPU | |
| Sarker et al. | Radical-Cylon: A Heterogeneous Data Pipeline for Scientific Computing | |
| Huang et al. | Performance optimization of High-Performance LINPACK based on GPU-centric model on heterogeneous systems | |
| CN113485802A (en) | Easy-to-use large number computing system | |
| Searles et al. | Creating a portable, high-level graph analytics paradigm for compute and data-intensive applications | |
| Tarakji et al. | Os support for load scheduling on accelerator-based heterogeneous systems | |
| Levchenko et al. | DDCI: Simple dynamic semiautomatic parallelizing for heterogeneous multicomputer systems | |
| Ahmed et al. | An automated lightweight framework for scheduling and profiling parallel workflows simultaneously on multiple hypervisors |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |