Abstract: Distributed shared memory (DSMEM) comprises blocks of memory that are distributed or scattered across a processor (such as a GPU). Threads executing on a processing core local to one memory block are able to access a memory block local to a different processing core. In one embodiment, shared access to these DSMEM allocations distributed across a collection of processing cores is implemented by communications between the processing cores. Such distributed shared memory provides very low latency memory access for processing cores located in proximity to the memory blocks, and also provides a way for more distant processing cores to also access the memory blocks in a manner and using interconnects that do not interfere with the processing cores' access to main or global memory such as hacked by an L2 cache.

Abstract: Systems and methods for efficient convolution based on matrix multiply and add (MMA) are described. An example processor having a plurality of processing lanes is configured to perform convolution of a matrix of activation elements and a filter matrix in accordance with a configurable series of instructions including a plurality of MMA instructions and shift instructions while reusing activation elements already loaded to the processor or associated memory over a plurality of MMA operations. The filter elements are held stationary at inputs to the processor for multiple cycles for multiple MMA operations while activations are streamed in. Associated methods are also described.

Abstract: Systems and methods for efficient convolution based on matrix multiply and add (MMA) are described. An example processor having a plurality of processing lanes is configured to perform convolution of a matrix of activation elements and a filter matrix in accordance with a configurable series of instructions including a plurality of MMA instructions and shift instructions while reusing activation elements already loaded to the datapath or associated memory over a plurality of MMA operations. Associated methods are also described.

Abstract: Various embodiments include techniques for performing self-synchronizing remote memory operations in a multiprocessor computing system. During a remote memory operation in the multiprocessor computing system, a source processing unit transmits multiple segments of data to a destination processing. For each segment of data, the source processing unit transmits a remote memory operation to the destination processing unit that includes associated metadata that identifies the memory location of a corresponding synchronization object. The remote memory operation along with the metadata is transmitted as a single unit to the destination processing unit. The destination processing unit splits the operation into the remote memory operation and the memory synchronization operation. As a result, the source processing unit avoids the need to perform a separate memory synchronization operation, thereby reducing inter-processor communications and increasing performance of remote memory operations.

Abstract: A parallel processing unit comprises a plurality of processors each being coupled to a memory access hardware circuitry. Each memory access hardware circuitry is configured to receive, from the coupled processor, a memory access request specifying a coordinate of a multidimensional data structure, wherein the memory access hardware circuit is one of a plurality of memory access circuitry each coupled to a respective one of the processors; and, in response to the memory access request, translate the coordinate of the multidimensional data structure into plural memory addresses for the multidimensional data structure and using the plural memory addresses, asynchronously transfer at least a portion of the multidimensional data structure for processing by at least the coupled processor. The memory locations may be in the shared memory of the coupled processor and/or an external memory.

Abstract: Various embodiments include techniques for performing self-synchronizing remote memory operations in a data center or multiprocessor computing system. During a remote memory operation, a source processor transmits multiple data segments to a destination processor. For each data segment, the source processor transmits a remote memory operation to the destination processor that includes associated metadata that identifies the memory location of a corresponding synchronization object representing a count of data segments to be stored or a flag for each data segment to be stored. The remote memory operation along with the metadata is transmitted as a single unit to the destination processor. The destination processor splits the operation into the remote memory operation and the memory synchronization operation.

Abstract: Various embodiments include techniques for performing self-synchronizing remote memory operations in a multiprocessor computing system. During a remote memory operation in the multiprocessor computing system, a source processing unit transmits multiple segments of data to a destination processing. For each segment of data, the source processing unit transmits a remote memory operation to the destination processing unit that includes associated metadata that identifies the memory location of a corresponding synchronization object. The remote memory operation along with the metadata is transmitted as a single unit to the destination processing unit. The destination processing unit splits the operation into the remote memory operation and the memory synchronization operation. As a result, the source processing unit avoids the need to perform a separate memory synchronization operation, thereby reducing inter-processor communications and increasing performance of remote memory operations.

Abstract: This specification describes a programmatic multicast technique enabling one thread (for example, in a cooperative group array (CGA) on a GPU) to request data on behalf of one or more other threads (for example, executing on respective processor cores of the GPU). The multicast is supported by tracking circuitry that interfaces between multicast requests received from processor cores and the available memory. The multicast is designed to reduce cache (for example, layer 2 cache) bandwidth utilization enabling strong scaling and smaller tile sizes.

Abstract: This specification describes a programmatic multicast technique enabling one thread (for example, in a cooperative group array (CGA) on a GPU) to request data on behalf of one or more other threads (for example, executing on respective processor cores of the GPU). The multicast is supported by tracking circuitry that interfaces between multicast requests received from processor cores and the available memory. The multicast is designed to reduce cache (for example, layer 2 cache) bandwidth utilization enabling strong scaling and smaller tile sizes.

Abstract: An application may store data to a dataset comprising a plurality of volumes stored on a plurality of storage systems. The application may request a dataset image of the dataset, the dataset image comprising a volume image of each volume of the dataset. A dataset image manager operates with a plurality of volume image managers in parallel to produce the dataset image, each volume image manager executing on a storage system. The plurality of volume image managers respond by performing requested operations and sending responses to the dataset image manager in parallel. Each volume image manager on a storage system may manage and produce a volume image for each volume of the dataset stored to the storage system. If a volume image for any volume of the dataset fails, or a timeout period expires, a cleanup procedure is performed to delete any successful volume images.

Abstract: Various embodiments include techniques for performing self-synchronizing remote memory operations in a multiprocessor computing system. During a remote memory operation in the multiprocessor computing system, a source processing unit transmits multiple segments of data to a destination processing. For each segment of data, the source processing unit transmits a remote memory operation to the destination processing unit that includes associated metadata that identifies the memory location of a corresponding synchronization object. The remote memory operation along with the metadata is transmitted as a single unit to the destination processing unit. The destination processing unit splits the operation into the remote memory operation and the memory synchronization operation. As a result, the source processing unit avoids the need to perform a separate memory synchronization operation, thereby reducing inter-processor communications and increasing performance of remote memory operations.

Abstract: A technique for block data transfer is disclosed that reduces data transfer and memory access overheads and significantly reduces multiprocessor activity and energy consumption. Threads executing on a multiprocessor needing data stored in global memory can request and store the needed data in on-chip shared memory, which can be accessed by the threads multiple times. The data can be loaded from global memory and stored in shared memory using an instruction which directs the data into the shared memory without storing the data in registers and/or cache memory of the multiprocessor during the data transfer.

Abstract: An application may store data to a dataset comprising a plurality of volumes stored on a plurality of storage systems. The application may request a dataset image of the dataset, the dataset image comprising a volume image of each volume of the dataset. A dataset image manager operates with a plurality of volume image managers in parallel to produce the dataset image, each volume image manager executing on a storage system. The plurality of volume image managers respond by performing requested operations and sending responses to the dataset image manager in parallel. Each volume image manager on a storage system may manage and produce a volume image for each volume of the dataset stored to the storage system. If a volume image for any volume of the dataset fails, or a timeout period expires, a cleanup procedure is performed to delete any successful volume images.

Abstract: A new synchronization system synchronizes data exchanges between producer processes and consumer processes which may be on the same or different processors in a multiprocessor system. The synchronization incurs less than one roundtrip of latency - in some implementations, in approximately 0.5 roundtrip times. A key aspect of the fast synchronization is that the producer’s data store is followed without delay with the updating of a barrier on which the consumer is waiting.

Abstract: An application may store data to a dataset comprising a plurality of volumes stored on a plurality of storage systems. The application may request a dataset image of the dataset, the dataset image comprising a volume image of each volume of the dataset. A dataset image manager operates with a plurality of volume image managers in parallel to produce the dataset image, each volume image manager executing on a storage system. The plurality of volume image managers respond by performing requested operations and sending responses to the dataset image manager in parallel. Each volume image manager on a storage system may manage and produce a volume image for each volume of the dataset stored to the storage system. If a volume image for any volume of the dataset fails, or a timeout period expires, a cleanup procedure is performed to delete any successful volume images.

Abstract: Various embodiments include techniques for utilizing resources on a processing unit. Thread groups executing on a processor begin execution with specified resources, such as a number of registers and an amount of shared memory. During execution, one or more thread groups may determine that the thread groups have excess resources needed to execute the current functions. Such thread groups can deallocate the excess resources to a free pool. Similarly, during execution, one or more thread groups may determine that the thread groups have fewer resources needed to execute the current functions. Such thread groups can allocate the needed resources from the free pool. Further, producer thread groups that generate data for consumer thread groups can deallocate excess resources prior to completion. The consumer thread groups can allocate the excess resources and initiate execution while the producer thread groups complete execution, thereby decreasing latency between producer and consumer thread groups.

Abstract: A parallel processing unit comprises a plurality of processors each being coupled to a memory access hardware circuitry. Each memory access hardware circuitry is configured to receive, from the coupled processor, a memory access request specifying a coordinate of a multidimensional data structure, wherein the memory access hardware circuit is one of a plurality of memory access circuitry each coupled to a respective one of the processors; and, in response to the memory access request, translate the coordinate of the multidimensional data structure into plural memory addresses for the multidimensional data structure and using the plural memory addresses, asynchronously transfer at least a portion of the multidimensional data structure for processing by at least the coupled processor. The memory locations may be in the shared memory of the coupled processor and/or an external memory.

Abstract: A new level(s) of hierarchy—Cooperate Group Arrays (CGAs)—and an associated new hardware-based work distribution/execution model is described. A CGA is a grid of thread blocks (also referred to as cooperative thread arrays (CTAs)). CGAs provide co-scheduling, e.g., control over where CTAs are placed/executed in a processor (such as a GPU), relative to the memory required by an application and relative to each other. Hardware support for such CGAs guarantees concurrency and enables applications to see more data locality, reduced latency, and better synchronization between all the threads in tightly cooperating collections of CTAs programmably distributed across different (e.g., hierarchical) hardware domains or partitions.

Abstract: A parallel processing unit comprises a plurality of processors each being coupled to a memory access hardware circuitry. Each memory access hardware circuitry is configured to receive, from the coupled processor, a memory access request specifying a coordinate of a multidimensional data structure, wherein the memory access hardware circuit is one of a plurality of memory access circuitry each coupled to a respective one of the processors; and, in response to the memory access request, translate the coordinate of the multidimensional data structure into plural memory addresses for the multidimensional data structure and using the plural memory addresses, asynchronously transfer at least a portion of the multidimensional data structure for processing by at least the coupled processor. The memory locations may be in the shared memory of the coupled processor and/or an external memory.

Abstract: Distributed shared memory (DSMEM) comprises blocks of memory that are distributed or scattered across a processor (such as a GPU). Threads executing on a processing core local to one memory block are able to access a memory block local to a different processing core. In one embodiment, shared access to these DSMEM allocations distributed across a collection of processing cores is implemented by communications between the processing cores. Such distributed shared memory provides very low latency memory access for processing cores located in proximity to the memory blocks, and also provides a way for more distant processing cores to also access the memory blocks in a manner and using interconnects that do not interfere with the processing cores' access to main or global memory such as hacked by an L2 cache.