33 results about "Speculative multithreading" patented technology

A system, method and computer program product for supporting thread level speculative execution in a computing environment having multiple processing units adapted for concurrent execution of threads in speculative and non-speculative modes. Each processing unit includes a cache memory hierarchy of caches operatively connected therewith. The apparatus includes an additional cache level local to each processing unit for use only in a thread level speculation mode, each additional cache for storing speculative results and status associated with its associated processor when handling speculative threads. The additional local cache level at each processing unit are interconnected so that speculative values and control data may be forwarded between parallel executing threads. A control implementation is provided that enables speculative coherence between speculative threads executing in the computing environment.

A cooperative multithreading architecture includes an instruction cache, capable of providing a micro-VLIW instruction; a first cluster, connects to the instruction cache to fetch the micro-VLIW instruction; and a second cluster, connects to the instruction cache to fetch the micro-VLIW instruction and capable of execution acceleration. The second cluster includes a second front-end module, connects to the instruction cache and capable of requesting and dispatching the micro-VLIW instruction; a helper dynamic scheduler, connects to the second front-end module and capable of dispatching the micro-VLIW instruction; a non-shared data path, connects to the second front-end module and capable of providing a wider data path; and a shared data path, connected to the helper dynamic scheduler and capable of assisting a control part of the non-shared data path. The first cluster and the second cluster carry out execution of the respective micro-instructions in parallel.

A method for analyzing a set of spawning pairs, where each spawning pair identifies at least one speculative thread. The analysis may be practiced via software in a compiler, binary optimizer, standalone modeler, or the like. The analysis may include determining a predicted execution time for a sequence of program instructions, given the set of spawning pairs, for a target processor having a known number of thread units, where the target processor supports speculative multithreading. The method is further to select a spawning pair, according to a greedy approach, if the spawning pair provides a performance enhancement, in terms of decreased execution time due to increased parallelism, when the speculative thread is spawned during execution of a code sequence. Other embodiments are also described and claimed.

Verifying speculative multithreading in an application executing in a computing system, including: executing one or more test instructions serially thereby producing a serial result, including insuring that all data dependencies among the test instructions are satisfied; executing the test instructions speculatively in a plurality of threads thereby producing a speculative result; and determining whether a speculative multithreading error exists including: comparing the serial result to the speculative result and, if the serial result does not match the speculative result, determining that a speculative multithreading error exists.

A method for analyzing a set of spawning pairs, where each spawning pair identifies at least one speculative thread. The method, which may be practiced via software in a compiler or standalone modeler, determines execution time for a sequence of program instructions, given the set of spawning pairs, for a target processor having a known number of thread units, where the target processor supports speculative multithreading. Other embodiments are also described and claimed.

Disclosed are selected embodiments of a processor that may include a plurality of thread units and a register file architecture to support speculative multithreading. For at least one embodiment, live-in values for a speculative thread are computed via execution of a precomputation slice and are stored in a validation buffer for later validation. A global register file holds the committed architecture state generated by a non-speculative thread. Each thread unit includes a local register file. A directory indicates, for each architectural register, which speculative thread(s) have generated a value for the architectural register. Other embodiments are also described and claimed.

One embodiment of the present invention provides a system that facilitates efficient join operations between a head thread and a speculative thread during speculative program execution, wherein the head thread executes program instructions and the speculative thread executes program instructions in advance of the head thread. The system operates by executing a primary version of a program using the head thread, and by executing a speculative version of the program using the speculative thread. When the head thread reaches a point in the program where the speculative thread began executing, the system performs a join operation between the head thread and the speculative thread. This join operation causes the speculative thread to act as a new head thread by switching from executing the speculative version of the program to executing the primary version of the program. To facilitate this switching operation, the system performs a lookup to determine where the new head thread is to commence executing within the primary version of the program based upon where the speculative thread is currently executing within the speculative version of the program.

The invention discloses an assembly level interprocedual pointer analysis method based on speculative multithreading. Through assembly level interprocedual points-to analysis, a procedure call graph of all processes is established in a source program before the interprocedual points-to analysis is carried out, and after establishment, points-to analysis is carried out in two stages. The method comprises: in the first stage, firstly, establishing a path profiling cost model based on probability, extracting speculative paths selected by all process nodes in the procedure call graph, and then performing points-to analysis in a process; in the second stage, performing interprocedual points-to analysis and interprocedual side-effect calculation, on each process calling point, performing mappingand inverse mapping on points-to relations using a points-to analysis result in a process, to obtain an interprocedual side-effect calculation result, and realize context points-to relation update ofa calling procedure. On this basis, the method guides a whole procedure to perform interprocedual data flow analysis, evaluates data dependence degree among pointer variable memory points-to relations, and obtains a more accurate thread partitioning result.

The invention discloses a CMP (Chip Multiprocessor)-based multi-speculative path thread partitioning method under a speculative multithreading mechanism. According to the method, a process is taken as a unit in thread partitioning; for each process, control-independence nodes of thread end points in the process are limited during partitioning; mutual exclusion path segments of thread excitation points in the process are limited, so that excitation of the thread is relatively strictly limited; meanwhile, the excitation points on the mutual exclusion path segments are made to correspond to the same thread end point; a plurality of continuous pre-computation slices are inserted behind the thread end points; the contents of the pre-computation slices are mutually different along with the changes of the speculative paths and the excitation points; and a simulator carries out different speculative paths when running, and selects corresponding pre-computation slices for executionaccording to the corresponding excitation points on the speculative paths. According to the method, thread partitioning can be carried out on a plurality of paths, so that the branch coverage rate of speculative parallel execution is increased.

Verifying speculative multithreading in an application executing in a computing system, including: executing one or more test instructions serially thereby producing a serial result, including insuring that all data dependencies among the test instructions are satisfied; executing the test instructions speculatively in a plurality of threads thereby producing a speculative result; and determining whether a speculative multithreading error exists including: comparing the serial result to the speculative result and, if the serial result does not match the speculative result, determining that a speculative multithreading error exists.

The invention discloses a speculative thread particle with restart optimization and a restart optimization method thereof. The speculative thread particle consists of an input parameter identification part, an execution part, and a result storage part, wherein the input parameter identification part consists of N+1 binary digits which are orderly arranged, and is used for identifying whether the speculative thread particle uses an input parameter corresponding to an input parameter identification byte; the execution part consists of a plurality of program statements in speculative thread, and is used for storing executive statements of the speculative thread particle; and the result storage part is used for storing an execution result of the speculative thread particle. The restart optimization method has the following operation steps of: A, detecting disabled speculative thread input parameters; B, searching speculative thread particles which depend on the failed speculative thread input parameters in step A; and C, restarting the speculative thread particles found in step B. The restart optimization method has the advantages that the restart optimization method can quickly judge the speculative thread particles which must be restarted when the speculative fails, and reduce the cost for thread restart.

Efficient rollback and retry of conflicted speculative threads using distributed tokens. A method for rolling back speculative threads in symmetric-multiprocessing (SMP) environments is disclosed. In one embodiment, such a method includes detecting an aborted thread at runtime and determining whether the aborted thread is an oldest aborted thread. In the event the aborted thread is the oldest aborted thread, the method sets a high-priority request for allocation to an absolute thread number associated with the oldest aborted thread. The method further detects that the high-priority request is set and, in response, modifies a local allocation token of the oldest aborted thread. The modification prompts the oldest aborted thread to retry a work unit associated with its absolute thread number. The oldest aborted thread subsequently initiates the retry of a successor thread by updating the successor thread's local allocation token. A corresponding apparatus and computer program product are also disclosed.