Abstract: Embodiments are generally directed to compression in machine learning and deep learning processing. An embodiment of an apparatus for compression of untyped data includes a graphical processing unit (GPU) including a data compression pipeline, the data compression pipeline including a data port coupled with one or more shader cores, wherein the data port is to allow transfer of untyped data without format conversion, and a 3D compression/decompression unit to provide for compression of untyped data to be stored to a memory subsystem and decompression of untyped data from the memory subsystem.

Abstract: A graphics processing device is provided that includes a set of compute units to execute a workload, a cache coupled with the set of compute units, and circuitry coupled with the cache and the set of compute units. The circuitry is configured to, in response to a cache miss for the read from a first cache, broadcast an event within the graphics processor device to identify data associated with the cache miss, receive the event at a second compute unit in the set of compute units, and prefetch the data identified by the event into a second cache that is local to the second compute unit before an attempt to read the instruction or data by the second thread.

Abstract: Embodiments are generally directed to compression in machine learning and deep learning processing. An embodiment of an apparatus for compression of untyped data includes a graphical processing unit (GPU) including a data compression pipeline, the data compression pipeline including a data port coupled with one or more shader cores, wherein the data port is to allow transfer of untyped data without format conversion, and a 3D compression/decompression unit to provide for compression of untyped data to be stored to a memory subsystem and decompression of untyped data from the memory subsystem.

Abstract: A graphics processing device comprises a set of compute units to execute multiple threads of a workload, a cache coupled with the set of compute units, and a prefetcher to prefetch instructions associated with the workload. The prefetcher is configured to use a thread dispatch command that is used to dispatch threads to execute a kernel to prefetch instructions, parameters, and/or constants that will be used during execution of the kernel. Prefetch operations for the kernel can then occur concurrently with thread dispatch operations.

Abstract: Embodiments are generally directed to compression in machine learning and deep learning processing. An embodiment of an apparatus for compression of untyped data includes a graphical processing unit (GPU) including a data compression pipeline, the data compression pipeline including a data port coupled with one or more shader cores, wherein the data port is to allow transfer of untyped data without format conversion, and a 3D compression/decompression unit to provide for compression of untyped data to be stored to a memory subsystem and decompression of untyped data from the memory subsystem.

Abstract: Apparatuses, systems, and techniques are presented to perform one or more operations. In at least one embodiment, one or more data values, to be used by one or more neural networks, are caused to be replaced by one or more invalid data values.

Abstract: Embodiments are generally directed to compression in machine learning and deep learning processing. An embodiment of an apparatus for compression of untyped data includes a graphical processing unit (GPU) including a data compression pipeline, the data compression pipeline including a data port coupled with one or more shader cores, wherein the data port is to allow transfer of untyped data without format conversion, and a 3D compression/decompression unit to provide for compression of untyped data to be stored to a memory subsystem and decompression of untyped data from the memory subsystem.

Abstract: A graphics processing device comprises a set of compute units to execute multiple threads of a workload, a cache coupled with the set of compute units, and a prefetcher to prefetch instructions associated with the workload. The prefetcher is configured to use a thread dispatch command that is used to dispatch threads to execute a kernel to prefetch instructions, parameters, and/or constants that will be used during execution of the kernel. Prefetch operations for the kernel can then occur concurrently with thread dispatch operations.

Abstract: A graphics processing device comprises a set of compute units to execute multiple threads of a workload, a cache coupled with the set of compute units, and a prefetcher to prefetch instructions associated with the workload. The prefetcher is configured to use a thread dispatch command that is used to dispatch threads to execute a kernel to prefetch instructions, parameters, and/or constants that will be used during execution of the kernel. Prefetch operations for the kernel can then occur concurrently with thread dispatch operations.

Abstract: Embodiments are generally directed to compression in machine learning and deep learning processing. An embodiment of an apparatus for compression of untyped data includes a graphical processing unit (GPU) including a data compression pipeline, the data compression pipeline including a data port coupled with one or more shader cores, wherein the data port is to allow transfer of untyped data without format conversion, and a 3D compression/decompression unit to provide for compression of untyped data to be stored to a memory subsystem and decompression of untyped data from the memory subsystem.

Abstract: A graphics processing device comprises a set of compute units to execute multiple threads of a workload, a cache coupled with the set of compute units, and a prefetcher to prefetch instructions associated with the workload. The prefetcher is configured to use a thread dispatch command that is used to dispatch threads to execute a kernel to prefetch instructions, parameters, and/or constants that will be used during execution of the kernel. Prefetch operations for the kernel can then occur concurrently with thread dispatch operations.

Abstract: A mechanism is described for facilitating efficient prediction of most commonly occurring values in data blocks in computing environments. An apparatus of embodiments, as described herein, includes one or more processors to perform parallel calculations on values associated with multiple sub-blocks of a data block, and predict, based on the parallel calculations, a most commonly-occurring value in the data block. The apparatus if further to classify the most commonly-occurring value as a mode value for one or more data types to be used with one or more applications.

Abstract: Embodiments are generally directed to compression in machine learning and deep learning processing. An embodiment of an apparatus for compression of untyped data includes a graphical processing unit (GPU) including a data compression pipeline, the data compression pipeline including a data port coupled with one or more shader cores, wherein the data port is to allow transfer of untyped data without format conversion, and a 3D compression/decompression unit to provide for compression of untyped data to be stored to a memory subsystem and decompression of untyped data from the memory subsystem.

Abstract: A mechanism is described for facilitating efficient prediction of most commonly occurring values in data blocks in computing environments. An apparatus of embodiments, as described herein, includes one or more processors to perform parallel calculations on values associated with multiple sub-blocks of a data block, and predict, based on the parallel calculations, a most commonly-occurring value in the data block. The apparatus if further to classify the most commonly-occurring value as a mode value for one or more data types to be used with one or more applications.

Abstract: Embodiments described herein provide techniques to maintain high temporal cache locality between independent threads having the same or similar memory access pattern. One embodiment provides a graphics processing unit comprising an instruction execution pipeline including hardware execution logic and a thread dispatcher to process a set of commands for execution and distribute multiple groups of hardware threads to the hardware execution logic to execute the set of commands. The thread dispatcher can be configured to concurrently distribute a first group of the multiple groups of hardware threads to the hardware execution logic and withhold distribution of additional hardware threads for the set of commands until after the first group completes execution.

Abstract: Embodiments are generally directed to compression in machine learning and deep learning processing. An embodiment of an apparatus for compression of untyped data includes a graphical processing unit (GPU) including a data compression pipeline, the data compression pipeline including a data port coupled with one or more shader cores, wherein the data port is to allow transfer of untyped data without format conversion, and a 3D compression/decompression unit to provide for compression of untyped data to be stored to a memory subsystem and decompression of untyped data from the memory subsystem.

Abstract: The present disclosure includes devices and methods for operating resistance variable memory. One device embodiment includes an array of memory cells wherein a number of the cells are commonly coupled to a select line, the number cells including a number of data cells programmable within a number of target threshold resistance (Rt) ranges which correspond to a number of data states, and a number of reference cells interleaved with the data cells and programmable within the number of target Rt ranges. The aforementioned device embodiment also includes control circuitry coupled to the array and configured to sense a level associated with at least one data cell and at least one reference cell, and compare the sensed level associated with the at least one data cell with the sensed level associated with the at least one reference cell to determine a data state of the at least one data cell.

Abstract: The present disclosure includes devices and methods for operating resistance variable memory. One device embodiment includes an array of memory cells wherein a number of the cells are commonly coupled to a select line, the number cells including a number of data cells programmable within a number of target threshold resistance (Rt) ranges which correspond to a number of data states, and a number of reference cells interleaved with the data cells and programmable within the number of target Rt ranges. The aforementioned device embodiment also includes control circuitry coupled to the array and configured to sense a level associated with at least one data cell and at least one reference cell, and compare the sensed level associated with the at least one data cell with the sensed level associated with the at least one reference cell to determine a data state of the at least one data cell.

Abstract: The present disclosure includes devices and methods for operating resistance variable memory. One device embodiment includes an array of memory cells wherein a number of the cells are commonly coupled to a select line, the number cells including a number of data cells programmable within a number of target threshold resistance (Rt) ranges which correspond to a number of data states, and a number of reference cells interleaved with the data cells and programmable within the number of target Rt ranges. The aforementioned device embodiment also includes control circuitry coupled to the array and configured to sense a level associated with at least one data cell and at least one reference cell, and compare the sensed level associated with the at least one data cell with the sensed level associated with the at least one reference cell to determine a data state of the at least one data cell.

Abstract: The present disclosure includes devices and methods for operating resistance variable memory cells. One or more embodiments include applying a programming signal to a resistance variable material of a memory cell, and decreasing a magnitude of a trailing portion of the applied programming signal successively according to a number of particular decrements. The magnitude and the duration of the number of particular decrements correspond to particular programmed values.