WO2000068784A1 - Data processing device, method for executing load or store instructions and method for compiling programs - Google Patents
Data processing device, method for executing load or store instructions and method for compiling programs Download PDFInfo
- Publication number
- WO2000068784A1 WO2000068784A1 PCT/EP2000/003781 EP0003781W WO0068784A1 WO 2000068784 A1 WO2000068784 A1 WO 2000068784A1 EP 0003781 W EP0003781 W EP 0003781W WO 0068784 A1 WO0068784 A1 WO 0068784A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- instruction
- memory
- pipeline
- memory access
- data
- Prior art date
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F9/00—Arrangements for program control, e.g. control units
- G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
- G06F9/30—Arrangements for executing machine instructions, e.g. instruction decode
- G06F9/38—Concurrent instruction execution, e.g. pipeline, look ahead
- G06F9/3824—Operand accessing
- G06F9/3834—Maintaining memory consistency
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F9/00—Arrangements for program control, e.g. control units
- G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
- G06F9/30—Arrangements for executing machine instructions, e.g. instruction decode
- G06F9/38—Concurrent instruction execution, e.g. pipeline, look ahead
- G06F9/3867—Concurrent instruction execution, e.g. pipeline, look ahead using instruction pipelines
Definitions
- the invention relates to a data processing device that has a forward load instruction that can be taken into execution before a store instruction that stores the data loaded by the forward load instruction.
- the invention also relates to a method of executing programs using a forward load instruction and to a method of generating machine code programs containing forward load instructions.
- load-store dependency where a first instruction stores data to memory and a second instruction loads da a from memory.
- the sequence of executing the load instruction and the store instruction cannot normally be changed without affecting the results of the program.
- This is a problem that is similar to the problems that occur in cache prefetching, which can be corrected by updating data in the cache when a store occurs.
- the article by Chen et al. applies this cache technique also to registers in the processor. Upon encountering a forward load instruction, the processor prefetches data from memory into a register. The load address used by the forward load instruction is saved after it has been used to load data.
- a store instruction when a store instruction is executed, the store address of the store instruction is compared with the addresses used to prefetch data into each register. If the load and store addresses address the same data, the prefetched data in the relevant register is replaced by the store data that is stored by the store instruction. The data is replaced from the time that the store instruction is completed.
- a register loaded with a forward load instruction always contains data that corresponds to the data that is actually in memory at the load address, no matter when the forward load instruction is executed.
- a "commit" instruction is added to prevent store instructions after that location from causing a substitution with store data. As a result the forward load instruction can be moved freely through the program, past any store instructions, without affecting the result of the program.
- the data processor device is described in Claim 1.
- compensation of the effect of out of order execution of memory access instructions is incorporated in pipelined execution of the memory access instruction.
- the memory address needs to be compared only with one or more memory addresses present in one or more of the stages downstream in the pipeline, and not with memory addresses for all available registers.
- At a pipeline stage that makes irreversible changes to memory or register content such changes are suppressed or data obtained from a different stage is substituted if the memory addresses the same memory location, so as to obtain the same effect as if the memory access instructions had been executed in the original order.
- substitution of load data is incorporated in pipelined execution of a forward load instruction before a store instruction that may affect a memory location from which the load instruction loads data.
- the loaded data or, if appropriate due to a store and load address match the stored data is written back into the result register of the forward load instruction.
- suppression or substitution of store data is incorporated in pipelined execution of a first store instruction executed after a forwarded store instruction that may affect a memory location from which the first store instruction stores data.
- the data of the first store instruction if appropriate due to an address match, no data or substituted data is written to memory.
- an instruction indicates for which pipeline stages relative to the pipeline stage that executes the instruction the effect of out of order execution is to be compensated.
- the instruction that would execute incorrectly due to out of order execution or the other instruction that causes this incorrect execution or both may be used to indicate that compensation is necessary for a certain pipeline distance. This makes it possible to process either a load/store instruction that involve forward loading/storing or a load/store instruction that does not involve such forward loading/storing alternatively at the same point in the pipeline.
- a compiler place either a memory access instruction that has been moved out of order, or a memory access instruction that has not been so moved at the same place in the program and the compiler can select the appropriate type for each such memory access instruction to indicate whether it is necessary to provide for correction of the effect of movement with respect to indicated other memory instructions at selected distances relative to the memory access instruction.
- Figure 1 shows a data processing architecture
- Figure 2 shows a pipeline structure
- Figure 3 shows a pipeline structure
- Figure 3 A shows a pipeline structure
- Figure 4 shows a pipeline structure
- Figure 5 shows a flow chart for generating a program of machine instructions.
- Figure 1 shows a data processing architecture.
- the architecture contains an instruction issue unit 10, a number of functional units 12a-c, a register file 14 and a memory 16.
- the instruction issue unit is capable of issuing a number of instructions in parallel to different functional units 12a-c.
- three functional units 12a-c are shown more or fewer functional units may be present without deviating from the invention.
- Different types of functional units such as ALU's (Arithmetic Logic Units), branch control units, shift units, vector processing units, DSP's (Digital Signal Processor units etc. may be provided.
- One of the functional units 12a in figure 1 is shown as a memory access control unit 12a, coupled to the memory 16.
- the instruction issue unit 10 has outputs coupled to the functional units 12a-c and read ports of the register file 14.
- the functional units 12a-c have inputs and outputs coupled to read ports and write ports respectively of the register file 14.
- the processing architecture of figure 1 operates in successive instruction cycles.
- the instruction issue unit 10 issues one or more instructions in parallel to the functional units 12a-c.
- the instruction issue unit 10 operates in a conventional way using an instruction memory (not shown) a program counter (not shown) etc.
- Each instruction indicates an operation code, a number of registers in the register file 14 and optionally some literal data.
- the indication of registers serves to identify registers that contain source operands of the instruction and a result register where the result of the instruction must be stored.
- the instruction issue unit 10 uses the indication of registers that contain source operands to address read ports of the register file.
- the instruction issue unit 10 supplies the operation code, the indication of the result register and optionally the literal data to the functional unit 12a-c for which that particular instruction is intended.
- the functional unit 12a-c uses data from the read ports of the register file 14, the operation code and optionally the literal data to process the instruction. If necessary, the functional unit 12a-c writes back a result to a write port of the register file 14.
- the memory access control unit 12a operates as one of the functional units. As shown in figure 1, there is only one memory access control unit 12a that accesses memory 16, but more than one such memory access control unit may be present among the functional units 12a-c.
- the memory access control unit 12a receives an operation code, which indicates whether a load or a store instruction should be executed, how the memory address should be calculated and what data length should loaded or stored.
- the memory access control unit 12a uses this information to supply an address to the memory 16, together with a read/write control signal and any other control signal as appropriate. Write data is also supplied to memory 16 in case of a store instruction.
- the memory access control unit 12a reads from memory or writes data to memory, dependent on whether a store or load instruction is executed.
- Write data is usually derived from operand data received from a read port of the register file 14.
- Read data is usually written to a write port of the register file 14.
- the sequence in which store and load instructions are executed cannot be changed without running the risk of affecting the outcome of program execution. This is because storing data to memory may affect the result of subsequent load instructions if these instructions load from the same location where the store instruction stores data. Also, the sequence in which two store instructions are executed may affect the content of memory after the two store instructions if these store instructions address the same memory location.
- the invention has the object to make it possible to change the sequence of executing load and store instructions without affecting the result of the program by this changing and/or to change the sequence of executing several load instructions without affecting the result of the program.
- Figure 2 shows a pipeline structure of a memory access control unit 12a, emphasizing only those parts that are used to forward data from store instructions.
- Figure 2 shows, at the beginning of the pipeline, an input register 200 that is the interface to the instruction issue unit 10.
- the register file 260 is shown (indicated at this position with a different reference number for the sake of clarity).
- a number of pipeline registers 210, 220, 230, 240 are shown, which mark the separations between different stages of the pipeline.
- an instruction decoder 202 is provided between the input register 200 and a first stage pipeline register 210.
- the first stage pipeline register 210 has inputs coupled to an output of the instruction decoder 202 and the output of one or more read ports of the register file 14.
- a second stage pipeline register 220 is coupled to the first stage pipeline register 210, partly via an address computation unit 212.
- the second stage pipeline register 220 has an address output, a data output and a control output coupled to a memory interface port.
- a third stage pipeline register 230 has inputs coupled to the outputs of the second stage pipeline register 220.
- a number of further pipeline registers may be present in between the second stage pipeline register 220 and the third stage pipeline register 230.
- a fourth stage pipeline register 240 follows the third stage pipeline register 230.
- the third stage pipeline register 230 has outputs coupled directly to the fourth stage pipeline register 240 and to an input of a first comparator 232.
- the first comparator 232 has a further input coupled to the address output of the second stage pipeline register 220.
- the first comparator has an output coupled to a control input of a first multiplexer 234.
- the first multiplexer has inputs coupled to a data output of the memory (not shown) and to the data output of the second stage pipeline register 220.
- An output of the first multiplexer 234 is coupled to the fourth stage pipeline register 240.
- a fifth stage pipeline register 250 follows the fourth stage pipeline register 240.
- the fourth stage pipeline register 240 has outputs coupled directly to the fifth stage pipeline register 250 and to a first input a second comparator 244.
- a second input of the second comparator 244 is coupled to the address output of the second stage pipeline register 220.
- An output of the second comparator 244 is coupled to a control input of a second multiplexer 246.
- a data output of the fourth stage pipeline register 240 is coupled to a first data input of the multiplexer 246 via a first data handling circuit 242a.
- a data output of the second stage pipeline register 220 is coupled to a second data input of the multiplexer 246 via a second data handling circuit 242b.
- the second multiplexer 246 has an output coupled to the fifth stage pipeline register 250.
- the fifth stage pipeline register 250 has outputs coupled to a write port of the register file 260.
- the pipeline shown in figure 2 processes memory access instructions, such as store, normal load and forward load.
- An instruction is processed in steps. Each step is processed in a different stage of the pipeline in a different instruction cycle. Different steps of different instructions can be processed in parallel.
- a preceding stage may process another step for a subsequent instruction.
- each stage may be processing a step for a different instruction.
- the decoder 202 decodes the instruction and generates a signal to indicate the result register; operand data is supplied from a read port or read ports of the register file
- a memory address is computed from the operands and possible further data such as an offset included in the load instruction, or the content of index registers in the memory access control unit 12a
- the memory address is supplied to the memory, with a read control signal - in a fourth step, data is received from memory.
- the fourth step may be executed in the instruction cycle directly following the instruction cycle in which the third step is executed, but the fourth step may also be executed one or more instruction cycles later in case the memory is too slow to provide data sooner.
- the data required by the load instruction is obtained by handling the data received from memory
- the extracted data is written into the register file 14 in the result register indicated by the load instruction (in this step data may also bypass the register file 14 and go directly to a functional unit that addresses the result register).
- the first and second steps are substantially the same (except that there need be no indication of a result register).
- the store instruction operand data is supplied to the memory together with the address and a write control signal. Normally, this terminates processing of the store instruction as far as the pipeline is concerned. Execution of the forward load instruction follows the same steps as the normal load instruction, but with some differences.
- the function of these differences is to replace data loaded from memory by data stored to memory by a subsequent store instruction in case that subsequent store instruction addresses the same memory location as the forward load instruction.
- the load address used by the forward load instruction is passed along the pipeline from one stage register 220, 230, 240 to another 230, 240. In a number of stages this load address is compared with the store address of a store instruction that is processed by an earlier stage. Roughly speaking, if the load and store address match, the store data of that store instruction is substituted for the load data loaded from memory when the result of the forward load instruction is written back to the register file 14 at the end of the pipeline.
- the load address from the third stage pipeline register 230 is compared to the store address from the second stage pipeline register 220 in the first comparator. If the second stage pipeline register 220 indicates that it processes a store instruction and the third stage pipeline register 230 indicates that it processes a forward store instruction, and the store address and the load address match, then the first comparator 232 generates a signal to the first multiplexer to pass the store data from the second stage pipeline register 220 instead of the data received from memory. Similarly, the load address from the fourth stage pipeline register 240 is compared to the store address from the second stage pipeline register 220 in the second comparator.
- the second comparator 244 If the second stage pipeline register 220 indicates that it processes a store instruction and the fourth stage pipeline register 240 indicates that it processes a forward store instruction, and the store address and the load address match, then the second comparator 244 generates a signal to the second multiplexer to pass store data from the second stage pipeline register 220 after handling by second handling circuit 242b instead of the data received from memory via the fourth stage pipeline register 240 and first handling circuit 242a.
- the match detected by the comparators 232, 244 should, in this case, determine not so much whether the addresses are precisely equal, but rather whether the forward load instruction and the store instruction address a same memory location. This may differ from testing for address equality if a sequence of more than one memory location is addressed with one address. Such a test can be readily realized when a length code is passed along the pipeline and to the comparators with the address to indicate the data length of the load and store operations. This comparison can be further simplified by restricting forward loads to full words at word boundaries.
- Figure 3 shows a similar pipeline as figure 2, but in this case showing those connections that allow a different sequence of execution of store instructions.
- store data from a later stage of the pipeline is substituted for store data from an earlier stage when data is applied to the memory from that earlier stage. This in contrast to figure 2, where store data was substituted for load data from memory at the later stage.
- the input stage 200, the first pipeline stage 210 register and the second pipeline stage register 220 are connected in the same way as in figure 2. These registers are followed by the third and fourth stage pipeline register 230, 240, which have inputs for receiving addresses from further upstream in the pipeline.
- the address output of the second stage pipeline register 220 is coupled directly to the memory (not shown) and to a first input of a three way address comparator 222.
- a second and third input of the comparator are coupled to address outputs of the third and fourth stage pipeline register 230, 240 respectively.
- the address comparator furthermore has control inputs coupled to the second, third and fourth stage pipeline register 220, 230, 240.
- the address comparator 222 has an output coupled to a control input of the memory.
- the second stage pipeline register contains a first store instruction
- the address and data of that store instruction are normally passed to memory and a "write" control signal is sent to memory.
- the comparator 222 gives off a control signal that disables writing to memory.
- the data stored by one of the further store instructions is not overwritten by the first store instruction.
- the pipeline shown in figure 3 does not do this for any store instruction.
- Write is disabled only if the need to prevent write is indicated by one of the store instructions involved, for example by the first store instruction in the second stage pipeline register.
- the pipeline accounts for store access to overlapping but not identical memory locations. This occurs if different word lengths can be used in different store instructions, or if store instructions need not address memory at addresses that are a multiple of the word length.
- the store instruction is only partially disabled (for the overlapped memory locations).
- data from a later pipeline stage is substituted in the second pipeline stage and the write goes ahead. This embodiment is shown in figure 3 A. The embodiment is similar to the embodiment of figure 3, but a substitution circuit 224 has been added. In operation data that is written to memory at the second stage 220 is passed with the address along the pipeline to the subsequent pipeline stages 230, 240. This data is supplied back from the subsequent pipeline stages 230, 240 to the second stage 220.
- the substitution circuit substitutes the data from the subsequent pipeline stage 230, 240 that has a matching address for data from the second pipeline stage 220, for those memory locations that are accessed by both store instructions.
- the data for other memory locations is taken from the second stage 220.
- the data from the second pipeline stage with substitutions from the subsequent pipeline stage is then written to memory. So for example if the second pipeline stage writes a 32 bit word and a subsequent pipeline stage contains a store instruction for writing an 8 bit byte, then the part of the 32 bit word that is to be stored in this byte is replaced by data from the subsequent pipeline stage.
- the invention is not limited to the specific embodiment of the pipeline shown in figures 2, 3 and 3 A. For example, if one wants to be able to substitute data from store operations that follow the load operation sooner, one might include additional comparators in one or more stages (not shown) before the stage that actually receives the corresponding load data addressed by the memory.
- a flag may be set and passed along the pipeline, together with the store data of the store instruction whose address matches.
- the third stage pipeline register 230 one should then select one of the data loaded from memory or the data thus passed through the pipeline dependent on the flag that is passed along the pipeline may be used to signal substitution by the data passed along the pipeline.
- the result of this selection or the data of the store instruction then in the second stage pipeline register 220, dependent on the output of the first comparator 232 is the passed to the next pipeline stage.
- the store addresses may be passed along the pipeline for comparison at a later pipeline stage.
- a multiplexer and possibly even a comparator may be inserted between the fifth stage pipeline register 250 and the write port of the register file 260.
- store instructions that follow the forward load at a greater distance can be accounted for.
- the last pipeline stage 242, 250 may even be deleted for straightforward word load instructions.
- Figure 4 shows a further pipeline for a memory access unit. In comparison with the pipeline of figure 2, this pipeline has been simplified and speeded up.
- the pipeline of figure 4 has three pipeline registers 310, 320, 330.
- a first stage pipeline register 310 receives instruction information via a decoder 301 and operand data from the register file 14.
- the first stage pipeline register 310 outputs control, address and (if necessary) data to the memory interface.
- a second stage pipeline register 320 receives instruction information and address data from the first stage pipeline register 310.
- the second stage pipeline register 320 outputs address data and control information to a first comparator 322 which also receives address data and control information from the first stage pipeline register 310.
- An output of the first comparator is coupled to a control input of a first multiplexer 324, which has inputs coupled to a memory data output and to the data output of the first stage pipeline register 310.
- a third stage pipeline register 330 receives instruction information and address data from the second stage pipeline register and data from the first multiplexer 324.
- the third stage pipeline register 330 outputs address data and instruction information to a second comparator 332, which also receives address a data and control information from the first stage pipeline register 310.
- the output of comparator 332 is coupled to a control input of a second multiplexer 334.
- the second multiplexer 334 has inputs coupled to a data output of the third stage pipeline register 330 and to a data output of the first stage pipeline register 310.
- An output of the second multiplexer 334 is coupled to a write port of the register file (not shown).
- the memory access control unit of figure 4 corresponds to the memory access control unit of figure 2, with the deletion of a number of functions, such as the address computation stage (210, 212): instead addresses are taken directly from operands. Also the data extraction stage (242, 250) is deleted: only full words are used.
- the memory access control unit of figure 4 works similar to the memory access control unit of figure 2, but faster and with less functionality.
- the memory access control units of figures 2 and 3 represent illustrations of possible variations in the design of the pipeline. Of course any combination of such variations can be used. Which combination is to be preferred depends on considerations such as whether pipeline steps can be executed sufficiently quickly and hardware cost.
- the A instruction uses the result register of the LD instruction. It is possible, but not certain, that the ST instruction stores data at the location where the LD instruction loads data. With normal load and store instructions, the pipeline of figure 2 would require that the program performing this sequence looks something like
- the space between the load instruction FLD can be filled with the store instruction and with instructions XI, X2 that produce results needed by the store instruction. But this is only possible with a forward load instruction FLD because the result of the load instruction may depend on the store instruction.
- the processor provides for an even more refined control over whether new data stored by the store instruction is loaded by the load instruction or whether "old" data from memory is loaded by the load instruction. For example, it may be desirable to provide for an instruction sequence like
- Such selective forwarding makes it possible to schedule store instructions that should not affect the load FLD, like ST2, in an earlier instruction cycle. In this way the instruction cycles can be used more efficiently, and instructions that depend on the store instruction ST2 can be scheduled sooner.
- the selective forwarding is realized by providing different versions of the load and/or store instructions.
- Each version defines one or more relative positions of the store and load in the pipeline. Only for instructions at those relative positions will forwarding of data from the store to the load be effected.
- Different versions define different sets of relative positions.
- the forward load instruction there may be two versions of the forward load instruction.
- One version loads data forwarded from store instructions that start execution at N and N+l instruction cycles after the forward load instruction.
- Another version loads data forwarded from a store instruction that start N instruction cycles after the forward load instruction, but not from a store instruction that starts N+l instruction cycles after the forward load instruction.
- One version forwards data to forward load instructions that start execution at N and N+l instruction cycles before this store instruction.
- Another version forwards to a forward load instruction that start N instruction cycles before this store instruction, but not to a forward load instruction that starts N+l instruction cycles before this store instruction.
- both store and load instructions can indicate whether forwarding should be used, forwarding taking place only when both instructions so indicate (or, in an alternative embodiment, when at least one so indicates).
- This provides for an even more refined control of forwarding, which is useful if more than one load and/or store instruction can start execution in the same instruction cycle, for example in a VLIW processor.
- the processor can implement such instructions by passing a code that identifies the version of the load or store instruction along the pipeline. This code is supplied as control information to the comparators. Each comparator controls substitution from a store to a load for a specific relative position of this store and load in the pipeline.
- the comparator is arranged to signal a need to substitute forwarded data only if the load and store address match and the code derived from the load and/or store instruction at the relevant stage of the pipeline indicates that the version of the load and/or store requires such a substitution.
- FIG. 5 shows a flowchart for generating a scheduled set of machine instructions for a processor with forward stores.
- a compiler or scheduler (herein both generally referred to as scheduler) typically selects the sequence of execution of instructions.
- a program in a high level language is converted into a set of machine instructions and information (such as instruction sequence) that represents a data dependency graph, which indicates data dependence relations between pairs of instructions (data user-producer relations between instructions).
- a second step 42 the instructions are scheduled in respective instruction cycles.
- each instruction cycle may even have more than one scheduled instruction.
- it must be ensured that dependent instructions will not be executed before data is available from instructions on which they depend.
- not more instructions may be scheduled together than the hardware can execute. Otherwise there is freedom in choosing the sequence in which the instructions are scheduled. This freedom is used to search for an assignment of instructions to instruction cycles that minimize the number of instruction cycles that is needed to execute the program for typical input data.
- the scheduler can schedule a load instruction in an instruction cycle after the store instruction cycle containing the store instruction that may affect the load data, but also in an earlier instruction cycle before such a store instruction cycle, if this earlier instruction cycle is not more than a pipeline delay before this store instruction cycle.
- the load store combination is not ambiguous, i.e. when it is known from the program that they cannot address the same memory location, then the load instruction can be moved even further ahead.
- the second step 32 keeps information indicating whether the load instruction depends on the store instruction (e.g. corresponds to a statement in the original program after the store instruction) or not.
- a third step 43 once the scheduler has decided the relative scheduling of load and store instructions, it subsequently selects the required version of each load and/or store instructions (e.g. forward load in general, or forward load from store instructions N cycles later) dependent on the required dependency indicated by the second step 32 and the relative position where the load and store instructions are scheduled.
- the required version of each load and/or store instructions e.g. forward load in general, or forward load from store instructions N cycles later
Abstract
Description
Claims
Priority Applications (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
EP00929437A EP1093611A1 (en) | 1999-05-06 | 2000-04-20 | Data processing device, method for executing load or store instructions and method for compiling programs |
JP2000616501A JP2002544588A (en) | 1999-05-06 | 2000-04-20 | Data processing apparatus, method of executing load or store instruction, and method of compiling program |
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
EP99201425.8 | 1999-05-06 | ||
EP99201425 | 1999-05-06 |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2000068784A1 true WO2000068784A1 (en) | 2000-11-16 |
Family
ID=8240186
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/EP2000/003781 WO2000068784A1 (en) | 1999-05-06 | 2000-04-20 | Data processing device, method for executing load or store instructions and method for compiling programs |
Country Status (4)
Country | Link |
---|---|
US (1) | US6615333B1 (en) |
EP (1) | EP1093611A1 (en) |
JP (1) | JP2002544588A (en) |
WO (1) | WO2000068784A1 (en) |
Families Citing this family (57)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US7962716B2 (en) | 2001-03-22 | 2011-06-14 | Qst Holdings, Inc. | Adaptive integrated circuitry with heterogeneous and reconfigurable matrices of diverse and adaptive computational units having fixed, application specific computational elements |
US7249242B2 (en) | 2002-10-28 | 2007-07-24 | Nvidia Corporation | Input pipeline registers for a node in an adaptive computing engine |
US6836839B2 (en) | 2001-03-22 | 2004-12-28 | Quicksilver Technology, Inc. | Adaptive integrated circuitry with heterogeneous and reconfigurable matrices of diverse and adaptive computational units having fixed, application specific computational elements |
US7752419B1 (en) | 2001-03-22 | 2010-07-06 | Qst Holdings, Llc | Method and system for managing hardware resources to implement system functions using an adaptive computing architecture |
US7653710B2 (en) | 2002-06-25 | 2010-01-26 | Qst Holdings, Llc. | Hardware task manager |
US6577678B2 (en) | 2001-05-08 | 2003-06-10 | Quicksilver Technology | Method and system for reconfigurable channel coding |
US7046635B2 (en) | 2001-11-28 | 2006-05-16 | Quicksilver Technology, Inc. | System for authorizing functionality in adaptable hardware devices |
US6986021B2 (en) | 2001-11-30 | 2006-01-10 | Quick Silver Technology, Inc. | Apparatus, method, system and executable module for configuration and operation of adaptive integrated circuitry having fixed, application specific computational elements |
US8412915B2 (en) | 2001-11-30 | 2013-04-02 | Altera Corporation | Apparatus, system and method for configuration of adaptive integrated circuitry having heterogeneous computational elements |
US7215701B2 (en) | 2001-12-12 | 2007-05-08 | Sharad Sambhwani | Low I/O bandwidth method and system for implementing detection and identification of scrambling codes |
US7403981B2 (en) | 2002-01-04 | 2008-07-22 | Quicksilver Technology, Inc. | Apparatus and method for adaptive multimedia reception and transmission in communication environments |
US7328414B1 (en) | 2003-05-13 | 2008-02-05 | Qst Holdings, Llc | Method and system for creating and programming an adaptive computing engine |
US7660984B1 (en) | 2003-05-13 | 2010-02-09 | Quicksilver Technology | Method and system for achieving individualized protected space in an operating system |
US8108656B2 (en) | 2002-08-29 | 2012-01-31 | Qst Holdings, Llc | Task definition for specifying resource requirements |
GB2424503B (en) * | 2002-09-17 | 2007-06-20 | Micron Technology Inc | An active memory device |
US7937591B1 (en) | 2002-10-25 | 2011-05-03 | Qst Holdings, Llc | Method and system for providing a device which can be adapted on an ongoing basis |
US8276135B2 (en) | 2002-11-07 | 2012-09-25 | Qst Holdings Llc | Profiling of software and circuit designs utilizing data operation analyses |
US7225301B2 (en) | 2002-11-22 | 2007-05-29 | Quicksilver Technologies | External memory controller node |
US20040154010A1 (en) * | 2003-01-31 | 2004-08-05 | Pedro Marcuello | Control-quasi-independent-points guided speculative multithreading |
US7657880B2 (en) * | 2003-01-31 | 2010-02-02 | Intel Corporation | Safe store for speculative helper threads |
US20060259742A1 (en) * | 2005-05-16 | 2006-11-16 | Infineon Technologies North America Corp. | Controlling out of order execution pipelines using pipeline skew parameters |
US20060277425A1 (en) * | 2005-06-07 | 2006-12-07 | Renno Erik K | System and method for power saving in pipelined microprocessors |
GB2462237B (en) * | 2007-06-04 | 2011-11-23 | Fujitsu Ltd | Packet switching system |
EP2526494B1 (en) | 2010-01-21 | 2020-01-15 | SVIRAL, Inc. | A method and apparatus for a general-purpose, multiple-core system for implementing stream-based computations |
WO2013100783A1 (en) | 2011-12-29 | 2013-07-04 | Intel Corporation | Method and system for control signalling in a data path module |
US10331583B2 (en) | 2013-09-26 | 2019-06-25 | Intel Corporation | Executing distributed memory operations using processing elements connected by distributed channels |
CN104978282B (en) * | 2014-04-04 | 2019-10-01 | 上海芯豪微电子有限公司 | A kind of caching system and method |
JP6422381B2 (en) * | 2015-03-18 | 2018-11-14 | ルネサスエレクトロニクス株式会社 | Processor, program code conversion device and software |
US10416999B2 (en) | 2016-12-30 | 2019-09-17 | Intel Corporation | Processors, methods, and systems with a configurable spatial accelerator |
US10474375B2 (en) * | 2016-12-30 | 2019-11-12 | Intel Corporation | Runtime address disambiguation in acceleration hardware |
US10572376B2 (en) | 2016-12-30 | 2020-02-25 | Intel Corporation | Memory ordering in acceleration hardware |
US10558575B2 (en) | 2016-12-30 | 2020-02-11 | Intel Corporation | Processors, methods, and systems with a configurable spatial accelerator |
US10515046B2 (en) | 2017-07-01 | 2019-12-24 | Intel Corporation | Processors, methods, and systems with a configurable spatial accelerator |
US10469397B2 (en) | 2017-07-01 | 2019-11-05 | Intel Corporation | Processors and methods with configurable network-based dataflow operator circuits |
US10445234B2 (en) | 2017-07-01 | 2019-10-15 | Intel Corporation | Processors, methods, and systems for a configurable spatial accelerator with transactional and replay features |
US10467183B2 (en) | 2017-07-01 | 2019-11-05 | Intel Corporation | Processors and methods for pipelined runtime services in a spatial array |
US10445451B2 (en) | 2017-07-01 | 2019-10-15 | Intel Corporation | Processors, methods, and systems for a configurable spatial accelerator with performance, correctness, and power reduction features |
US10515049B1 (en) | 2017-07-01 | 2019-12-24 | Intel Corporation | Memory circuits and methods for distributed memory hazard detection and error recovery |
US10496574B2 (en) | 2017-09-28 | 2019-12-03 | Intel Corporation | Processors, methods, and systems for a memory fence in a configurable spatial accelerator |
US11086816B2 (en) | 2017-09-28 | 2021-08-10 | Intel Corporation | Processors, methods, and systems for debugging a configurable spatial accelerator |
US10445098B2 (en) | 2017-09-30 | 2019-10-15 | Intel Corporation | Processors and methods for privileged configuration in a spatial array |
US10565134B2 (en) | 2017-12-30 | 2020-02-18 | Intel Corporation | Apparatus, methods, and systems for multicast in a configurable spatial accelerator |
US10445250B2 (en) | 2017-12-30 | 2019-10-15 | Intel Corporation | Apparatus, methods, and systems with a configurable spatial accelerator |
US10417175B2 (en) | 2017-12-30 | 2019-09-17 | Intel Corporation | Apparatus, methods, and systems for memory consistency in a configurable spatial accelerator |
US11307873B2 (en) | 2018-04-03 | 2022-04-19 | Intel Corporation | Apparatus, methods, and systems for unstructured data flow in a configurable spatial accelerator with predicate propagation and merging |
US10564980B2 (en) | 2018-04-03 | 2020-02-18 | Intel Corporation | Apparatus, methods, and systems for conditional queues in a configurable spatial accelerator |
US11200186B2 (en) | 2018-06-30 | 2021-12-14 | Intel Corporation | Apparatuses, methods, and systems for operations in a configurable spatial accelerator |
US10459866B1 (en) | 2018-06-30 | 2019-10-29 | Intel Corporation | Apparatuses, methods, and systems for integrated control and data processing in a configurable spatial accelerator |
US10853073B2 (en) | 2018-06-30 | 2020-12-01 | Intel Corporation | Apparatuses, methods, and systems for conditional operations in a configurable spatial accelerator |
US10891240B2 (en) | 2018-06-30 | 2021-01-12 | Intel Corporation | Apparatus, methods, and systems for low latency communication in a configurable spatial accelerator |
US10678724B1 (en) | 2018-12-29 | 2020-06-09 | Intel Corporation | Apparatuses, methods, and systems for in-network storage in a configurable spatial accelerator |
US11029927B2 (en) | 2019-03-30 | 2021-06-08 | Intel Corporation | Methods and apparatus to detect and annotate backedges in a dataflow graph |
US10915471B2 (en) | 2019-03-30 | 2021-02-09 | Intel Corporation | Apparatuses, methods, and systems for memory interface circuit allocation in a configurable spatial accelerator |
US10817291B2 (en) | 2019-03-30 | 2020-10-27 | Intel Corporation | Apparatuses, methods, and systems for swizzle operations in a configurable spatial accelerator |
US10965536B2 (en) | 2019-03-30 | 2021-03-30 | Intel Corporation | Methods and apparatus to insert buffers in a dataflow graph |
US11037050B2 (en) | 2019-06-29 | 2021-06-15 | Intel Corporation | Apparatuses, methods, and systems for memory interface circuit arbitration in a configurable spatial accelerator |
US11907713B2 (en) | 2019-12-28 | 2024-02-20 | Intel Corporation | Apparatuses, methods, and systems for fused operations using sign modification in a processing element of a configurable spatial accelerator |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
EP0568842A2 (en) * | 1992-05-06 | 1993-11-10 | International Business Machines Corporation | A method and apparatus for improving performance of out of sequence load operations in a computer system |
US5280597A (en) * | 1990-03-30 | 1994-01-18 | Mitsubishi Denki Kabushiki Kaisha | Pipeline processor with self timed data transfer |
Family Cites Families (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5588126A (en) * | 1993-12-30 | 1996-12-24 | Intel Corporation | Methods and apparatus for fordwarding buffered store data on an out-of-order execution computer system |
US5666506A (en) * | 1994-10-24 | 1997-09-09 | International Business Machines Corporation | Apparatus to dynamically control the out-of-order execution of load/store instructions in a processor capable of dispatchng, issuing and executing multiple instructions in a single processor cycle |
US5751946A (en) * | 1996-01-18 | 1998-05-12 | International Business Machines Corporation | Method and system for detecting bypass error conditions in a load/store unit of a superscalar processor |
US5758051A (en) * | 1996-07-30 | 1998-05-26 | International Business Machines Corporation | Method and apparatus for reordering memory operations in a processor |
US5841998A (en) * | 1996-12-31 | 1998-11-24 | Metaflow Technologies, Inc. | System and method of processing instructions for a processor |
US5931957A (en) * | 1997-03-31 | 1999-08-03 | International Business Machines Corporation | Support for out-of-order execution of loads and stores in a processor |
US6058472A (en) * | 1997-06-25 | 2000-05-02 | Sun Microsystems, Inc. | Apparatus for maintaining program correctness while allowing loads to be boosted past stores in an out-of-order machine |
-
2000
- 2000-04-20 JP JP2000616501A patent/JP2002544588A/en active Pending
- 2000-04-20 WO PCT/EP2000/003781 patent/WO2000068784A1/en not_active Application Discontinuation
- 2000-04-20 EP EP00929437A patent/EP1093611A1/en not_active Withdrawn
- 2000-05-03 US US09/563,369 patent/US6615333B1/en not_active Expired - Lifetime
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5280597A (en) * | 1990-03-30 | 1994-01-18 | Mitsubishi Denki Kabushiki Kaisha | Pipeline processor with self timed data transfer |
EP0568842A2 (en) * | 1992-05-06 | 1993-11-10 | International Business Machines Corporation | A method and apparatus for improving performance of out of sequence load operations in a computer system |
Non-Patent Citations (2)
Title |
---|
"BYPASS-LOAD INSTRUCTION", IBM TECHNICAL DISCLOSURE BULLETIN,US,IBM CORP. NEW YORK, vol. 31, no. 10, pages 276-277, XP000097378, ISSN: 0018-8689 * |
NAIR R ET AL: "EXPLOITING INSTRUCTION LEVEL PARALLELISM IN PROCESSORS BY CACHING SCHEDULED GROUPS", COMPUTER ARCHITECTURE NEWS,US,ASSOCIATION FOR COMPUTING MACHINERY, NEW YORK, vol. 25, no. 2, pages 13-25, XP000656563, ISSN: 0163-5964 * |
Also Published As
Publication number | Publication date |
---|---|
JP2002544588A (en) | 2002-12-24 |
US6615333B1 (en) | 2003-09-02 |
EP1093611A1 (en) | 2001-04-25 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US6615333B1 (en) | Data processing device, method of executing a program and method of compiling | |
US5692167A (en) | Method for verifying the correct processing of pipelined instructions including branch instructions and self-modifying code in a microprocessor | |
JP3120152B2 (en) | Computer system | |
US8443171B2 (en) | Run-time updating of prediction hint instructions | |
JP5638107B2 (en) | Data processing circuit | |
US8131982B2 (en) | Branch prediction instructions having mask values involving unloading and loading branch history data | |
US20060090063A1 (en) | Method for executing structured symbolic machine code on a microprocessor | |
JP2816248B2 (en) | Data processor | |
JPH11513164A (en) | Programmable branch prediction system and method for computer system | |
JP2834292B2 (en) | Data processor | |
JPH11212788A (en) | Data supplying device for processor | |
US6205544B1 (en) | Decomposition of instructions into branch and sequential code sections | |
KR100303712B1 (en) | Method and apparatus for an address pipeline in a pipelined machine | |
EP0742517B1 (en) | A program translating apparatus and a processor which achieve high-speed execution of subroutine branch instructions | |
US6725365B1 (en) | Branching in a computer system | |
US6871343B1 (en) | Central processing apparatus and a compile method | |
US20070174592A1 (en) | Early conditional selection of an operand | |
CN116414458A (en) | Instruction processing method and processor | |
JP7048612B2 (en) | Vector generation instruction | |
US6807624B1 (en) | Instruction control device and method therefor | |
US6016544A (en) | Apparatus and method for tracking changes in address size and for different size retranslate second instruction with an indicator from address size | |
JP2002024008A (en) | Data processor and program conversion device | |
US6604193B1 (en) | Processor in which register number translation is carried out | |
JP2006525583A (en) | Zero-overhead branching and looping in time-stationary processors | |
KR101084728B1 (en) | Pocessor supporting dynamic implied adressing mode |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
AK | Designated states |
Kind code of ref document: A1 Designated state(s): JP |
|
AL | Designated countries for regional patents |
Kind code of ref document: A1 Designated state(s): AT BE CH CY DE DK ES FI FR GB GR IE IT LU MC NL PT SE |
|
WWE | Wipo information: entry into national phase |
Ref document number: 2000929437 Country of ref document: EP |
|
ENP | Entry into the national phase |
Ref country code: JP Ref document number: 2000 616501 Kind code of ref document: A Format of ref document f/p: F |
|
121 | Ep: the epo has been informed by wipo that ep was designated in this application | ||
WWP | Wipo information: published in national office |
Ref document number: 2000929437 Country of ref document: EP |
|
WWW | Wipo information: withdrawn in national office |
Ref document number: 2000929437 Country of ref document: EP |