摘要:
One embodiment of the present disclosure sets forth an optimized way to execute pre-scheduled replay operations for divergent operations in a parallel processing subsystem. Specifically, a streaming multiprocessor (SM) includes a multi-stage pipeline configured to insert pre-scheduled replay operations into a multi-stage pipeline. A pre-scheduled replay unit detects whether the operation associated with the current instruction is accessing a common resource. If the threads are accessing data which are distributed across multiple cache lines, then the pre-scheduled replay unit inserts pre-scheduled replay operations behind the current instruction. The multi-stage pipeline executes the instruction and the associated pre-scheduled replay operations sequentially. If additional threads remain unserviced after execution of the instruction and the pre-scheduled replay operations, then additional replay operations are inserted via the replay loop, until all threads are serviced. One advantage of the disclosed technique is that divergent operations requiring one or more replay operations execute with reduced latency.
摘要:
One embodiment of the present invention sets forth a technique for processing load instructions for parallel threads of a thread group when a sub-set of the parallel threads request the same memory address. The load/store unit determines if the memory addresses for each sub-set of parallel threads match based on one or more uniform patterns. When a match is achieved for at least one of the uniform patterns, the load/store unit transmits a read request to retrieve data for the sub-set of parallel threads. The number of read requests transmitted is reduced compared with performing a separate read request for each thread in the sub-set. A variety of uniform patterns may be defined based on common access patterns present in program instructions. A variety of uniform patterns may also be defined based on interconnect constraints between the load/store unit and the memory when a full crossbar interconnect is not available.
摘要:
One embodiment of the present invention sets forth an approach for executing replay operations for divergent operations in a parallel processing subsystem. Specifically, the streaming multiprocessor (SM) includes a multistage pipeline configured to batch two or more replay operations for processing via replay loop. A logic element within the multistage pipeline detects whether the current pipeline stage is accessing a shared resource, such as loading data from a shared memory. If the threads are accessing data which are distributed across multiple cache lines, then the multistage pipeline batches two or more replay operations, where the replay operations are inserted into the pipeline back-to-back.
摘要:
One embodiment of the present disclosure sets forth an effective way to maintain fairness and order in the scheduling of common resource access requests related to replay operations. Specifically, a streaming multiprocessor (SM) includes a total order queue (TOQ) configured to schedule the access requests over one or more execution cycles. Access requests are allowed to make forward progress when needed common resources have been allocated to the request. Where multiple access requests require the same common resource, priority is given to the older access request. Access requests may be placed in a sleep state pending availability of certain common resources. Deadlock may be avoided by allowing an older access request to steal resources from a younger resource request. One advantage of the disclosed technique is that older common resource access requests are not repeatedly blocked from making forward progress by newer access requests.
摘要:
One embodiment of the present invention sets forth a technique for performing a shaped access of a register file that includes a set of N registers, wherein N is greater than or equal to two. The technique involves, for at least one thread included in a group of threads, receiving a request to access a first amount of data from each register in the set of N registers, and configuring a crossbar to allow the at least one thread to access the first amount of data from each register in the set of N registers.
摘要:
Techniques for vector completion mask (VCM) handling are provided. A data structure includes a mask field for each operand of a particular operation. A processor attempts to execute the operation with multiple operands, which are identified in the data structure by the mask fields. If operands are successfully retrieved for execution with the operation, then the corresponding mask field within the data structure is cleared. The processor can reset if any field remains set within the data structure and can re-process the operation with operands that were not previously handled with the operation.
摘要:
In one embodiment, the present invention includes a method for executing an operation on low order portions of first and second source operands using a first execution stack of a processor and executing the operation on high order portions of the first and second source operands using a second execution stack of the processor, where the operation in the second execution stack is staggered by one or more cycles from the operation in the first execution stack. Other embodiments are described and claimed.
摘要:
A method is disclosed. The method includes scheduling a load operation at least twice the size of a maximum access supported by a memory device, dividing the load operation into a plurality of separate load operation segments having a size equivalent to the maximum access supported by the memory device, and performing each of the plurality of load operation segments. A further method is disclosed where a temporary register is used to minimize the number of memory accesses to support unaligned accesses.
摘要:
In one embodiment, a method for flow optimization and prediction for vector streaming single instruction, multiple data (SIMD) extension (VSSE) memory operations is disclosed. The method comprises generating an optimized micro-operation (μop) flow for an instruction to operate on a vector if the instruction is predicted to be unmasked and unit-stride, the instruction to access elements in memory, and accessing via the optimized μop flow two or more of the elements at the same time without determining masks of the two or more elements. Other embodiments are also described.
摘要:
One embodiment of the present disclosure sets forth an effective way to maintain fairness and order in the scheduling of common resource access requests related to replay operations. Specifically, a streaming multiprocessor (SM) includes a total order queue (TOQ) configured to schedule the access requests over one or more execution cycles. Access requests are allowed to make forward progress when needed common resources have been allocated to the request. Where multiple access requests require the same common resource, priority is given to the older access request. Access requests may be placed in a sleep state pending availability of certain common resources. Deadlock may be avoided by allowing an older access request to steal resources from a younger resource request. One advantage of the disclosed technique is that older common resource access requests are not repeatedly blocked from making forward progress by newer access requests.