Abstract: Disclosed herein are systems, apparatuses, and methods for enabling efficient reads to a local memory of a processing unit. In an embodiment, a processing unit includes an interface and a buffer. The interface is configured to (i) send a request for a portion of data in a region of a local memory of an other processing unit and (ii) receive, responsive to the request, all the data from the region. The buffer is configured to store the data from the region of the local memory of the other processing unit.

Abstract: A central processor unit (CPU) is connected to a system/graphics controller generally comprising a monolithic semiconductor device. The system/graphics controller is connected to an input output (IO) controller via a high-speed PCI bus. The IO controller interfaces to the system graphics controller via the high-speed PCI bus. The IO controller includes a lower speed PCI port controlled by an arbiter within the IO controller. Generally, the low speed PCI arbiter of the IO controller will interface to standard 33 MHz PCI cards. In addition, the IO controller interfaces to an external storage device, such as a hard drive, via either a standard or a proprietary bus protocol. A unified system/graphics memory which is accessed by the system/graphics controller. The unified memory contains both system data and graphics data. In a specific embodiment, two channels, CH0 and CH1 access the unified memory.

Abstract: In a hardware-based virtualization system, a hypervisor switches out of a first function into a second function. The first function is one of a physical function and a virtual function and the second function is one of a physical function and a virtual function. During the switching a malfunction of the first function is detected. The first function is reset without resetting the second function. The switching, detecting, and resetting operations are performed by a hypervisor of the hardware-based virtualization system. Embodiments further include a communication mechanism for the hypervisor to notify a driver of the function that was reset to enable the driver to restore the function without delay.

Abstract: The present system enables passing a pointer, associated with accessing data in a memory, to an input/output (I/O) device via an input/output memory management unit (IOMMU). The I/O device accesses the data in the memory via the IOMMU without copying the data into a local I/O device memory. The I/O device can perform an operation on the data in the memory based on the pointer, such that I/O device accesses the memory without expensive copies.

Abstract: A method for power management is disclosed. The method may include monitoring requests for access to a memory of a memory subsystem by one or more processor cores; and monitoring requests for access to the memory conveyed by an input/output (I/O) unit. The method may further include determining if at least a first amount of time has elapsed since any one of the processor cores has asserted a memory access request and determining if at least a second amount of time has elapsed since the I/O unit has conveyed a memory access request. A first signal may be asserted if the first and second amounts of time have elapsed. A memory subsystem may be transitioned from operating in a full power state to a first low power state responsive to assertion of the first signal.

Abstract: An input/output memory management unit (IOMMU) having an “invalidate all” command available to clear the contents of cache memory is presented. The cache memory provides fast access to address translation data that has been previously obtained by a process. A typical cache memory includes device tables, page tables and interrupt remapping entries. Cache memory data can become stale or be compromised from security breaches or malfunctioning devices. In these circumstances, a rapid approach to clearing cache memory content is provided.

Abstract: Methods and systems are provided for mapping a memory instruction to a shared memory address space in a computer arrangement having a CPU and an APD. A method includes receiving a memory instruction that refers to an address in the shared memory address space, mapping the memory instruction based on the address to a memory resource associated with either the CPU or the APD, and performing the memory instruction based on the mapping.

Abstract: A method and system for allocating memory to a memory operation executed by a processor in a computer arrangement having a first processor configured for unified operation with a second processor. The method includes receiving a memory operation from a processor and mapping the memory operation to one of a plurality of memory heaps. The mapping produces a mapping result. The method also includes providing the mapping result to the processor.

Abstract: Embodiments of the present invention provide a method of a first processor using a memory resource associated with a second processor. The method includes receiving a memory instruction from a first processor process, wherein the memory instruction refers to a shared memory address (SMA) that maps to a second processor memory. The method also includes mapping the SMA to the second processor memory, wherein the mapping produces a mapping result and providing the mapping result to the first processor.

Abstract: Provided is a method of permitting the reordering of a visibility order of operations in a computer arrangement configured for permitting a first processor and a second processor threads to access a shared memory. The method includes receiving in a program order, a first and a second operation in a first thread and permitting the reordering of the visibility order for the operations in the shared memory based on the class of each operation. The visibility order determines the visibility in the shared memory, by a second thread, of stored results from the execution of the first and second operations.

Abstract: Existing multiprocessor computing systems often have insufficient memory coherency and, consequently, are unable to efficiently utilize separate memory systems. Specifically, a CPU cannot effectively write to a block of memory and then have a GPU access that memory unless there is explicit synchronization. In addition, because the GPU is forced to statically split memory locations between itself and the CPU, existing multiprocessor computing systems are unable to efficiently utilize the separate memory systems. Embodiments described herein overcome these deficiencies by receiving a notification within the GPU that the CPU has finished processing data that is stored in coherent memory, and invalidating data in the CPU caches that the GPU has finished processing from the coherent memory. Embodiments described herein also include dynamically partitioning a GPU memory into coherent memory and local memory through use of a probe filter.

Abstract: The present system enables receiving a request from an I/O device to translate a virtual address to a physical address to access the page in system memory. One or more memory attributes of the page defining a cacheability characteristic of the page is identified. A response including the physical address and the cacheability characteristic of the page is sent to the I/O device.

Abstract: Embodiments of the present invention provides for the execution of threads and/or workitems on multiple processors of a heterogeneous computing system in a manner that they can share data correctly and efficiently. Disclosed method, system, and article of manufacture embodiments include, responsive to an instruction from a sequence of instructions of a work-item, determining an ordering of visibility to other work-items of one or more other data items in relation to a particular data item, and performing at least one cache operation upon at least one of the particular data item or the other data items present in any one or more cache memories in accordance with the determined ordering. The semantics of the instruction includes a memory operation upon the particular data item.

Abstract: A method and apparatus for repopulating a cache are disclosed. At least a portion of the contents of the cache are stored in a location separate from the cache. Power is removed from the cache and is restored some time later. After power has been restored to the cache, it is repopulated with the portion of the contents of the cache that were stored separately from the cache.

Abstract: A method for changing between virtual machines on a graphics processing unit (GPU) includes requesting to switch from a first virtual machine (VM) with a first global context to a second VM with a second global context; stopping taking of new commands in the first VM; saving the first global context; and switching out of the first VM.

Abstract: The present system enables an input/output (I/O) device to request memory for performing a direct memory access (DMA) of system memory. Further, the system uses an input/output memory management unit (IOMMU) to determine whether or not the system memory is available. The IOMMU notifies an operating system associated with the system memory if the system memory is not available, such that the operating system allocates non-system memory for use by the I/O device to perform the DMA.

Abstract: The present system enables passing a pointer, associated with accessing data in a memory, to an input/output (I/O) device via an input/output memory management unit (IOMMU). The I/O device accesses the data in the memory via the IOMMU without copying the data into a local I/O device memory. The I/O device can perform an operation on the data in the memory based on the pointer, such that I/O device accesses the memory without expensive copies.

Abstract: A method for power management is disclosed. The method may include monitoring requests for access to a memory of a memory subsystem by one or more processor cores; and monitoring requests for access to the memory conveyed by an input/output (I/O) unit. The method may further include determining if at least a first amount of time has elapsed since any one of the processor cores has asserted a memory access request and determining if at least a second amount of time has elapsed since the I/O unit has conveyed a memory access request. A first signal may be asserted if the first and second amounts of time have elapsed. A memory subsystem may be transitioned from operating in a full power state to a first low power state responsive to assertion of the first signal.

Abstract: A method and apparatus for repopulating a cache are disclosed. At least a portion of the contents of the cache are stored in a location separate from the cache. Power is removed from the cache and is restored some time later. After power has been restored to the cache, it is repopulated with the portion of the contents of the cache that were stored separately from the cache.

Abstract: Described are a system and method for broadcasting write requests to a plurality of graphics devices. A different address range of graphics device addresses is associated with each graphics device of the plurality of graphics devices. A controller receives a write request directed to a memory address and generates a plurality of graphics device addresses based on the memory address of the write request when the memory address is within a particular range of broadcast addresses. An offset may be applied to a reference address in each address range associated with one of the graphics devices when generating the plurality of graphics device addresses. The write request is forwarded to each graphics device of the plurality of graphics devices associated with one of the generated graphics device addresses.