Abstract: Examples herein describe techniques for reducing the amount of memory used during weight sparsity. When decompressing the weights, the uncompressed weight data typically has many zero values. By knowing the location of these zero values (e.g., their indices in a weight matrix), the processor core can prune some of the activations (e.g., logically reduce the size of the activation matrix) which improves the efficiency of the processor core. In embodiments herein, the processor core includes logic for identifying the indices of the non-zero value after decompressing the compressed weights. These indices can then be used to prune the activations to improve the efficiency of the processor core.

Abstract: Using multiple overlays with a data processing array includes loading an application in a data processing array. The data processing array includes a plurality of compute tiles each having a processor. The application specifies kernels executable by the processors and implements stream channels that convey data to the plurality of compute tiles. During runtime of the application, a plurality of overlays are sequentially implemented in the data processing array. Each overlay implements a different mode of data movement in the data processing array via the stream channels. For each overlay implemented, a workload is performed by moving data to the plurality of compute tiles based on the respective mode of data movement.

Abstract: Examples herein describe techniques for adapting a multiplier array (e.g., a systolic array implemented in a processing core) to perform different dot products. The processing core can include data selection logic that enables different configurations of the multiplier array in the core. For example, the data selection logic can enable different configurations of the multiplier array while using the same underlying hardware. That is, the multiplier array is fixed hardware but the data selection can transmit data into the matrix multiplier such that it is configured to perform different length dot products, perform more dot products in parallel, or change its output precision. In this manner, the same underlying hardware (i.e., the multiplier array) can be reconfigured for different dot products which can result in much more efficient use of the hardware.

Abstract: Examples herein describe techniques for reducing the amount of memory used during weight sparsity. When decompressing the weights, the uncompressed weight data typically has many zero values. By knowing the location of these zero values (e.g., their indices in a weight matrix), the processor core can prune some of the activations (e.g., logically reduce the size of the activation matrix) which improves the efficiency of the processor core. In embodiments herein, the processor core includes logic for identifying the indices of the non-zero value after decompressing the compressed weights. These indices can then be used to prune the activations to improve the efficiency of the processor core.

Abstract: Using multiple overlays with a data processing array includes loading an application in a data processing array. The data processing array includes a plurality of compute tiles each having a processor. The application specifies kernels executable by the processors and implements stream channels that convey data to the plurality of compute tiles. During runtime of the application, a plurality of overlays are sequentially implemented in the data processing array. Each overlay implements a different mode of data movement in the data processing array via the stream channels. For each overlay implemented, a workload is performed by moving data to the plurality of compute tiles based on the respective mode of data movement.