WO2012129886A1 - 一种跳转指令编码的方法和系统 - Google Patents

一种跳转指令编码的方法和系统 Download PDF

Info

Publication number
WO2012129886A1
WO2012129886A1 PCT/CN2011/078846 CN2011078846W WO2012129886A1 WO 2012129886 A1 WO2012129886 A1 WO 2012129886A1 CN 2011078846 W CN2011078846 W CN 2011078846W WO 2012129886 A1 WO2012129886 A1 WO 2012129886A1
Authority
WO
WIPO (PCT)
Prior art keywords
loop
instruction
advance
start instruction
memory
Prior art date
Application number
PCT/CN2011/078846
Other languages
English (en)
French (fr)
Inventor
黎立煌
帕瓦旺萨克⋅保罗
曹庆新
Original Assignee
中兴通讯股份有限公司
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by 中兴通讯股份有限公司 filed Critical 中兴通讯股份有限公司
Priority to EP11862348.7A priority Critical patent/EP2696279A4/en
Publication of WO2012129886A1 publication Critical patent/WO2012129886A1/zh

Links

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F9/00Arrangements for program control, e.g. control units
    • G06F9/06Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
    • G06F9/30Arrangements for executing machine instructions, e.g. instruction decode
    • G06F9/30003Arrangements for executing specific machine instructions
    • G06F9/3005Arrangements for executing specific machine instructions to perform operations for flow control
    • G06F9/30065Loop control instructions; iterative instructions, e.g. LOOP, REPEAT
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F9/00Arrangements for program control, e.g. control units
    • G06F9/06Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
    • G06F9/30Arrangements for executing machine instructions, e.g. instruction decode
    • G06F9/38Concurrent instruction execution, e.g. pipeline or look ahead
    • G06F9/3802Instruction prefetching
    • G06F9/3804Instruction prefetching for branches, e.g. hedging, branch folding

Definitions

  • the present invention relates to the field of processors, and in particular, to a method and system for encoding a jump instruction. Background technique
  • the main object of the present invention is to provide a method and system for skip instruction encoding to simplify hardware logic and improve the reliability and performance of the implementation.
  • a method for encoding a jump instruction comprising:
  • the loop start instruction is acquired in advance with respect to the operation cycle; when the last instruction of the loop body is executed, the acquisition of the loop start instruction is completed.
  • the method for obtaining a loop start instruction in advance to the memory relative to the operation cycle is:
  • the N week corresponds to the cycle of the cycle start instruction to the last instruction of the loop body.
  • the process of completing the acquisition of the loop start instruction includes:
  • the hardware finds the last instruction of the loop body N weeks in advance, thereby acquiring the instruction from the memory MEMORY N weeks in advance.
  • the method further includes:
  • the instruction pipeline involved in executing the loop PIPELINE continues to work without waiting.
  • the method is implemented by a sink compiler ASS sink compiler.
  • a system for jumping instruction encoding comprising an operation cycle advance unit and a loop start instruction acquisition unit;
  • the operation cycle advancement unit is configured to, when executing the loop, confirm that the loop start instruction is to be acquired in the memory in advance with respect to the operation cycle;
  • the loop start instruction acquisition unit is configured to complete the acquisition of the loop start instruction when the last instruction of the loop body is executed.
  • the operation cycle advancement unit is used for: when the cycle start instruction is acquired in the memory in advance with respect to the operation cycle,
  • the N week corresponds to the cycle of the cycle start instruction to the last instruction of the loop body.
  • the loop start instruction acquiring unit completes the acquisition of the loop start instruction, it is used to: trigger execution of the exe.bin instruction, so that the hardware finds the last instruction of the loop body N weeks in advance, thereby acquiring the MEMORY in advance N weeks in advance. instruction.
  • the system also includes PIPELINE, which involves executing loops, to continue working without waiting.
  • the system is set in the ASS.
  • the method and system for the jump instruction encoding of the present invention simplifies the hardware logic from the software level by means of the pre-processing of the assembly compiler (ASS), ensuring that no additional BUFFER and counter logic are needed to solve the loopback. Pipeline needs to be paused, eliminating a lot of hardware logic and hardware resources, saving area and cost, making implementation simple and improving the reliability of the implementation.
  • ASS assembly compiler
  • FIG. 1 is a schematic diagram of a jump instruction encoding process according to an embodiment of the present invention.
  • FIG. 2 is a diagram of a jump instruction encoding system according to an embodiment of the present invention. detailed description
  • the LOOP_BEGIN instruction Since the LOOP_BEGIN instruction is taken from MEMORY during the loopback, it takes N cycles. If the MEMORY instruction is started after waiting for the LOOP-END instruction, the pipeline must be stopped for N cycles, and the performance loss is large. In order to avoid this loss, you can go to MEMORY fetch instructions N times in hardware, so when LOOP END is executed, The LOOP-BEGIN command has also been retrieved from MEMORY, so that PIPELINE does not need to pause.
  • ASS can encode assembly instructions written by programmers into binary or hexadecimal.
  • the LOOP_END flag can be advanced N weeks (this cycle is the operation cycle of LOOP_BEGIN to LOOP-END) to generate the required executable file exe.bin.
  • the hardware can see LOOP_END in advance N weeks, so that the instruction is taken from MEMORY N weeks in advance, eliminating the PIPELINE pause and achieving high performance.
  • sequence of assembly code instructions written can be exemplified as:
  • Inst2 to Inst6 represent loop body instructions.
  • the machine code sequence after ASS assembly can be exemplified as:
  • LOOP-MARKER loop tag occupies the lower 2 bits of each instruction, 1 means LOOP-BEGIN, 2 means LOOP- END, 0 means no loop label.
  • 1 means LOOP-BEGIN
  • 2 means LOOP- END
  • 0 means no loop label.
  • LOOP_END is advanced by 2 instructions during encoding.
  • Inst6 corresponds to LOOP_END, which is encoded in Inst4 in the machine code, and two instructions are advanced. This process is done automatically by the compiler, and the loop body written by the programmer is still Inst2-Inst6.
  • LOOP_END that is, the machine code is 2 bits lower than 2
  • Inst2 has been fetched from MEMORY (this assumes that it takes 2 weeks to fetch instructions from MEMORY), and the instructions are executed sequentially from Inst2.
  • Inst4 go to MEMORY and take Inst2, which forms a loop execution of Inst2 to Inst6 until the loop exits.
  • Step 110 When the loop is executed, the loop start instruction is acquired in advance with respect to the operation cycle.
  • Step 120 When the last instruction of the loop body is executed, the acquisition of the loop start instruction is completed.
  • the method for obtaining the loop start instruction in advance in the memory may be: advance the flag of the last instruction of the loop body by N weeks, and generate a required execution file exe.bin; N week corresponds to the cycle of the cycle start instruction to the last instruction of the loop body.
  • the process of completing the acquisition of the loop start instruction may include: when the exe.bin instruction is executed, the hardware finds the last instruction of the loop body N weeks in advance, thereby acquiring the instruction from the MEMORY N weeks in advance. It is also possible to continue working without waiting by PIPELINE, which involves executing the loop.
  • FIG. 2 is a diagram of a jump instruction encoding system according to an embodiment of the present invention.
  • the system includes a connected operation cycle advance unit and a loop start instruction acquisition unit.
  • the operation cycle advancement unit can confirm the advance to the memory to obtain the loop start instruction relative to the operation cycle when executing the loop; the loop start instruction acquisition unit can complete the loop when the last instruction of the loop body is executed. Get the acquisition of the instruction.
  • the operation cycle advance unit may be used to advance the flag of the last instruction of the loop body by N weeks and generate the required execution file when the cycle start instruction is acquired in advance in the operation cycle.
  • N week corresponds to the operation cycle of the loop start instruction to the last instruction of the loop body.
  • the loop start instruction acquisition unit may be used to: trigger execution of the exe.bin instruction, so that the hardware finds the last instruction of the loop body N weeks in advance, thereby acquiring from the MEMORY N weeks in advance.
  • the instruction It is also possible to continue working with PIPELINE, which involves executing the loop, without waiting.
  • system of the present invention can be placed in an ASS.
  • the method and system for the jump instruction encoding of the present invention simplifies the hardware logic from the software level by the preprocessing of the ASS assembly code, ensuring that no additional BUFFER and counters need to be added.
  • the logic solves the problem that the pipeline needs to be paused when the loop is folded back, which saves a lot of hardware logic and hardware resources, saves area and cost, makes the implementation simple, and improves the reliability of the implementation scheme.

Landscapes

  • Engineering & Computer Science (AREA)
  • Software Systems (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Advance Control (AREA)
  • Devices For Executing Special Programs (AREA)
  • Executing Machine-Instructions (AREA)

Description

一种跳转指令编码的方法和系统 技术领域
本发明涉及处理器领域, 具体涉及一种跳转指令编码的方法和系统。 背景技术
目前, 处理器性能损失的原因之一可能是循环指令的执行, 一个循环 体包含 N条指令, 它需要被重复执行多次。 循环体最后一条指令可以用 LOOP— END 表示, 当在 PIPELINE (指令流水线) 的译码阶段看到 L00P END时, 如果此时才从 MEMORY (存储器)取 LOOP_BEGIN (循 环体第一条指令), PIPELINE必须停下来等待 MEMORY, 等待的周期取决 于 MEMORY 的速度和架构。 对于常见的处理器, 循环指令的使用频率是 非常频繁的, 所以这样的开销带来的性能损失是非常大的。 现有的解决办 法有两种, 第一种是添加 LOOP— BUFFER (循环緩冲器), 当循环体第一次 执行时, 把循环体的指令保存到 LOOP— BUFFER, 第二次执行时就不需要 从 MEMORY取 LOOP— BEGIN,而是直接从 BUFFER获取,这样 PIPELINE 就不需要停顿。 这种实现方案需要添加 BUFFER (緩冲器) 以及相应的硬 件逻辑, 电路也非常复杂, 实现的代价大。 第二种是添加计数器, 比如添 加一条专门的 LOOP— BEGIN指令, 它包含循环体指令条数的信息, 这样从 第一条开始计数,每执行一条就减一,因而可以提前知道 LOOP— END到达, 提前从 MEMORY取指令, 其间 PIPELINE不需要停顿。这种方案同样需要 增加硬件逻辑, 对于循环嵌套, 需要的硬件资源更多, 而且如果循环体指 令条数小于 N ( N表示从 MEMORY取指令需要的周数)时, 这种方案不具 有适应性。 发明内容
有鉴于此, 本发明的主要目的在于提供一种跳转指令编码的方法和系 统, 以简化硬件逻辑, 提高实现的可靠性和性能。
为达到上述目的, 本发明的技术方案是这样实现的:
一种跳转指令编码的方法, 包括:
在执行循环时, 相对于操作周期, 提前到内存中获取循环开始指令; 当循环体最后一条指令执行完时, 完成循环开始指令的获取。
其中, 所述相对于操作周期, 提前到内存中获取循环开始指令的方法 为:
将循环体最后一条指令的标志位提前 N 周, 并生成需要的执行文件 exe.bin;
所述 N周对应于循环开始指令至循环体最后一条指令的操作周期。 其中, 所述完成循环开始指令的获取的过程包括:
所述 exe.bin指令在被执行时,硬件提前 N周发现循环体最后一条指令, 从而提前 N周从存储器 MEMORY获取该指令。
其中, 该方法还包括:
涉及执行循环的指令流水线 PIPELINE在不需要等待的情况下继续工 作。
其中, 所述方法通过汇编译器 ASS汇编译器实现。
一种跳转指令编码的系统, 包括操作周期提前单元、 循环开始指令获 取单元; 其中,
所述操作周期提前单元, 用于在执行循环时, 相对于操作周期, 确认 要提前到内存中获取循环开始指令;
所述循环开始指令获取单元, 用于在循环体最后一条指令执行完时, 完成循环开始指令的获取。 其中, 所述操作周期提前单元, 相对于操作周期, 提前到内存中获取 循环开始指令时, 用于:
将循环体最后一条指令的标志位提前 N 周, 并生成需要的执行文件 exe.bin;
所述 N周对应于循环开始指令至循环体最后一条指令的操作周期。 其中, 所述循环开始指令获取单元完成循环开始指令的获取时, 用于: 触发所述 exe.bin指令的执行,使硬件提前 N周发现循环体最后一条指 令, 从而提前 N周从 MEMORY获取该指令。
其中, 该系统还包括涉及执行循环的 PIPELINE, 用于在不需要等待的 情况下继续工作。
其中, 所述系统设置于 ASS中。
本发明跳转指令编码的方法和系统, 通过汇编译器(ASS )汇编编码的 预处理,从软件的层面简化了硬件逻辑,保证不需要去添加额外的 BUFFER 和计数器等逻辑来解决循环折回时流水线需要停顿的问题, 省去了很多的 硬件逻辑和硬件资源, 节省了面积和成本, 使得实现简单, 也提高了实现 方案的可靠性。 附图说明
图 1为本发明实施例的跳转指令编码流程简图;
图 2为本发明实施例的跳转指令编码系统图。 具体实施方式
由于在循环折回时从 MEMORY取 LOOP— BEGIN的指令需要消耗 N个 周期, 如果在等到 LOOP— END指令时才开始去 MEMORY取指令, 那么流 水线必须停顿 N个周期, 性能的损失很大。 为了避免这种损失, 可以在硬 件中提前 N个周期去 MEMORY取指令, 这样当 LOOP END执行完时, LOOP— BEGIN的指令也已经从 MEMORY取回来了, 使得 PIPELINE不需 要停顿。
上述方案可以通过汇编译器 (ASS ) 来实现。 ASS 能够将程序员写的 汇编指令编码成二进制或十六进制。 ASS在编码时, 可以将 LOOP— END的 标志位提前 N周 (该周期即 LOOP— BEGIN至 LOOP— END的操作周期), 生成需要的执行文件 exe.bin。 exe.bin指令在被执行时, 硬件就能提前 N周 看到 LOOP— END, 从而提前 N周从 MEMORY取该指令, 消除 PIPELINE 的停顿, 实现高性能。
所编写的汇编代码指令序列可以例举为:
InstO;
Instl
Inst2 LOOP START;
Inst3
Inst4
Inst5
Inst6; LOOP— END
其中, Inst2至 Inst6表示循环体指令。
ASS汇编后的机器码序列可以例举为:
Figure imgf000006_0001
20000000
30000000
40000002
50000000
60000000 其中, LOOP— MARKER (循环标签) 占用每条指令的低 2bit, 1 表示 LOOP— BEGIN, 2表示 LOOP— END , 0表示没有循环标签。 上述表示方式 只是一种编码方式之一, 并不是唯一的。
可见,在编码时将 LOOP— END提前了 2条指令。 Inst6对应 LOOP— END , 在机器码中被编码在了 Inst4上, 提前了 2条指令。 这个过程是编译器自动 完成的, 程序员写的循环体依旧是 Inst2-Inst6。 当机器码指令被执行时, 当 执行到 Inst4时, 就看到了 LOOP— END (即机器码低 2bit的值为 2 ), 这时 就开始从 MEMORY取 Inst2 指令。 当执行到 Inst6 时, Inst2 就已经从 MEMORY取出来了 (这里假定从 MEMORY取指令需要 2周), 指令就又 从 Inst2开始顺序执行。 当执行到 Inst4时, 又去 MEMORY取 Inst2 , 这样 就形成了 Inst2至 Inst6的循环执行, 直到循环退出。
由上述技术描述可见,本发明跳转指令编码的操作思路可以表示如图 1 所示的流程, 该流程包括以下步骤:
步骤 110: 在执行循环时, 相对于操作周期, 提前到内存中获取循环开 始指令。
步骤 120:当循环体最后一条指令执行完时,完成循环开始指令的获取。 需要说明的是, 所述相对于操作周期, 提前到内存中获取循环开始指 令的方法可以为: 将循环体最后一条指令的标志位提前 N周, 并生成需要 的执行文件 exe.bin; 所述 N周对应于循环开始指令至循环体最后一条指令 的操作周期。
并且, 所述完成循环开始指令的获取的过程可以包括: 所述 exe.bin指 令在被执行时, 硬件提前 N周发现循环体最后一条指令, 从而提前 N周从 MEMORY获取该指令。 还可以由涉及执行循环的 PIPELINE在不需要等待 的情况下继续工作。
另外, 本发明可以通过 ASS汇编译器实现。 为了保证上述技术描述以及操作思路能够顺利实现, 可以进行如图 2 所示的设置。 参见图 2, 图 2为本发明实施例的跳转指令编码系统图, 该系 统包括相连的操作周期提前单元、 循环开始指令获取单元。
在实际应用时, 操作周期提前单元能够在执行循环时, 相对于操作周 期, 确认要提前到内存中获取循环开始指令; 循环开始指令获取单元则能 够在循环体最后一条指令执行完时, 完成循环开始指令的获取。
需要说明的是, 所述操作周期提前单元, 相对于操作周期, 提前到内 存中获取循环开始指令时, 可以用于: 将循环体最后一条指令的标志位提 前 N周, 并生成需要的执行文件 exe.bin; 所述 N周对应于循环开始指令至 循环体最后一条指令的操作周期。
并且, 所述循环开始指令获取单元完成循环开始指令的获取时, 可以 用于: 触发所述 exe.bin指令的执行, 使硬件提前 N周发现循环体最后一条 指令, 从而提前 N周从 MEMORY获取该指令。 还可以进一步由涉及执行 循环的 PIPELINE在不需要等待的情况下继续工作。
另外, 本发明所述系统可以设置于 ASS中。
综上所述可见, 无论是方法还是系统, 本发明跳转指令编码的方法和 系统, 通过 ASS汇编编码的预处理, 从软件的层面简化了硬件逻辑, 保证 不需要去添加额外的 BUFFER和计数器等逻辑来解决循环折回时流水线需 要停顿的问题, 省去了很多的硬件逻辑和硬件资源, 节省了面积和成本, 使得实现简单, 也提高了实现方案的可靠性。
以上所述, 仅为本发明的较佳实施例而已, 并非用于限定本发明的保 护范围。

Claims

权利要求书
1、 一种跳转指令编码的方法, 包括:
在执行循环时, 相对于操作周期, 提前到内存中获取循环开始指令; 当循环体最后一条指令执行完时, 完成循环开始指令的获取。
2、 根据权利要求 1所述的方法, 其中, 所述相对于操作周期, 提前到 内存中获取循环开始指令的方法为:
将循环体最后一条指令的标志位提前 N 周, 并生成需要的执行文件 exe.bin;
所述 N周对应于循环开始指令至循环体最后一条指令的操作周期。
3、 根据权利要求 2所述的方法, 其中, 所述完成循环开始指令的获取 的过程包括:
所述 exe.bin指令在被执行时,硬件提前 N周发现循环体最后一条指令, 从而提前 N周从存储器 MEMORY获取该指令。
4、 根据权利要求 1至 3任一项所述的方法, 其中, 该方法还包括: 涉及执行循环的指令流水线 PIPELINE在不需要等待的情况下继续工 作。
5、 根据权利要求 1至 3任一项所述的方法, 其中, 所述方法通过汇编 译器 ASS汇编译器实现。
6、 一种跳转指令编码的系统, 包括操作周期提前单元、 循环开始指令 获取单元; 其中,
所述操作周期提前单元, 用于在执行循环时, 相对于操作周期, 确认 要提前到内存中获取循环开始指令;
所述循环开始指令获取单元, 用于在循环体最后一条指令执行完时, 完成循环开始指令的获取。
7、 根据权利要求 6所述的系统, 其中, 所述操作周期提前单元, 相对 于操作周期, 提前到内存中获取循环开始指令时, 用于:
将循环体最后一条指令的标志位提前 N 周, 并生成需要的执行文件 exe.bin;
所述 N周对应于循环开始指令至循环体最后一条指令的操作周期。
8、 根据权利要求 7所述的系统, 其中, 所述循环开始指令获取单元完 成循环开始指令的获取时, 用于:
触发所述 exe.bin指令的执行,使硬件提前 N周发现循环体最后一条指 令, 从而提前 N周从 MEMORY获取该指令。
9、 根据权利要求 6至 8任一项所述的系统, 其中, 该系统还包括涉及 执行循环的 PIPELINE, 用于在不需要等待的情况下继续工作。
10、 根据权利要求 6至 8任一项所述的系统, 其中, 所述系统设置于 ASS中。
PCT/CN2011/078846 2011-04-01 2011-08-24 一种跳转指令编码的方法和系统 WO2012129886A1 (zh)

Priority Applications (1)

Application Number Priority Date Filing Date Title
EP11862348.7A EP2696279A4 (en) 2011-04-01 2011-08-24 METHOD AND SYSTEM FOR CODING JUMPING INSTRUCTIONS

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201110081939.1 2011-04-01
CN201110081939.1A CN102736894B (zh) 2011-04-01 2011-04-01 一种跳转指令编码的方法和系统

Publications (1)

Publication Number Publication Date
WO2012129886A1 true WO2012129886A1 (zh) 2012-10-04

Family

ID=46929380

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2011/078846 WO2012129886A1 (zh) 2011-04-01 2011-08-24 一种跳转指令编码的方法和系统

Country Status (3)

Country Link
EP (1) EP2696279A4 (zh)
CN (1) CN102736894B (zh)
WO (1) WO2012129886A1 (zh)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112445520B (zh) * 2019-08-27 2022-11-15 无锡江南计算技术研究所 针对循环内条件转移指令的转移预测优化方法

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6564313B1 (en) * 2001-12-20 2003-05-13 Lsi Logic Corporation System and method for efficient instruction prefetching based on loop periods
CN1255723C (zh) * 1999-05-28 2006-05-10 英特尔公司 减少软件数据预取的开销的设备和方法
CN1928810A (zh) * 2005-09-09 2007-03-14 上海采微电子科技有限公司 带有循环跳转预测单元的微处理器

Family Cites Families (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5742804A (en) * 1996-07-24 1998-04-21 Institute For The Development Of Emerging Architectures, L.L.C. Instruction prefetch mechanism utilizing a branch predict instruction
US6523110B1 (en) * 1999-07-23 2003-02-18 International Business Machines Corporation Decoupled fetch-execute engine with static branch prediction support
JP4134179B2 (ja) * 2005-02-04 2008-08-13 株式会社ソニー・コンピュータエンタテインメント ソフトウエアによる動的予測方法および装置

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1255723C (zh) * 1999-05-28 2006-05-10 英特尔公司 减少软件数据预取的开销的设备和方法
US6564313B1 (en) * 2001-12-20 2003-05-13 Lsi Logic Corporation System and method for efficient instruction prefetching based on loop periods
CN1928810A (zh) * 2005-09-09 2007-03-14 上海采微电子科技有限公司 带有循环跳转预测单元的微处理器

Also Published As

Publication number Publication date
EP2696279A1 (en) 2014-02-12
EP2696279A4 (en) 2014-03-19
CN102736894B (zh) 2017-10-24
CN102736894A (zh) 2012-10-17

Similar Documents

Publication Publication Date Title
JP3702813B2 (ja) マルチスレッド実行方法及び並列プロセッサシステム
CN103729288B (zh) 一种嵌入式多核环境下应用程序的调试方法
KR101486025B1 (ko) 프로세서에서의 쓰레드 스케쥴링
JP5889986B2 (ja) 実行された命令の結果を選択的にコミットするためのシステムおよび方法
US9558000B2 (en) Multithreading using an ordered list of hardware contexts
JP6267513B2 (ja) 同一のデータ条件に依存する、隣接する命令の実行の制御
KR101748506B1 (ko) 실시간 명령어 추적 프로세서들, 방법들 및 시스템들
US8527969B2 (en) Systems, methods, and computer programs for dynamic binary translation in an interpreter
US11366669B2 (en) Apparatus for preventing rescheduling of a paused thread based on instruction classification
JP3669884B2 (ja) 処理装置
GB2374694A (en) Storing instruction set information
US8276128B2 (en) Systems, methods, and computer programs for dynamic binary translation in a master control program interpreter
US20090089555A1 (en) Methods and apparatus for executing or converting real-time instructions
US20140129805A1 (en) Execution pipeline power reduction
JP2014099192A (ja) 効率的な割込み復帰アドレス保存メカニズム
WO2012129886A1 (zh) 一种跳转指令编码的方法和系统
US20170060582A1 (en) Arbitrary instruction execution from context memory
US20070186084A1 (en) Circuit and method for loop control
US8896610B2 (en) Error recovery operations for a hardware accelerator
CN114218067A (zh) 一种异构众核软件调试装置及调试方法
CN116841614B (zh) 乱序访存机制下的顺序向量调度方法
EP2605134B1 (en) Systems, methods, and computer programs for dynamic binary translation in an interpreter
US20100153688A1 (en) Apparatus and method for data process
JP2009054032A (ja) 並列プロセッサ
KR101033591B1 (ko) 임베디드 시스템의 디버깅 방법

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 11862348

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

WWE Wipo information: entry into national phase

Ref document number: 2011862348

Country of ref document: EP