WO2011076602A1 - Predicting and avoiding operand-store-compare hazards in out-of-order microprocessors - Google Patents
Predicting and avoiding operand-store-compare hazards in out-of-order microprocessors Download PDFInfo
- Publication number
- WO2011076602A1 WO2011076602A1 PCT/EP2010/069496 EP2010069496W WO2011076602A1 WO 2011076602 A1 WO2011076602 A1 WO 2011076602A1 EP 2010069496 W EP2010069496 W EP 2010069496W WO 2011076602 A1 WO2011076602 A1 WO 2011076602A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- instruction
- store
- load
- hazard
- entry
- Prior art date
Links
- 238000000034 method Methods 0.000 claims abstract description 39
- 230000001419 dependent effect Effects 0.000 claims abstract description 21
- 230000002085 persistent effect Effects 0.000 claims description 21
- 230000015654 memory Effects 0.000 claims description 15
- 230000004044 response Effects 0.000 claims description 12
- 238000004364 calculation method Methods 0.000 claims description 8
- 230000003111 delayed effect Effects 0.000 claims description 8
- 238000004590 computer program Methods 0.000 claims 2
- 230000010365 information processing Effects 0.000 abstract description 13
- 238000010586 diagram Methods 0.000 description 14
- 230000007246 mechanism Effects 0.000 description 11
- 238000012545 processing Methods 0.000 description 7
- 238000013500 data storage Methods 0.000 description 5
- 230000008901 benefit Effects 0.000 description 3
- 230000005540 biological transmission Effects 0.000 description 2
- 239000000872 buffer Substances 0.000 description 2
- 230000001934 delay Effects 0.000 description 2
- 238000013461 design Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 238000011084 recovery Methods 0.000 description 2
- 230000015556 catabolic process Effects 0.000 description 1
- 238000006731 degradation reaction Methods 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 238000007726 management method Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000006855 networking Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000000717 retained effect Effects 0.000 description 1
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
- G06F9/30003—Arrangements for executing specific machine instructions
- G06F9/3004—Arrangements for executing specific machine instructions to perform operations on memory
- G06F9/30043—LOAD or STORE instructions; Clear instruction
-
- 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
- G06F9/383—Operand prefetching
-
- 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
- G06F9/3826—Bypassing or forwarding of data results, e.g. locally between pipeline stages or within a pipeline stage
-
- 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
- G06F9/3836—Instruction issuing, e.g. dynamic instruction scheduling or out of order instruction execution
-
- 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
- G06F9/3836—Instruction issuing, e.g. dynamic instruction scheduling or out of order instruction execution
- G06F9/3838—Dependency mechanisms, e.g. register scoreboarding
Definitions
- the present invention generally relates to microprocessors, and more particularly relates to managing load and store operations executed out-of-order.
- a microprocessor that is capable of issuing and executing machine instructions out of order will in general permit loads to be executed ahead of stores. This feature permits a large performance advantage provided that the load address and the store address do not both have the same physical address. In typical programs, the frequency that a load proceeds ahead of the store and that their physical address matches is low. However, since the discovery of this store violation condition is typically late in the instruction execution pipeline, the recovery penalty can be quite severe. For example, the recovery process typically involves invalidating the load instruction that caused the violation and all newer instructions in program order beyond the load instruction, and second reissuing the load instruction.
- a method for managing load and store operations executed out-of-order comprises executing at least one of a load instruction and a store instruction. A determination is made, based on the executing, that an operand store compare hazard has been encountered. An entry within an operand store compare hazard prediction table is created based on the determination. The entry comprises at least an instruction address of the instruction that has been executed and a hazard indicating flag (such as a bit) associated with the instruction. The hazard indicating flag indicates that the instruction has encountered the operand store compare hazard.
- a method for managing load and store operations executed out-of- order is disclosed. The method comprises fetching an instruction. The instruction is one of a load instruction and a store instruction. The instruction is decoded.
- An operand store compare hazard prediction table is queried with an instruction address of the instruction in response to the decoding.
- the operand store compare hazard prediction table comprises a first entry for a load instruction and a second entry for a store instruction.
- the first and second entries are independent of each other.
- the first and second entries indicate that the load instruction and the store instruction, respectively, have previously encountered an operand store compare hazard.
- the first and second entries comprise at least an instruction address of one of the load instruction and store instruction, respectively, and a hazard indicating flag associated with an operand store compare hazard.
- a determination is made, in response to querying the prediction table, that the instruction is associated with one of the first and second entries in the operand store compare hazard prediction table.
- the hazard indicating flag included within the one of the first and second entries associated with the instruction is identified based on the determination.
- the instruction is a load instruction.
- the instruction is marked based on the hazard indicating flag that has been identified. The marking makes an execution of the instruction dependent upon at least one store instruction, associated with an entry of the prediction table comprising a hazard indicating flag substantially similar to the hazard indicating flag associated with the instruction, having reached a given execution stage.
- the instruction is associated with the second entry.
- an information processing system for managing load and store operations executed out-of-order.
- the information processing system comprises a memory and a processor that is communicatively coupled to the memory.
- the processor is configured to perform a method comprising executing at least one of a load instruction and a store instruction.
- a determination is made, based on the executing, that an operand store compare hazard has been encountered.
- An entry within an operand store compare hazard prediction table is created based on the determination.
- the entry comprises at least an instruction address of the instruction that has been executed and a hazard indicating flag (such as a bit) associated with the instruction.
- the hazard indicating flag indicates that the instruction has encountered the operand store compare hazard.
- FIG. 1 illustrates one example of an operating environment according to one embodiment of the present invention
- FIG. 2 shows one example of a load queue entry according to one embodiment of the present invention
- FIG. 3 shows one example of a store queue entry according to one embodiment of the present invention
- FIG. 4 shows one example of an operand store compare hazard prediction table entry according to one embodiment of the present invention
- FIGs. 5-7 are operational flow diagrams illustrating various examples of creating an entry in an operand store compare hazard prediction table according to various embodiments of the present invention
- FIG. 8 is an operational flow diagram illustrating one example of predicting and preventing operand store compare hazards according to various embodiments of the present invention.
- FIG. 9 is a block diagram illustrating one example of an information processing system according to one embodiment of the present invention. DETAILED DESCRIPTION
- three operand-store- compare hazards can occur due to reordering between dependent loads and stores. For example, assume that a Store to address A is followed by a Load to address A. In one situation the Load can execute before the Store, i.e., the Store Queue (STQ) does not comprise the store address information. Therefore, the store queue does not indicate a conflict when the load executes. Once the Load finishes execution, the Store executes and detects the conflict against the already finished Load in the load queue and flushes the pipeline to stop the Load and any subsequent instruction.
- STQ Store Queue
- the situation above is referred to as a Store- hit-Load (SHL).
- the Store executes its address calculation, but the data for the Store is delayed, e.g. because the data-producing instruction is has a long latency (e.g. divide).
- the Load executes before the store data is written into the STQ.
- the Load detects that it is dependent on the Store, but the Load cannot perform store-data-forwarding since the data is not available. Therefore, the Load needs to reject and retry later on after the store data has become available.
- This situation is referred to as a non-forwardable Load-hit-Store (nf- LHS).
- a prediction table in at least one embodiment, is created that predicts which Loads and Stores have dependencies, and the type of these dependencies (such as e-bit or w-bit dependencies). Then after instruction decoding, e-bit Loads are made dependent on all prior e-bit Stores, and are treated by the instruction issue logic as if there was a regular register dependency. This effectively delays execution of the e-bit Load instruction until after all e-bit Stores have executed their address calculation, and written their data into the STQ. This in effect removes SHL and nf-LHS hazards. For w-bit dependencies, the Load is made dependent on the LI cache writeback of the last store that was predicted as w-bit Store. This effectively prevents persistent nf-LHS hazards.
- FIG. 1 is a block diagram illustrating one example of an operating environment 100 applicable to one or more processes instructions and data in accordance with one or more embodiments of the present invention.
- the processor 101 comprises a single integrated circuit processor such as a superscalar processor, which, includes various execution units, registers, buffers, memories, and other functional units that are all formed by integrated circuitry.
- the processor 101 in one embodiment, is capable of issuing and executing instructions out-of-order.
- the processor 101 in one embodiment, comprises an instruction fetch unit (IFU) 102, an instruction decode unit (IDU) 104, an instruction issue unit (ISU) 106, a load/store unit
- IFU instruction fetch unit
- IDU instruction decode unit
- ISU instruction issue unit
- the IFU 102 comprises an operand-store-compare (OSC) prediction table 116.
- the OSC prediction table 116 is discussed in greater detail below.
- the issue unit 106 comprises an issue queue 118.
- the LSU 106 in this embodiment, comprises a load queue (LDQ) 120, a store queue (STQ) 122, and an LI cache 124.
- the LDQ 120 and the STQ 122 each comprise entries 126, 128, respectively, that track additional information associated with outstanding load and store instructions. It should be noted that various embodiments of the present invention are not limited to the configuration of the processor 101 as shown in FIG. 1. The embodiments of the present invention are applicable to a variety of architectures which can vary from the example shown in FIG. 1.
- the IFU 102 fetches instruction codes stored in an I-cache, which can be part of the LI cache 124. These fetched instruction codes are decoded by the IDU 104 into instruction processing data. Once decoded, the instructions are dispatched and temporarily placed in an appropriate issue queue 118. The instructions are held in the issue queue 118 until all their required operands are available. From the issue queue(s) 118, instructions can be issued opportunistically to the execution units, e.g., LSU 108, FXU 112, etc., of the processor 100 for execution. In other words, the instructions can be issued out-of-order. The instructions, however, are maintained in the issue queue(s) 118 until execution of the instructions is complete, and the result data, if any, are written back, in case any of the instructions needs to be reissued.
- the execution units e.g., LSU 108, FXU 112, etc.
- an instruction receives operands, if any, from one or more architected and/or rename registers within a register file coupled to the execution unit. After an execution unit finishes execution of an instruction, the execution unit writes the result to the designated destination as specified by the instruction and removes the instruction from the issue queue and the completion of instructions can then be scheduled in program order.
- the operand address generation unit 110 generates operand address information for load and store instructions and writes these addresses into the respective LDQ 120 and the STQ 122.
- the FXU 112 writes data values in the STQ 122.
- the LSU 108 receives load and store instructions from the ISU 106, and executes the load and store instructions.
- each load instruction includes address information specifying an address of needed data.
- the LSU 108 supports out of order executions of load and store instructions, thereby achieving a high level of performance.
- the LSU 108 is pipelined. That is, the LSU 108 executes load and store instructions via a set of ordered pipeline stages performed in sequence.
- OSC Hazard Management As discussed above, three types of hazards (store-hit-load, non-forwardable load-hit store, and persistent non-forwardable load-hit store) can occur in a processor that executes load and store instructions out-of-order. Therefore, in addition to the general processing mechanisms discussed above with respect to FIG. 1, one or more of the following
- embodiments can also be implemented within the processor 100 to predict and avoid these OSC hazards.
- Every Load is allocated an entry in the LDQ 120, which saves the address of each load after it executed until completion.
- Every Store is allocated an entry in the STQ 122, which similarly saves the store address, from execution of the store address computation until the store completes and has written its data to the LI cache 124.
- an LDQ entry and an STQ entry can also comprise additional information to predict and avoid OSC hazards.
- the LSU 108 executes a load instruction and compares this load to a corresponding entry in the STQ 122.
- the load instruction determines that store- data-forwarding cannot be performed. For example, the load is executing prior to the store data being written to the STQ (nf-LHS) or store-data- forwarding is not allowed even when the data is available (persistent nf-LHS).
- the load instruction sets an OSC hazard bit such as an "e-flag" (e.g., an execution flag) in the STQ entry it compared against if the load instruction detected an nf-LHS hazard.
- the load instruction sets an OSC hazard bit such as a "w-fiag" (e.g., a write flag) in the STQ entry it compared against if the load instruction detected a persistent nf-LHS hazard.
- the load instruction also sets the same OSC hazard bit such as the e-fiag or the w-fiag in its own entry in the LDQ 120.
- the store instruction When an executed store instruction detects an SHL hazard and performs an SHL flush against an LDQ entry, the store instruction sets an OSC hazard bit such as the "e-fiag” in its own STQ entry, and also sets an OSC hazard bit such as the "e-fiag” in the (oldest) LDQ entry the instructions compares against. It should be noted that this LDQ entry is invalidated due to the resulting flush, but the "e-flag" is retained in the LDQ 120.
- the processor pipeline starts refetching and re-executing the flushed instructions, the same load is allocated the same LDQ entry, which now has the "e-fiag" set from before the flush.
- FIGs 2-3 show one example of an LDQ 226 and STQ 328 entry, respectively, according to one embodiment of the present invention.
- an entry 226 in the LDQ queue 120 also comprises one or more OSC hazard bits 208.
- this OSC hazard indicating bit 208 can be an e- flag or a w-fiag depending on whether the load instruction encountered an nf-LHS hazard or a persistent nf-LHS hazard.
- This OSC hazard bit 208 can also be set by a store instruction, as discussed above.
- an entry 328 in the STQ queue 120 also comprises one or more OSC hazard bits 310.
- this OSC hazard bit 310 can be an e-flag or a w-flag depending on whether a load instruction encountered an nf-LHS hazard or a persistent nf- LHS hazard.
- this OSC hazard bit 310 can be an e-flag if the store instruction encountered an SHL hazard, as discussed above.
- an OSC hazard indicating bit 208, 310 in one of the queues 120, 122 will match at least one OSC hazard indicating bit 208, 310 in the other queue 120, 122 since the load or store instruction sets the same bit in an entry of the other queue as it sets in an entry of its own queue. Also, a discussion on how the OSC hazard bit information 208, 310 is used to predict and avoid OSC hazards is given below.
- the load instruction determines if it has OSC hazard bit information, such as an e-flag or a w-flag, in the LDQ 120. If so, the load instruction indicates this to the IFU 102.
- the IFU 102 in one embodiment, then generates an entry in an OSC prediction table 116 for this particular load instruction.
- the OSC prediction table will create an entry based on the instruction address of the Load and remember the one or more flags for this Load. For example, an entry in OSC prediction table indicates whether a Load is associated with an e-flag and/or a w-flag, where the Load can have both flags if the Load compares against multiple store queues.
- the store instruction determines if it has OSC hazard bit information, such as an e-flag or a w-flag, in the STQ 122. If so, the store instruction indicates this to the IFU 102. The IFU 102 then generates an entry in an OSC prediction table 116, for this particular store instruction comprising the instruction address of the store instruction and the one or more flags under the instruction address of the Store. Also, when the store instruction is written back to the LI cache 124, the STQ 122 informs the ISU 106 of the STQ-entry-number (stag) of that given store instruction within the STQ 122.
- FIG. 4 shows one example of a Load instruction entry and a Store instruction entry within the OSC prediction table 116.
- the Load instruction entry 402 comprises the instruction address 404 of the instruction, an entry valid bit 406, and one or more hazard indicating bits such as a load "e” bit 408 and a load "w” bit 410.
- the Store instruction entry 403 comprises the instruction address 412 of the instruction, an entry valid bit 414, and one or more hazard indicating bits such as a store "e” bit 416 and a store "w” bit 418.
- the OSC hazard indicating bits 408, 410, 414, 416 are set based on the information obtained from the load instructions and the STQ 122, as discussed above.
- Each Load and Store entry within the OSC prediction table 116 are created independent of each other. In other words, a Load entry in the OSC prediction table 116 does not reference and is not referenced by a Store entry in the OSC prediction table 116 or any other table for that matter, and vice versa.
- a load instruction that has set an "e" dependency bit in its LDQ entry and an "e” bit in an STQ entry of a corresponding store instruction notifies the IFU 102 of this, which then creates an entry 402, 403 for each of the load and store instructions in the prediction table 116, as shown in FIG. 4.
- the load "e" bit 408 is set in the Load instruction entry 402 and the store "e” bit 414 is set in the Store instruction entry 403.
- various techniques can be used to generate the prediction table based on instruction addresses. For example, techniques directed to overflows (least- recently-used techniques), read/write conflicts (queuing), and similar techniques for prediction tables can be used.
- the IFU 102 each time an instruction is fetched by the IFU 102 and sent into the pipeline for decoding by the IDU 104, the IFU 102, in parallel, queries the OSC prediction table 116 and sends this information to the IDU 104. This query is used by the IDU 104 to determine whether the given fetched instruction is recognized as a load and/or store comprising an "e" or "w" bit. For example, the IFU 102 uses the instruction address of an instruction to query the prediction table 116 to identify an entry with the corresponding instruction address. The information obtained from the OSC prediction table 116 is passed from the IDU 104 to the ISU 106.
- the ISU 106 uses this dependency information to perform the following. If the instruction comprises a store-e-bit the ISU 106 remembers the instruction as a store-e-bit instruction. If the instruction has a store-w-bit, the ISU 106 remembers the STQ entry number ("stag") that is allocated for the instruction. The ISU 106, in one embodiment, remembers the youngest such stag, referred to as the w-stag. If the instruction has a load-e-bit, the ISU 106 marks this instruction as being dependent on any prior instruction that -was marked with a store-e- bit. This dependency is very similar to the dependency that is created between the writer and the reader of a given General Purpose Register.
- the ISU 106 ensures that the load instruction does not execute before the store instruction has successfully executed, and has written the store address and data into its STQ entry. By delaying the load until that point, the ISU 106 guarantees that the Load can obtain store- forwarded data from the STQ entry, and thus, an SHL hazard or an nf-LHS hazard is prevented.
- the Load is marked such that the ISU 106 does not allow the load to execute until the stag of the last store-w-bit Store before the Load (i.e. the w-stag) has written back to the LI cache 124 (as discussed above, the STQ 122 informs the ISU 106 when the writeback occurs). By delaying the load until after the store has written back to the LI cache 124, persistent nf-LHS hazards are prevented. Effectively the above process performed by the ISU 106 makes all e-bit-Loads dependent on all prior e-bit-Store's execution, and all w-bit-Loads dependent on all prior w-bit-Store's write back.
- the youngest non- flushed w-bit store is designated as the saved youngest w-bit store. This option may require significant tracking expense.
- the youngest non-flushed stag becomes the saved youngest w-bit store. This store may not have actually been marked as a w-bit store. This option does not require much tracking expense, but treats a store as a w-store even though that store may not have had a w- bit prediction. This leads to a slight performance degradation.
- the saved youngest w-bit store is invalidated. W-bit loads dispatched while the saved youngest w-bit store is still in an invalid state will not mark a stag dependency.
- steps are taken to ensure that these instructions do not mark a w-bit dependency on themselves. This is particularly a problem if the instruction is made of multiple parts. Preventing this dependency can be accomplished, in one embodiment, by ensuring that w-bit marked store-parts always follow the w-bit marked load-parts, or by ensuring that the saved youngest w-bit store is not updated until all of the parts of an instruction have dispatched.
- marking e-bit dependencies within an instruction consisting of multiple parts is safe because the parts will simply be issued in- order with respect to each other. However, it is possible to use similar methods to prevent this dependency, if desired.
- a vector of e-bit marked stores is maintained (other register dependency-like tracking mechanisms are possible, 1 bit per issue queue entry, and e-bit marked loads are made dependent on all older valid issue queue entries for which the corresponding bit in the vector is set. Bits in the vector are written when an instruction is dispatched into them, and are put in the set state if the instruction is an e-bit store or in the unset state otherwise.
- Loads and Stores can be tracked in groups of e and w bits (e.g. there could be 5 e-bits el ...e5, and only el -stores and el -loads are paired, and e2-stores and a2-loads are paired, and so on).
- a random e-bit e.g. e2
- This design can be extended to the method discussed above where certain instructions are both loads and stores (e.g. the CS instruction in System z), by treating the two aspects separately, but enforcing the dependencies both as a Store and a Load.
- the above embodiments of the present invention are advantageous in that a prediction table is created that predicts which Loads and Stores have dependencies, and the type of these dependencies (such as e-bit or w-bit dependencies. Then after instruction decoding, e-bit Loads are made dependent on all prior e-bit Stores, and are treated by the instruction issue logic as if there was a regular register dependency. This effectively delays execution of the e-bit Load instruction until after all e-bit Stores have executed their address calculation, and written their data into the STQ. This in effect removes SHL and nf-LHS hazards.
- the Load is made dependent on the LI cache writeback of the last store that was predicted as w-bit Store. This effectively prevents persistent nf-LHS hazards.
- Each Load entry and each Store entry are independent of each other within the OSC prediction table. In other words, a Load instruction entry does not reference a Store instruction entry and vice versa. This allows dependencies to be created between multiple store instructions and multiple loads.
- the Load if a Load has both an e-bit and a w-bit set, the Load is delayed until after all e-bit Stores and after all w-bit Stores indicated in the OSC prediction table. That is, the Load is delayed until after all e-bit Stores have executed their address calculation, and written their data into the STQ, and the Load is also made dependent on the LI cache writeback of the last w-bit Store.
- a Load has an e-bit set in the OSC table and a Store has both an e-bit and a w-bit set in the OSC table
- the Load is delayed until after the Store with the e-bit set, has executed its address calculation, and written its data into the STQ (e.g., the data is forwarded).
- the Load is delayed until after the Store with the w-bit set, has executed the
- one or more Loads and one or more Stores can be included in one complex instruction.
- this type of complex instruction may be found in the following publication entitled "z/ Architecture Principles of Operation", SA22- 7832-07, Eighth Edition, published February 2009, by International Business Machines. That is, one instruction can be considered a Load and a Store. The same instruction can create one or more Load entries and one or more Store entries in the OSC table.
- Loads and Stores can be tracked according to their respective e-bits and w-bits in the OSC table.
- FIG. 5 is an operational flow diagram illustrating one example of generating an entry in an
- OSC prediction table 116 for predicting and preventing OSC hazards.
- the operational flow diagram of FIG. 5 begins at step 502 and flows directly into step 504.
- a load instruction begins executing prior to an associated store instruction.
- the load instruction at step 506, obtains data from a memory location where the store instruction will write to in the future.
- the load instruction at step 508, finishes executing.
- the store instruction at step
- the store instruction begins to execute.
- the store instruction determines that the load instruction has previously obtained data from a memory location that the store instruction is currently writing to.
- the store instruction determines that an SHL hazard has been encountered.
- the store instruction at step 516, then sets a flag bit such as an e-flag bit to indicate this instruction is a candidate for an OSC hazard situation. This e-flag bit is set in the oldest LDQ entry that store instruction compares against.
- the store instruction at step 518, sets an e-flag bit in the STQ entry associated with the store instruction in the STQ 122.
- the store instruction at step 520, then flushes the load instruction and all younger load instructions from the pipeline.
- the store instruction writes back to the LI cache 124.
- the store instruction informs the IFU 102 that the store instruction has an e-flag (or w-f ag as set by a load instruction) bit pending.
- the IFU 102 at step 526, generates an entry for the store instruction in the OSC prediction table 116.
- This entry includes an instruction address of the store instruction, a valid bit, and an indication that the store instruction is associated with an e-flag. For example, a bit or flag can be set in the entry indicating that the instruction is associated with a store-e-bit.
- the IFU 102 at step 527, also generates an entry for the load instruction in the OSC prediction table 116.
- the pipeline re-executes the load, which uses the same LDQ entry as before.
- the load writes its flag (e.g., e-bit) into the IFU prediction table 116.
- This entry includes an instruction address of the load instruction, a valid bit, and an indication that the load instruction is associated with an e-flag.
- the store and load entries are independent of each other and do not reference each other in anyway.
- the STQ 122 at step 528, informs the ISU 106 of the STQ entry number (stag) of the store instruction that has written back to the LI cache 124.
- the control flow then exits at step 530.
- FIG. 6 is an operational flow diagram illustrating another example of generating an entry in an OSC prediction table 116 for predicting and preventing OSC hazards.
- the operational flow diagram of FIG. 6 begins at step 602 and flows directly into step 604.
- a store instruction at step 604, executes its address calculation.
- the data for the store instruction at step 606, is delayed.
- the load instruction at step 608, executes before the store data is written into the STQ 122.
- the load instruction, at step 610 determines that it is dependent on the store instruction and cannot perform store-data-forwarding.
- the load instruction, at step 612 determines that an nf-LHS situation has been encountered.
- the load instruction, at step 614 sets an e-flag bit in the STQ entry of the store instruction.
- the load instruction sets an e-flag bit in a corresponding LDQ entry.
- the load instruction finishes executing.
- the load instruction sends information to the IFU 102 that it has set an e-flag bit in the LDQ 120.
- the IFU at step 622, generates an entry for the load instruction in an OSC prediction table 116.
- This entry includes an instruction address of the load instruction, a valid bit, and an indication that the load instruction is associated with an e-flag bit. For example, a bit can or flag can be set in the entry indicating that the instruction is associated with a load-e-bit.
- the IFU 102 also generates an entry for the store instruction in the OSC prediction table 116. For example, when the store instruction writes back into the LI -cache 206 (which can happen before or after step 620) the flag (e-bit) in the STQ is communicated to the IFU 102 and an entry for the store is created in the table 116. This entry includes an instruction address of the store instruction, a valid bit, and an indication that the store instruction is associated with an e-flag bit. The load and store entries are independent of each other and do not reference each other in anyway. The control flow then exits at step 624.
- FIG. 7 is an operational flow diagram illustrating yet another example of generating an entry in an OSC prediction table 116 for predicting and preventing OSC hazards.
- the operational flow diagram of FIG. 7 begins at step 702 and flows directly into step 704.
- a store instruction at step 704, executes its address calculation.
- a load instruction at step 706, begins its execution.
- the load instruction at step 708, determines that it is dependent on the store instruction and cannot perform store-data-forwarding.
- the load instruction determines that a persistent nf-LHS situation has been encountered.
- the load instruction at step 712, sets a w-flag bit in the STQ entry of the store instruction.
- the load instruction sets a w-flag bit in a corresponding LDQ entry.
- the load instruction at step 716, finishes executing.
- the load instruction at step 718, informs the IFU 102 that it has set a w-flag bit in the LDQ 120.
- the IFU at step 720, generates an entry for the load instruction in an OSC prediction table 116. This entry includes an instruction address of the load instruction, a valid bit, and an indication that the load instruction is associated with a w-flag bit.
- a bit can or flag can be set in the entry indicating that the instruction is associated with a load-w-bit.
- the IFU at step 721, generates an entry for the store instruction in an OSC prediction table 116. For example, when the store instruction writes back into the LI -cache 206 (which can happen before or after step 718) the flag (e-bit) in the STQ is communicated to the IFU 102 and an entry for the store is created in the table 116.
- This entry includes an instruction address of the store instruction, a valid bit, and an indication that the store instruction is associated with a w-flag bit.
- the load and store entries are independent of each other and do not reference each other in anyway. The control flow then exits at step 722.
- FIG. 8 is an operational flow diagram illustrating one example of predicting and preventing OSC hazards.
- the operational flow diagram of FIG. 8 begins at step 802 and flows directly into step 804.
- the IFU 102 fetches an instruction.
- the IFU 102 at step 806, in parallel, queries the OSC prediction table 116 with the instruction address of the instruction.
- the IFU 102 determines if the instruction comprises an entry in the prediction table 116. If the result of this determination is negative, conventional processing, at step 810, is performed.
- the control flow then exits at step 812.
- the IFU 102 sends the instruction and the OSC hazard information associated with the instruction obtained from the OSC prediction table 116 to the IDU 104.
- the IDU 104 decodes the instruction.
- the IDU 104 determines if the instruction comprises a store-e-bit (e.g., the instruction is a store with an e-flag bit). If the result of this determination is positive, the ISU 106, at step 818, remembers the store instruction as a store-e-bit. If this instruction only comprises a single bit then the control flow then returns to step 804. However, if the instruction comprises multiple bits because it performs both one or more loads and one or more stores, the IDU performs steps 826 and 828 if the other bit is a load "e" bit or performs step 832 if the other bit is a load "w" bit.
- a store-e-bit e.g., the instruction is a store with an e-flag bit.
- the IDU 104 determines if the instruction comprises a store- w-bit (e.g., the instruction is a store with a w- flag bit). If the result of this determination is positive, the ISU 106, at step 822, remembers the STQ entry number (stag) that is allocated to the store instruction. The control flow then returns to step 804. If the result of the determination at step 820 is negative, the IDU 104, at step 824, determines if the instruction comprises a load-e-bit (e.g., the instruction is a load with an e- flag bit).
- the ISU 106 marks the load instruction as being dependent on any prior instruction that was marked with a store-e- bit.
- the load instruction, as a result of being marked, at step 828, is prevented from executing before the Store has successfully executed and written the store address and data into its STQ entry. The control flow then returns to step 804.
- the instruction, at step 830 is determined to be a load with a w-flag bit.
- the ISU 106 at step 832 , marks this instruction so as not to execute until the stag of the last store-w-bit Store before the Load (i.e., the w- stag) has written back to the LI cache.
- the control flow then returns to step 804.
- FIG. 9 is a block diagram illustrating detailed view an information processing system 900 according to one embodiment of the present invention.
- the information processing system 900 is based upon a suitably configured processing system adapted to implement one or more embodiments of the present invention. Any suitably configured processing system is similarly able to be used as the information processing system 900 by embodiments of the present invention.
- the information processing system 900 includes a computer 902.
- the computer 902 has a processor(s) 101 such as the processor of FIG. 1.
- the processor 101 comprises the IFU 102 including the OSC prediction table 116; the IDU 104; the ISU 106 comprising the issue queue 118; the LSU 108 comprising the LDQ 120, the STQ 122, and the LI cache 124; the operand address generating unit 110, the FXU 112, and various other components 114, as shown in FIG. 1.
- the processor 101 is connected to a main memory 906, mass storage interface 908, and network adapter hardware 910.
- a system bus 912 interconnects these system components.
- the mass storage interface 908 is used to connect mass storage devices, such as data storage device 914, to the information processing system 900.
- One specific type of data storage device is an optical drive such as a CD/DVD drive, which may be used to store data to and read data from a computer readable medium or storage product such as (but not limited to) a CD/DVD 916.
- Another type of data storage device is a data storage device configured to support, for example, file system operations.
- the information processing system 600 utilizes conventional virtual addressing mechanisms to allow programs to behave as if they have access to a large, single storage entity, referred to herein as a computer system memory, instead of access to multiple, smaller storage entities such as the main memory 906 and data storage device 916.
- a computer system memory is used herein to generically refer to the entire virtual memory of the information processing system 900.
- processor 101 Although only one processor 101 is illustrated for computer 902, computer systems with multiple processors can be used equally effectively.
- Various embodiments of the present invention further incorporate interfaces that each includes separate, fully programmed microprocessors that are used to off-load processing from the processor 101.
- An operating system (not shown) included in the main memory is a suitable multitasking operating system such as, and not for limitation, the GNU/Linux, AIX, Solaris, and HP-UX.
- Various embodiments of the present invention are able to use any other suitable operating system.
- Some embodiments of the present invention utilize architectures, such as an object oriented framework mechanism, that allow instructions of the components of operating system (not shown) to be executed on any processor located within the information processing system 900.
- the network adapter hardware 910 is used to provide an interface to one or more networks 918.
- Various embodiments of the present invention are able to be adapted to work with any data communications connections including present day analog and/or digital techniques or via a future networking mechanism.
Abstract
Description
Claims
Priority Applications (4)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
DE112010004322T DE112010004322T5 (en) | 2009-12-22 | 2010-12-13 | Predict and avoid operand-store-compare comparison hazards in microprocessors with different order |
JP2012545216A JP5850532B2 (en) | 2009-12-22 | 2010-12-13 | Predict and avoid operand store comparison hazards in out-of-order microprocessors |
GB1206367.3A GB2486155B (en) | 2009-12-22 | 2010-12-13 | Predicting and avoiding operand-store-compare hazards in out- of-order microprocessors |
CN201080056309.2A CN102652304B (en) | 2009-12-22 | 2010-12-13 | Predicting and avoiding operand-store-compare hazards in out-of-order microprocessors |
Applications Claiming Priority (4)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US12/644,923 | 2009-12-22 | ||
US12/644,923 US8468325B2 (en) | 2009-12-22 | 2009-12-22 | Predicting and avoiding operand-store-compare hazards in out-of-order microprocessors |
US12/822,960 | 2010-06-24 | ||
US12/822,960 US8521992B2 (en) | 2009-12-22 | 2010-06-24 | Predicting and avoiding operand-store-compare hazards in out-of-order microprocessors |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2011076602A1 true WO2011076602A1 (en) | 2011-06-30 |
Family
ID=43416544
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/EP2010/069496 WO2011076602A1 (en) | 2009-12-22 | 2010-12-13 | Predicting and avoiding operand-store-compare hazards in out-of-order microprocessors |
Country Status (6)
Country | Link |
---|---|
US (2) | US8521992B2 (en) |
JP (2) | JP5850532B2 (en) |
CN (1) | CN102652304B (en) |
DE (1) | DE112010004322T5 (en) |
GB (1) | GB2486155B (en) |
WO (1) | WO2011076602A1 (en) |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2017019008A1 (en) * | 2015-07-27 | 2017-02-02 | Applied Micro Circuits Corporation | Hazard prediction for a group of memory access instructions using a buffer associated with branch prediction |
US10929142B2 (en) | 2019-03-20 | 2021-02-23 | International Business Machines Corporation | Making precise operand-store-compare predictions to avoid false dependencies |
US10977040B2 (en) | 2019-02-19 | 2021-04-13 | International Business Machines Corporation | Heuristic invalidation of non-useful entries in an array |
US11093401B2 (en) | 2014-03-11 | 2021-08-17 | Ampere Computing Llc | Hazard prediction for a group of memory access instructions using a buffer associated with branch prediction |
US11243774B2 (en) | 2019-03-20 | 2022-02-08 | International Business Machines Corporation | Dynamic selection of OSC hazard avoidance mechanism |
Families Citing this family (56)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US9069563B2 (en) * | 2011-09-16 | 2015-06-30 | International Business Machines Corporation | Reducing store-hit-loads in an out-of-order processor |
US9128725B2 (en) * | 2012-05-04 | 2015-09-08 | Apple Inc. | Load-store dependency predictor content management |
US9600289B2 (en) | 2012-05-30 | 2017-03-21 | Apple Inc. | Load-store dependency predictor PC hashing |
US9158691B2 (en) | 2012-12-14 | 2015-10-13 | Apple Inc. | Cross dependency checking logic |
US9880849B2 (en) * | 2013-12-09 | 2018-01-30 | Macom Connectivity Solutions, Llc | Allocation of load instruction(s) to a queue buffer in a processor system based on prediction of an instruction pipeline hazard |
US9274970B2 (en) | 2013-12-31 | 2016-03-01 | Samsung Electronics Co., Ltd. | Method and apparatus for handling processor read-after-write hazards with cache misses |
US9710268B2 (en) | 2014-04-29 | 2017-07-18 | Apple Inc. | Reducing latency for pointer chasing loads |
CN104063329B (en) * | 2014-06-30 | 2017-04-12 | 龙芯中科技术有限公司 | 64-bit immediate operand processing method and device |
US10209995B2 (en) | 2014-10-24 | 2019-02-19 | International Business Machines Corporation | Processor core including pre-issue load-hit-store (LHS) hazard prediction to reduce rejection of load instructions |
US9483409B2 (en) | 2015-02-05 | 2016-11-01 | International Business Machines Corporation | Store forwarding cache |
US9928075B2 (en) | 2015-02-13 | 2018-03-27 | International Business Machines Corporation | Load queue entry reuse for operand store compare history table update |
US9710281B2 (en) * | 2015-07-30 | 2017-07-18 | International Business Machines Corporation | Register comparison for operand store compare (OSC) prediction |
US20170046167A1 (en) * | 2015-08-14 | 2017-02-16 | Qualcomm Incorporated | Predicting memory instruction punts in a computer processor using a punt avoidance table (pat) |
US9606805B1 (en) | 2015-10-19 | 2017-03-28 | International Business Machines Corporation | Accuracy of operand store compare prediction using confidence counter |
US10514925B1 (en) | 2016-01-28 | 2019-12-24 | Apple Inc. | Load speculation recovery |
US9983875B2 (en) | 2016-03-04 | 2018-05-29 | International Business Machines Corporation | Operation of a multi-slice processor preventing early dependent instruction wakeup |
US10437595B1 (en) | 2016-03-15 | 2019-10-08 | Apple Inc. | Load/store dependency predictor optimization for replayed loads |
US10037211B2 (en) | 2016-03-22 | 2018-07-31 | International Business Machines Corporation | Operation of a multi-slice processor with an expanded merge fetching queue |
US10346174B2 (en) | 2016-03-24 | 2019-07-09 | International Business Machines Corporation | Operation of a multi-slice processor with dynamic canceling of partial loads |
US10761854B2 (en) | 2016-04-19 | 2020-09-01 | International Business Machines Corporation | Preventing hazard flushes in an instruction sequencing unit of a multi-slice processor |
US10037229B2 (en) | 2016-05-11 | 2018-07-31 | International Business Machines Corporation | Operation of a multi-slice processor implementing a load/store unit maintaining rejected instructions |
US10740107B2 (en) * | 2016-06-01 | 2020-08-11 | International Business Machines Corporation | Operation of a multi-slice processor implementing load-hit-store handling |
US9934033B2 (en) | 2016-06-13 | 2018-04-03 | International Business Machines Corporation | Operation of a multi-slice processor implementing simultaneous two-target loads and stores |
US10042647B2 (en) | 2016-06-27 | 2018-08-07 | International Business Machines Corporation | Managing a divided load reorder queue |
US10318419B2 (en) | 2016-08-08 | 2019-06-11 | International Business Machines Corporation | Flush avoidance in a load store unit |
US10684859B2 (en) * | 2016-09-19 | 2020-06-16 | Qualcomm Incorporated | Providing memory dependence prediction in block-atomic dataflow architectures |
GB2563582B (en) * | 2017-06-16 | 2020-01-01 | Imagination Tech Ltd | Methods and systems for inter-pipeline data hazard avoidance |
US10572256B2 (en) | 2017-10-06 | 2020-02-25 | International Business Machines Corporation | Handling effective address synonyms in a load-store unit that operates without address translation |
US10606591B2 (en) | 2017-10-06 | 2020-03-31 | International Business Machines Corporation | Handling effective address synonyms in a load-store unit that operates without address translation |
US10417002B2 (en) | 2017-10-06 | 2019-09-17 | International Business Machines Corporation | Hazard detection of out-of-order execution of load and store instructions in processors without using real addresses |
US10606590B2 (en) | 2017-10-06 | 2020-03-31 | International Business Machines Corporation | Effective address based load store unit in out of order processors |
US11175924B2 (en) | 2017-10-06 | 2021-11-16 | International Business Machines Corporation | Load-store unit with partitioned reorder queues with single cam port |
US10394558B2 (en) | 2017-10-06 | 2019-08-27 | International Business Machines Corporation | Executing load-store operations without address translation hardware per load-store unit port |
US10564976B2 (en) | 2017-11-30 | 2020-02-18 | International Business Machines Corporation | Scalable dependency matrix with multiple summary bits in an out-of-order processor |
US10929140B2 (en) | 2017-11-30 | 2021-02-23 | International Business Machines Corporation | Scalable dependency matrix with a single summary bit in an out-of-order processor |
US10572264B2 (en) | 2017-11-30 | 2020-02-25 | International Business Machines Corporation | Completing coalesced global completion table entries in an out-of-order processor |
US10901744B2 (en) | 2017-11-30 | 2021-01-26 | International Business Machines Corporation | Buffered instruction dispatching to an issue queue |
US10802829B2 (en) | 2017-11-30 | 2020-10-13 | International Business Machines Corporation | Scalable dependency matrix with wake-up columns for long latency instructions in an out-of-order processor |
US10564979B2 (en) * | 2017-11-30 | 2020-02-18 | International Business Machines Corporation | Coalescing global completion table entries in an out-of-order processor |
US10884753B2 (en) | 2017-11-30 | 2021-01-05 | International Business Machines Corporation | Issue queue with dynamic shifting between ports |
US10922087B2 (en) | 2017-11-30 | 2021-02-16 | International Business Machines Corporation | Block based allocation and deallocation of issue queue entries |
US10942747B2 (en) | 2017-11-30 | 2021-03-09 | International Business Machines Corporation | Head and tail pointer manipulation in a first-in-first-out issue queue |
CN108519906B (en) * | 2018-03-20 | 2022-03-22 | 东南大学 | Superscalar out-of-order processor steady state instruction throughput rate modeling method |
US10725783B2 (en) * | 2018-11-02 | 2020-07-28 | International Business Machines Corporation | Splitting load hit store table for out-of-order processor |
CN111221579B (en) * | 2018-11-27 | 2022-04-26 | 展讯通信(上海)有限公司 | Method and system for predicting Load instruction execution delay |
JP7151439B2 (en) * | 2018-12-06 | 2022-10-12 | 富士通株式会社 | Arithmetic processing device and method of controlling arithmetic processing device |
US10802830B2 (en) * | 2019-03-05 | 2020-10-13 | International Business Machines Corporation | Imprecise register dependency tracking |
US11099989B2 (en) | 2019-03-12 | 2021-08-24 | International Business Machines Corporation | Coherency maintenance via physical cache coordinate comparison |
US10831661B2 (en) | 2019-04-10 | 2020-11-10 | International Business Machines Corporation | Coherent cache with simultaneous data requests in same addressable index |
US10824430B1 (en) * | 2019-04-25 | 2020-11-03 | International Business Machines Corporation | Resolving operand store compare conflicts |
EP3812891A1 (en) * | 2019-10-21 | 2021-04-28 | ARM Limited | Decoupled access-execute processing |
US11803637B2 (en) | 2020-07-23 | 2023-10-31 | Ventana Micro Systems Inc. | Microprocessor that prevents store-to-load forwarding between different translation contexts |
US11755732B2 (en) * | 2020-07-23 | 2023-09-12 | Ventana Micro Systems Inc. | Microprocessor that conditions store-to-load forwarding on circumstances associated with a translation context update |
US11803638B2 (en) | 2020-07-23 | 2023-10-31 | Ventana Micro Systems Inc. | Microprocessor core with a store dependence predictor accessed using a translation context |
US11755731B2 (en) | 2020-07-23 | 2023-09-12 | Ventana Micro Systems Inc. | Processor that prevents speculative execution across translation context change boundaries to mitigate side channel attacks |
CN115576610B (en) * | 2022-11-18 | 2023-03-10 | 北京数渡信息科技有限公司 | Instruction distribution processing method and device suitable for general sequence emission processor |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5781752A (en) * | 1996-12-26 | 1998-07-14 | Wisconsin Alumni Research Foundation | Table based data speculation circuit for parallel processing computer |
US5913048A (en) * | 1997-03-31 | 1999-06-15 | International Business Machines Corporation | Dispatching instructions in a processor supporting out-of-order execution |
WO2001035212A1 (en) * | 1999-11-10 | 2001-05-17 | Advanced Micro Devices, Inc. | Store buffer which forwards data based on index and optional way match |
US6463523B1 (en) * | 1999-02-01 | 2002-10-08 | Compaq Information Technologies Group, L.P. | Method and apparatus for delaying the execution of dependent loads |
WO2005111794A1 (en) * | 2004-05-05 | 2005-11-24 | Advanced Micro Devices, Inc. | System and method for validating a memory file that links speculative results of load operations to register values |
US20090210679A1 (en) * | 2008-02-15 | 2009-08-20 | International Business Machines Corporation | Processor and method for store data forwarding in a system with no memory model restrictions |
Family Cites Families (11)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5488729A (en) * | 1991-05-15 | 1996-01-30 | Ross Technology, Inc. | Central processing unit architecture with symmetric instruction scheduling to achieve multiple instruction launch and execution |
US5619662A (en) * | 1992-11-12 | 1997-04-08 | Digital Equipment Corporation | Memory reference tagging |
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 |
US5999727A (en) * | 1997-06-25 | 1999-12-07 | Sun Microsystems, Inc. | Method for restraining over-eager load boosting using a dependency color indicator stored in cache with both the load and store instructions |
US6219742B1 (en) * | 1998-04-29 | 2001-04-17 | Compaq Computer Corporation | Method and apparatus for artificially generating general purpose events in an ACPI environment |
US6999952B1 (en) * | 2001-04-18 | 2006-02-14 | Cisco Technology, Inc. | Linear associative memory-based hardware architecture for fault tolerant ASIC/FPGA work-around |
US20030126409A1 (en) * | 2001-12-28 | 2003-07-03 | Toni Juan | Store sets poison propagation |
US7028166B2 (en) * | 2002-04-30 | 2006-04-11 | Advanced Micro Devices, Inc. | System and method for linking speculative results of load operations to register values |
US20050223292A1 (en) * | 2004-02-17 | 2005-10-06 | Lee Chee S | Single instruction type based hardware patch controller |
US7716528B2 (en) * | 2004-09-07 | 2010-05-11 | Broadcom Corporation | Method and system for configurable trigger logic for hardware bug workaround in integrated circuits |
US8443227B2 (en) * | 2008-02-15 | 2013-05-14 | International Business Machines Corporation | Processor and method for workaround trigger activated exceptions |
-
2010
- 2010-06-24 US US12/822,960 patent/US8521992B2/en active Active
- 2010-12-13 JP JP2012545216A patent/JP5850532B2/en active Active
- 2010-12-13 DE DE112010004322T patent/DE112010004322T5/en not_active Ceased
- 2010-12-13 CN CN201080056309.2A patent/CN102652304B/en active Active
- 2010-12-13 WO PCT/EP2010/069496 patent/WO2011076602A1/en active Application Filing
- 2010-12-13 GB GB1206367.3A patent/GB2486155B/en active Active
-
2013
- 2013-07-29 US US13/953,303 patent/US9430235B2/en active Active
-
2015
- 2015-07-23 JP JP2015146076A patent/JP6143306B2/en active Active
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5781752A (en) * | 1996-12-26 | 1998-07-14 | Wisconsin Alumni Research Foundation | Table based data speculation circuit for parallel processing computer |
US5913048A (en) * | 1997-03-31 | 1999-06-15 | International Business Machines Corporation | Dispatching instructions in a processor supporting out-of-order execution |
US6463523B1 (en) * | 1999-02-01 | 2002-10-08 | Compaq Information Technologies Group, L.P. | Method and apparatus for delaying the execution of dependent loads |
WO2001035212A1 (en) * | 1999-11-10 | 2001-05-17 | Advanced Micro Devices, Inc. | Store buffer which forwards data based on index and optional way match |
WO2005111794A1 (en) * | 2004-05-05 | 2005-11-24 | Advanced Micro Devices, Inc. | System and method for validating a memory file that links speculative results of load operations to register values |
US20090210679A1 (en) * | 2008-02-15 | 2009-08-20 | International Business Machines Corporation | Processor and method for store data forwarding in a system with no memory model restrictions |
Non-Patent Citations (2)
Title |
---|
"z/Architecture Principles of Operation", February 2009, INTERNATIONAL BUSINESS MACHINES, pages: SA22 - 7832,07 |
MOSHOVOS A ET AL: "Streamlining inter-operation memory communication via data dependence prediction", PROCEEDINGS OF THE 30TH ANNUAL IEEE/ACM INTERNATIONAL SYMPOSIUM ON MICROARCHITECTURE. MICRO-30. RESEARCH TRIANGLE PARK, NC, DEC. 1 - 3, 1997; [PROCEEDINGS OF THE ANNUAL INTERNATIONAL SYMPOSIUM ON MICROARCHITEC TURE], LOS ALAMITOS, CA : IEEE COMPUTER, vol. 30TH CONF, 1 December 1997 (1997-12-01), pages 235 - 245, XP010261300, ISBN: 978-0-8186-7977-3, DOI: DOI:10.1109/MICRO.1997.645814 * |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US11093401B2 (en) | 2014-03-11 | 2021-08-17 | Ampere Computing Llc | Hazard prediction for a group of memory access instructions using a buffer associated with branch prediction |
WO2017019008A1 (en) * | 2015-07-27 | 2017-02-02 | Applied Micro Circuits Corporation | Hazard prediction for a group of memory access instructions using a buffer associated with branch prediction |
US10977040B2 (en) | 2019-02-19 | 2021-04-13 | International Business Machines Corporation | Heuristic invalidation of non-useful entries in an array |
US10929142B2 (en) | 2019-03-20 | 2021-02-23 | International Business Machines Corporation | Making precise operand-store-compare predictions to avoid false dependencies |
US11243774B2 (en) | 2019-03-20 | 2022-02-08 | International Business Machines Corporation | Dynamic selection of OSC hazard avoidance mechanism |
Also Published As
Publication number | Publication date |
---|---|
US20110153986A1 (en) | 2011-06-23 |
JP2015228237A (en) | 2015-12-17 |
JP2013515306A (en) | 2013-05-02 |
US8521992B2 (en) | 2013-08-27 |
DE112010004322T5 (en) | 2012-08-23 |
GB2486155A (en) | 2012-06-06 |
CN102652304A (en) | 2012-08-29 |
US9430235B2 (en) | 2016-08-30 |
CN102652304B (en) | 2015-04-15 |
GB201206367D0 (en) | 2012-05-23 |
US20130318330A1 (en) | 2013-11-28 |
JP5850532B2 (en) | 2016-02-03 |
GB2486155B (en) | 2017-04-19 |
JP6143306B2 (en) | 2017-06-07 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US9430235B2 (en) | Predicting and avoiding operand-store-compare hazards in out-of-order microprocessors | |
US10963248B2 (en) | Handling effective address synonyms in a load-store unit that operates without address translation | |
US10534616B2 (en) | Load-hit-load detection in an out-of-order processor | |
US10977047B2 (en) | Hazard detection of out-of-order execution of load and store instructions in processors without using real addresses | |
US10776113B2 (en) | Executing load-store operations without address translation hardware per load-store unit port | |
JP2013515306A5 (en) | ||
US20120023314A1 (en) | Paired execution scheduling of dependent micro-operations | |
US11175925B2 (en) | Load-store unit with partitioned reorder queues with single cam port | |
US10310988B2 (en) | Address translation for sending real address to memory subsystem in effective address based load-store unit | |
US10606593B2 (en) | Effective address based load store unit in out of order processors | |
US8468325B2 (en) | Predicting and avoiding operand-store-compare hazards in out-of-order microprocessors | |
JPH07325716A (en) | Pipeline processor and its operating method | |
US10572257B2 (en) | Handling effective address synonyms in a load-store unit that operates without address translation | |
JP7064273B2 (en) | Read / store unit with split reorder queue using a single CAM port |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
WWE | Wipo information: entry into national phase |
Ref document number: 201080056309.2 Country of ref document: CN |
|
121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 10788342 Country of ref document: EP Kind code of ref document: A1 |
|
ENP | Entry into the national phase |
Ref document number: 1206367 Country of ref document: GB Kind code of ref document: A Free format text: PCT FILING DATE = 20101213 |
|
WWE | Wipo information: entry into national phase |
Ref document number: 1206367.3 Country of ref document: GB |
|
WWE | Wipo information: entry into national phase |
Ref document number: 2012545216 Country of ref document: JP |
|
WWE | Wipo information: entry into national phase |
Ref document number: 1120100043221 Country of ref document: DE Ref document number: 112010004322 Country of ref document: DE |
|
122 | Ep: pct application non-entry in european phase |
Ref document number: 10788342 Country of ref document: EP Kind code of ref document: A1 |