CN112486424A - Parallel printing order-preserving agent method for user-transparent operation cores - Google Patents
Parallel printing order-preserving agent method for user-transparent operation cores Download PDFInfo
- Publication number
- CN112486424A CN112486424A CN201910861483.7A CN201910861483A CN112486424A CN 112486424 A CN112486424 A CN 112486424A CN 201910861483 A CN201910861483 A CN 201910861483A CN 112486424 A CN112486424 A CN 112486424A
- Authority
- CN
- China
- Prior art keywords
- core
- parameter
- printing
- data
- statement
- 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
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F3/00—Input arrangements for transferring data to be processed into a form capable of being handled by the computer; Output arrangements for transferring data from processing unit to output unit, e.g. interface arrangements
- G06F3/12—Digital output to print unit, e.g. line printer, chain printer
- G06F3/1201—Dedicated interfaces to print systems
- G06F3/1223—Dedicated interfaces to print systems specifically adapted to use a particular technique
- G06F3/1237—Print job management
- G06F3/124—Parallel printing or parallel ripping
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F3/00—Input arrangements for transferring data to be processed into a form capable of being handled by the computer; Output arrangements for transferring data from processing unit to output unit, e.g. interface arrangements
- G06F3/12—Digital output to print unit, e.g. line printer, chain printer
- G06F3/1201—Dedicated interfaces to print systems
- G06F3/1278—Dedicated interfaces to print systems specifically adapted to adopt a particular infrastructure
- G06F3/1281—Multi engine printer devices, e.g. one entity having multiple output engines
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F3/00—Input arrangements for transferring data to be processed into a form capable of being handled by the computer; Output arrangements for transferring data from processing unit to output unit, e.g. interface arrangements
- G06F3/12—Digital output to print unit, e.g. line printer, chain printer
- G06F3/1293—Printer information exchange with computer
- G06F3/1295—Buffering means
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F3/00—Input arrangements for transferring data to be processed into a form capable of being handled by the computer; Output arrangements for transferring data from processing unit to output unit, e.g. interface arrangements
- G06F3/12—Digital output to print unit, e.g. line printer, chain printer
- G06F3/1297—Printer code translation, conversion, emulation, compression; Configuration of printer parameters
- G06F3/1298—Printer language recognition, e.g. programme control language, page description language
Abstract
The invention discloses a user transparent arithmetic core parallel printing order-preserving agent method, which carries out extension replacement on a printout statement on an arithmetic core, and converts the printout statement into a thread printing processing function in a system during calling operation; performing parameter analysis on the thread printing processing function, applying for a data unloading buffer area, and unloading data according to the parameter type; the operation core requests the control core to perform printing order-preserving proxy and transmits the first address of the applied data transfer buffer area to the control core; and the control core receives the interrupt signal, the interrupt processing is carried out, the embedding assembly instruction is executed according to the parameter type of the data unloading buffer to load the printing function parameter, and the printf agent is called to realize the printing output of the operation core. The invention adopts the combination of a compiler and a runtime to automatically complete the parallel and ordered output of all the operation cores; the method is transparent to users and does not need to additionally manage the printout of the operation cores; the basic compiler of the operation core is only required to perform simple program transformation on the operation core program.
Description
Technical Field
The invention belongs to the technical field of heterogeneous many-core, and particularly relates to a parallel printing order-preserving agent method for a user-transparent operation core.
Background
The operation cores of the heterogeneous many-core do not support the operation of an operation system in the general sense, only one simple management library is provided, the support of a user program printing function is limited, when the printing function is called by a plurality of operation cores simultaneously, screen outputs of the plurality of operation cores are staggered together, useful information cannot be identified, and the debugging and operation control of the program are influenced. Fig. 1 shows the print information of eight arithmetic cores, and it can be seen that the print information is staggered and disordered, and useful information cannot be effectively identified. The failure of multiple computational cores to effectively recognize print information can seriously affect the debugging and operation control of a program.
Although, in 2011, 11 th month, volume 37, 21 nd of computer engineering, an article, "I/O agent technology for compiler guidance in heterogeneous multi-core systems" is disclosed, in which an implementation scheme of compiler guidance and main processor agent I/O is described, where guidance and analysis of a parallel compiler are important, a main processor c program and a slave processor c program are generated by analyzing and processing the parallel program through the parallel compiler, an I/O statement in the slave processor program is copied in the main processor program, and when an agent is implemented, it is necessary to identify which I/O statement the main processor should execute in the agent at a current time point. The I/O proxy technology has the following defects: (1) the I/O agent technology needs a parallel compiler to perform source-to-source processing on a parallel program, and the processing of the compiler is emphasized, so that the problem of the language can be solved only by performing complex analysis and program transformation on the compiler of each parallel language, and the application range is not wide; (2) the processing method is based on static program transformation, so that each I/O statement needs to be subjected to secondary identification for distinguishing real agent statements, and the processing is complex.
Disclosure of Invention
The invention aims to provide a user-transparent parallel printing order-preserving agent method for operation cores, which aims to solve the problem that the printing output of all operation cores of a heterogeneous many-core is orderly and readable without mutual influence under the condition of being transparent to a user.
In order to achieve the purpose, the invention adopts the technical scheme that: a transparent user operating core parallel printing order-preserving agent method comprises the following steps,
s1, the compiler performs expansion replacement on the printout statement on the operation core, and converts the printout statement into a thread printing processing function in a calling runtime system;
s2, the run-time system performs data unloading and control core interruption according to the parameters of the thread printing processing function of S1, and sends an agent request, which comprises the following steps:
s21, performing parameter analysis by a thread printing processing function in the running system, applying for a data unloading buffer area, and unloading data according to the parameter type;
s22, the operation core sends a signal to interrupt the control core, requests the control core to perform printing order-preserving proxy, and simultaneously transmits the first address of the applied data dump buffer to the control core through an interrupt signal;
and S3, the control core switches to interrupt processing after receiving the interrupt signal of S22, executes a corresponding assembly embedding instruction according to the parameter type in the data unloading buffer area to load the printing function parameters, and calls a printf function agent to realize the printing output of the operation core.
The technical scheme of further improvement in the technical scheme is as follows:
1. in the above scheme, the expanding and replacing of the printout statement comprises the following steps,
replacing the function name, and replacing the printing output statement by a thread printing processing function of the runtime system by the compiler;
and expanding a parameter chain, wherein a compiler analyzes variable information in a printout statement, expands a type parameter for each variable needing to be printed out, explains the data type of the variable, provides required space length information for the data dump during operation, counts the total number of the variables in the printout statement, adds the total number of the variables as an expanded parameter into the parameter chain, and provides control information for the data dump during operation.
2. In the above scheme, the data dump is that all source variables printed and output by the operation core are copied in a space accessible by both the operation core and the control core, where the space is a dump buffer, and the dump buffer is divided into three areas: the system comprises a parameter information area, an output variable index area and an output variable pool, wherein the parameter information area explains the number of parameters and the data type of each parameter and is used for constructing an output variable when a printing output statement is constructed; the output variable index area stores the index value of each output variable in the dump buffer for use and positioning of the output variable; and the output variable pool stores all output variable values.
3. In the above scheme, the parameter data types in the data dump are different, and the used assembler instructions are different.
Due to the application of the technical scheme, compared with the prior art, the invention has the following advantages:
1) the invention discloses a transparent parallel printing order-preserving agent method for operation cores, which automatically completes the parallel and ordered output of all operation cores by adopting a mode of combining a compiler and a runtime.
2) The transparent operation core parallel printing order-preserving agent method is transparent to the user, the user does not need to additionally manage the printing output of the operation core, and only needs to write according to the standard I/O format, thereby reducing the burden of the user and increasing the friendliness.
3) The transparent parallel printing order-preserving agent method for the operation cores ensures that the output is orderly and readable, each sentence of each operation core is printed completely and is output according to the printing format requirement, the printing information of different operation cores cannot interfere with each other, the readability is good, and the debugging and the operation of a user program are effectively guaranteed.
4) The transparent parallel printing order-preserving agent method for the operation core has wide applicability, does not need a parallel compiler to perform source-to-source processing on a parallel program, does not need to perform complex identification on each I/O statement to distinguish and position real agent statements, only needs a basic compiler of the operation core to perform simple program transformation on the operation core program, and realizes real-time corresponding processing of a system when the system is mainly operated, so that the method can support multiple parallel languages.
Drawings
FIG. 1 is a schematic diagram of printed information from a plurality of computational cores without an order preserving agent.
FIG. 2 is a flow chart of the present invention.
FIG. 3 is a schematic diagram of the present invention for expanding and replacing the printout sentence.
FIG. 4 is a diagram of a buffer structure according to the present invention.
Detailed Description
In the description of this patent, it is noted that the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer", etc., indicate orientations or positional relationships based on the orientations or positional relationships shown in the drawings, and are only for convenience in describing the present invention and simplifying the description, but do not indicate or imply that the device or element being referred to must have a particular orientation, be constructed and operated in a particular orientation, and thus should not be construed as limiting the present invention; the terms "first," "second," and "third" are used for descriptive purposes only and are not to be construed as indicating or implying relative importance; furthermore, unless expressly stated or limited otherwise, the terms "mounted," "connected," and "connected" are to be construed broadly, as they may be fixedly connected, detachably connected, or integrally connected, for example; can be mechanically or electrically connected; they may be connected directly or indirectly through intervening media, or they may be interconnected between two elements. The meaning of the above terms in this patent may be specifically understood by those of ordinary skill in the art.
The invention is further described below with reference to the following examples:
example (b): as shown in fig. 2, a parallel printing order-preserving agent method for a user transparent arithmetic core comprises the following steps,
s1, the compiler performs expansion replacement on the printout statement on the operation core, and converts the printout statement into a thread printing processing function in a calling runtime system;
s2, the run-time system performs data unloading and control core interruption according to the parameters of the thread printing processing function of S1, and sends an agent request, which comprises the following steps:
s21, performing parameter analysis by a thread printing processing function in the running system, applying for a data unloading buffer area, and unloading data according to the parameter type;
s22, the operation core sends a signal to interrupt the control core, requests the control core to perform printing order-preserving proxy, and simultaneously transmits the first address of the applied data dump buffer to the control core through an interrupt signal;
and S3, the control core switches to interrupt processing after receiving the interrupt signal of S22, executes a corresponding assembly embedding instruction according to the parameter type in the data unloading buffer area to load the printing function parameters, and calls a printf function agent to realize the printing output of the operation core.
As shown in fig. 3, the expanded replacement of the printout sentence includes the steps of,
replacing the function name, and replacing the printing output statement by a thread printing processing function of the runtime system by the compiler;
and expanding a parameter chain, wherein a compiler analyzes variable information in a printout statement, expands a type parameter for each variable needing to be printed out, explains the data type of the variable, provides required space length information for the data dump during operation, counts the total number of the variables in the printout statement, adds the total number of the variables as an expanded parameter into the parameter chain, and provides control information for the data dump during operation.
As shown in fig. 4, the data dump is a copy of all source variables printed and output by the operation core in a space accessible to both the operation core and the control core, where the space is a dump buffer, and the dump buffer is divided into three areas: the system comprises a parameter information area, an output variable index area and an output variable pool, wherein the parameter information area explains the number of parameters and the data type of each parameter and is used for constructing an output variable when a printing output statement is constructed; the output variable index area stores the index value of each output variable in the dump buffer for use and positioning of the output variable; and the output variable pool stores all output variable values.
The parameter data types in the data unloading are different, and the used embedded assembly instructions are different.
The examples are further explained below: the parallel printing order-preserving agent method of the invention is to transmit the printing output of the operation core to the control core agent for completion, wherein the printing output is completed together when the compiler is required to run cooperatively, the compiler used by the method is the operation core basic compiler, and the user does not need to participate, and only needs to write and print the statement in a standard I/O mode.
The conversion of the printout statement is automatically completed by a compiler on the basis of program analysis, and not only is the function name replaced, but also a parameter chain is changed and parameter information is expanded so as to facilitate the processing of a system during running. The method for replacing the expansion of the printout statement comprises the following steps: (1) replacing the function name, and replacing the printing output statement with a thread printing processing function of the runtime system; (2) the extended parameter chain is used for analyzing variable information in the printout statement, extending a type parameter for each variable to be printed and output, explaining the data type of the variable, providing required space length information for data dump during operation and providing type information for printing agent parameter reconstruction; and counting the total number of variables in the printout statement, adding the total number into a parameter chain as an expansion parameter, and providing control information for data unloading during operation.
In a heterogeneous many-core system, storage spaces accessible by an operation core and a control core are intersected, that is, a space accessible only by the operation core exists, and a space accessible by both the operation core and the control core also exists. In order to effectively perform the proxy of the printout, it is necessary to ensure that all the printout variables are in the space accessible to both the operation core and the control core, which requires that the runtime system must perform data dump.
The data dump is completed by the operation core, and the basic idea is to copy one copy of all source variables needing to be printed and output by the operation core in a printout statement in a space accessible to both the operation core and the control core, wherein the space is called a dump buffer, so that the number of the variables needs to be known for confirming the size of the application dump buffer, and the type information and the value of each variable need to be known for obtaining the correct value of each variable, and all the information is obtained by setting parameters of a thread printing processing function.
The space attribute of the source variable printed and output by the operation core is not considered when the data dump design is carried out, even if the variable is originally in the space which can be accessed by both the operation core and the control core, the variable still needs to be copied into the dump buffer of the new application, so that all the parameters are uniformly managed in the dump buffer of the new application, when the control core performs proxy printing and output, the access index value of the current parameter is obtained according to the position of each output parameter, the parameter value can be conveniently and quickly obtained, and the efficiency of printing and output is improved.
The dump buffer is divided into three regions: the device comprises a parameter information area, an output variable index area and an output variable pool. The parameter information area explains the number of parameters and the data type of each parameter and is used for constructing an output variable when a printing output statement is constructed; the output variable index area stores the index value of each output variable in the dump buffer for use and positioning of the output variable; the output variable pool stores all output variable values, including the format control strings of the print statements.
And after the operation core finishes data transfer, the operation core interrupts the control core and simultaneously transmits the first address of the applied transfer buffer to the control core. And after receiving the interrupt, the control core shifts to interrupt processing, and the agent realizes the printing output of the operation core. The control core runs a complete operating system, the interrupt processing mechanism is complete, and the order retention and integrity of the interrupt processing when a plurality of operation cores send interrupts to the control core are guaranteed. The control core proxies the target of the printout to complete the printout by calling the print function through the control core.
Printf is a variable parameter function, the number of parameters is uncertain, the agent of the control core is operated during operation, all information comes from the dump buffer, and when the control core calls printf, the parameters can be constructed only through the dump buffer. Because the number of parameters in the dump buffer and the type of each parameter are selected in a plurality of ways, the output parameters of the printf statement need to directly reflect the type of the parameters, the conditions are too complex, and the printf is not suitable for writing.
The invention solves the problem of writing and expressing the variable parameter function printf by means of assembly. When a printout statement in a high-level language is translated into an assembler, each parameter is filled one by one, each parameter can be filled into a corresponding register or a corresponding position of a memory as long as the position and the type of the parameter are known, and the filling of the parameters is independent of each other, so that various types of problems of filling each parameter, variable parameters and parameters in sequence by using the assembly embedding implementation are solved easily.
Each processor architecture has its own features for the delivery and receipt of function parameters. How to construct the parameters and print out the parameters by using the embedded assembly is described below by taking an Shenwei heterogeneous many-core processor as an example.
The function parameter transfer of the Shenwei processor uses a specific register, the first six parameters of the function are put in $16 to $21, and if floating point parameters exist, the function parameters are put in the corresponding $ f16 to $ f21 according to the positions of the parameters. If the number of parameters of the function is more than 6, the rest of the parameters are placed at the bottom of the stack of the parent stack. Therefore, for the first 6 parameters, only the corresponding registers need to be filled, for the later parameters, the current stack needs to be dug before printf is called, the parameters are filled to the bottom of the stack, and the current stack is recovered after the calling is finished.
The data types of the parameters are different, and the assembler instructions used are also different. So at parameter loading time, it is necessary to refer to the parameter type information in the dump buffer, and use the correct instruction and the correct register.
After the parameters are filled by using the embedded assembly statement, the c statement printf (ss) is directly called to complete the printing output, and the ss points to the printing output format control character string. When the Printf statement is executed, the format control character string is analyzed, parameters are read from a corresponding register or the bottom of a parent stack according to an analysis result, and the parameters are prepared by using an embedded assembly statement. The advantage of directly calling the c statement instead of embedding assembly calling is that the change of gp pointers related before and after function calling is ensured by the compiler to be correct, safe and easy to use.
When a user transparent arithmetic core parallel printing order-preserving agent method is adopted, (1) an expansion replacement method of a printing output statement is adopted, and the method provides effective parameter information for a printing agent during operation; (2) the structure setting method of the dump buffer is characterized in that the structure directly serves a printing agent, and effective structure setting can ensure rapid construction of parameters and promote efficient completion of the printing agent; (3) the parameter construction strategy of the assembly embedding method solves the writing expression problem of the variable parameter function printf in the process of agent operation.
The above embodiments are merely illustrative of the technical ideas and features of the present invention, and the purpose thereof is to enable those skilled in the art to understand the contents of the present invention and implement the present invention, and not to limit the protection scope of the present invention. All equivalent changes and modifications made according to the spirit of the present invention should be covered within the protection scope of the present invention.
Claims (4)
1. A transparent user operation core parallel printing order-preserving agent method is characterized in that: comprises the following steps of (a) carrying out,
s1, the compiler performs expansion replacement on the printout statement on the operation core, and converts the printout statement into a thread printing processing function in a calling runtime system;
s2, the run-time system performs data unloading and control core interruption according to the parameters of the thread printing processing function of S1, and sends an agent request, which comprises the following steps:
s21, performing parameter analysis by a thread printing processing function in the running system, applying for a data unloading buffer area, and unloading data according to the parameter type;
s22, the operation core sends a signal to interrupt the control core, requests the control core to perform printing order-preserving proxy, and simultaneously transmits the first address of the applied data dump buffer to the control core through an interrupt signal;
and S3, the control core switches to interrupt processing after receiving the interrupt signal of S22, executes a corresponding assembly embedding instruction according to the parameter type in the data unloading buffer area to load the printing function parameters, and calls a printf function agent to realize the printing output of the operation core.
2. The user-transparent arithmetic core parallel printing order-preserving agent method according to claim 1, characterized in that: the expanded replacement of the printout sentence includes the steps of,
replacing the function name, and replacing the printing output statement by a thread printing processing function of the runtime system by the compiler;
and expanding a parameter chain, wherein a compiler analyzes variable information in a printout statement, expands a type parameter for each variable needing to be printed out, explains the data type of the variable, provides required space length information for the data dump during operation, counts the total number of the variables in the printout statement, adds the total number of the variables as an expanded parameter into the parameter chain, and provides control information for the data dump during operation.
3. The user-transparent arithmetic core parallel printing order-preserving agent method according to claim 2, characterized in that: the data unloading is that all source variables printed and output by the operation core are copied in a space which can be accessed by the operation core and the control core, wherein the space is unloading buffer which is divided into three areas: the system comprises a parameter information area, an output variable index area and an output variable pool, wherein the parameter information area explains the number of parameters and the data type of each parameter and is used for constructing an output variable when a printing output statement is constructed; the output variable index area stores the index value of each output variable in the dump buffer for use and positioning of the output variable; and the output variable pool stores all output variable values.
4. The parallel printing order-preserving agent method for the user-transparent arithmetic core according to any one of claims 1 to 3, characterized in that: the parameter data types in the data unloading are different, and the used embedded assembly instructions are different.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910861483.7A CN112486424B (en) | 2019-09-12 | 2019-09-12 | User transparent arithmetic core parallel printing order-preserving proxy method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910861483.7A CN112486424B (en) | 2019-09-12 | 2019-09-12 | User transparent arithmetic core parallel printing order-preserving proxy method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112486424A true CN112486424A (en) | 2021-03-12 |
| CN112486424B CN112486424B (en) | 2022-10-04 |
Family
ID=74920367
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910861483.7A Active CN112486424B (en) | 2019-09-12 | 2019-09-12 | User transparent arithmetic core parallel printing order-preserving proxy method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112486424B (en) |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20130346732A1 (en) * | 2011-03-04 | 2013-12-26 | Nec Corporation | Deadlock avoidance method and deadlock avoidance mechanism |
| CN106776304A (en) * | 2016-12-05 | 2017-05-31 | 云鸟（上海）能源科技有限公司 | Method based on printout information is controlled in multithreading exploitation under built-in Linux |
-
2019
- 2019-09-12 CN CN201910861483.7A patent/CN112486424B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20130346732A1 (en) * | 2011-03-04 | 2013-12-26 | Nec Corporation | Deadlock avoidance method and deadlock avoidance mechanism |
| CN106776304A (en) * | 2016-12-05 | 2017-05-31 | 云鸟（上海）能源科技有限公司 | Method based on printout information is controlled in multithreading exploitation under built-in Linux |
Non-Patent Citations (2)
| Title |
|---|
| JIAN-HONG ZHENG 等: "《Study and design of embedded trace debugging technology based on virtual technology》", 《2012 INTERNATIONAL CONFERENCE ON WAVELET ACTIVE MEDIA TECHNOLOGY AND INFORMATION PROCESSING (ICWAMTIP)》 * |
| 巩琛 等: "《基于ARM的Linux驱动调试技术研究》", 《计算机应用与软件》 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112486424B (en) | 2022-10-04 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| RU2550558C2 (en) | Comparing and replacing dynamic address translation table entry | |
| Simpson et al. | MemSafe: ensuring the spatial and temporal memory safety of C at runtime | |
| CN103116529B (en) | The method and apparatus of heap memory damage is caused for detecting buffer overflow in real time | |
| US9530006B2 (en) | Method and system for performing a memory safety check of a program written in an unmanaged programming language | |
| US9471291B2 (en) | Multi-processor code for modification for storage areas | |
| CN109684794A (en) | A kind of code protection virtual machine KVM network system realization, device, computer equipment and storage medium | |
| CN103440457A (en) | Binary program analytic system based on process simulation | |
| Hasabnis et al. | Extracting instruction semantics via symbolic execution of code generators | |
| US8732684B2 (en) | Program conversion apparatus and computer readable medium | |
| Kloos et al. | Asynchronous liquid separation types | |
| CN105573763A (en) | Embedded system modeling method supporting RTOS | |
| CN112486424B (en) | User transparent arithmetic core parallel printing order-preserving proxy method | |
| Lowther et al. | Morello MicroPython: A Python Interpreter for CHERI | |
| Cogumbreiro et al. | Memory access protocols: certified data-race freedom for GPU kernels | |
| Wang et al. | Automatic translation of data parallel programs for heterogeneous parallelism through OpenMP offloading | |
| Selfridge | GRIFT: A richly-typed, deeply-embedded RISC-V semantics written in Haskell | |
| Ročkai | Model checking software | |
| CN101763292B (en) | Filtering device for processor presumed access and filtering method thereof based on address window | |
| CN103049302B (en) | The method of the strcpy standard library function assembly code optimized by Program Generating | |
| Chatterjee et al. | A low-level memory model and an accompanying reachability predicate | |
| Kistler et al. | Detecting race conditions in asynchronous DMA operations with full system simulation | |
| CN110362501B (en) | Apparatus and method for performing saturated addressing load and store operations | |
| Guo et al. | OpenMP offloading data transfer optimization for DCUs | |
| Qian et al. | A measurable refinement method of design and verification for micro-kernel operating systems in communication network | |
| Liu et al. | The Complexity of Weak Consistency |
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 |