Intel Corporation

William Rash

Martin Guy Dixon

Yazmin A Santiago

Subramaniam M Maiyuran

Varghese George

Rajesh M Sankaran

Bret L Toll

Glenn J Hinton

Edward Thomas Grochowski

Nhon Toai Quach

Elmoustapha Ould-Ahmed-Vall

David B Papworth

Scott D Hahn

Bharath Muthiah

Hang T Nguyen

Gang Y Chen

Andres Rabago

Scott Hayn

Gang Chen

Graphics processing systems and methods are described. A graphics processing apparatus may comprise one or more graphics processing engines, a memory, a memory management unit (MMU) including a GPU second level page table and GPU dirty bit tracking, and a provisioning agent to receive a request from a virtual machine monitor (VMM) to provision a subcluster of graphics processing apparatuses, the subcluster including a plurality of graphics processing engines from a plurality of graphics processing apparatuses connected using a scale-up fabric, provision the scale-up fabric to route data within the subcluster of graphics processing apparatuses, and provision a plurality of resources on the graphics processing apparatus for the subcluster based on the request from the VMM.

Virtualization and multi-tenancy support in graphics processors

Disclosed embodiments relate to systems and methods to skip inconsequential matrix operations. In one example, a processor includes decode circuitry to decode an instruction having fields to specify an opcode and locations of first source, second source, and destination matrices, the opcode indicating that the processor is to multiply each element at row M and column K of the first source matrix with a corresponding element at row K and column N of the second source matrix, and accumulate a resulting product with previous contents of a corresponding element at row M and column N of the destination matrix, the processor to skip multiplications that, based on detected values of corresponding multiplicands, would generate inconsequential results; scheduling circuitry to schedule execution of the instruction; and execution circuitry to execute the instructions as per the opcode.

Systems and methods to skip inconsequential matrix operations

Disclosed embodiments relate to systems and methods to skip inconsequential matrix operations. In one example, a processor includes decode circuitry to decode an instruction having fields to specify an opcode and locations of first source, second source, and destination matrices, the opcode indicating that the processor is to multiply each element at row M and column K of the first source matrix with a corresponding element at row K and column N of the second source matrix, and accumulate a resulting product with previous contents of a corresponding element at row M and column N of the destination matrix, the processor to skip multiplications that, based on detected values of corresponding multiplicands, would generate inconsequential results, scheduling circuitry to schedule execution of the instruction; and execution circuitry to execute the instructions as per the opcode.

In one embodiment, Quality of Service (QoS) criteria based server side binary translation and execution of applications is performed on multiple servers utilizing distributed translation and execution in either a virtualized or native execution environment. The translated applications are executed to generate output display data, the output display data is encoded in a media format suitable for video streaming, and the video stream is delivered over a network to a client device. In one embodiment, one or more graphics processors assist the central processors of the servers by accelerating the rendering of the application output, and a media encoder encodes the application output into a media format.

QoS based binary translation and application streaming

A processor of an aspect includes a plurality of logical processors. A first logical processor of the plurality is to execute software that includes a memory access synchronization instruction that is to synchronize accesses to a memory. The processor also includes memory access synchronization relaxation logic that is to prevent the memory access synchronization instruction from synchronizing accesses to the memory when the processor is in a relaxed memory access synchronization mode.

Processors, methods, and systems to relax synchronization of accesses to shared memory

In an embodiment, a processor includes a plurality of cores. Each core includes a core power unit to detect one or more power management events, and in response to the one or more power management events, initiate a protected power management mode in the core. Software interrupts to the core may be disabled during the protected power management mode. The core is to execute power management code during the protected power management mode. Other embodiments are described and claimed.

Protected power management mode in a processor

A processor of an aspect includes decode logic to receive a first instruction and to determine that the first instruction is to be emulated. The processor also includes emulation mode aware post-decode instruction processor logic coupled with the decode logic. The emulation mode aware post-decode instruction processor logic is to process one or more control signals decoded from an instruction. The instruction is one of a set of one or more instructions used to emulate the first instruction. The one or more control signals are to be processed differently by the emulation mode aware post-decode instruction processor logic when in an emulation mode than when not in the emulation mode. Other apparatus are also disclosed as well as methods and systems.

Instruction emulation processors, methods, and systems

A method of an aspect is performed by a processor. The method includes receiving a partial width load instruction. The partial width load instruction indicates a memory location of a memory as a source operand and indicates a register as a destination operand. The method includes loading data from the indicated memory location to the processor in response to the partial width load instruction. The method includes writing at least a portion of the loaded data to a partial width of the register in response to the partial width load instruction. The method includes finishing writing the register with a set of bits stored in a remaining width of the register that have bit values that depend on a partial width load mode of the processor. The partial width load instruction does not indicate the partial width load mode. Other methods, processors, and systems are also disclosed.

Mode dependent partial width load to wider register processors, methods, and systems

A processor of an aspect includes an instruction fetch unit to fetch a pair of instruction order enforcement instructions. The pair of instruction order enforcement instructions are part of an instruction set of the processor. The pair of instruction order enforcement instructions includes an activation instruction and an enforcement instruction. The activation instruction is to occur before the enforcement instruction in a program order. The processor also includes an instruction order enforcement module. The instruction order enforcement module, in response to the pair of the instruction order enforcement instructions, is to prevent instructions occurring after the enforcement instruction in the program order, from being processed prior to the activation instruction, in an out-of-order portion of the processor. Other processors are also disclosed, as are various methods, systems, and instructions.

Instruction order enforcement pairs of instructions, processors, methods, and systems

Embodiments of an invention for inter-processor attestation hardware are disclosed. In one embodiment, an apparatus includes first attestation hardware associated with a first portion of a system. The first attestation hardware is to attest to a second portion of the system that the first portion of the system is secure.

Inter-processor attestation hardware

A processor is provided that implements a replay mechanism to recover from soft errors. The processor includes a protected execution unit, a check unit to detect errors in results generated by the protected execution unit, and a replay unit to track selected instructions issued to the protected execution unit. When the check unit detects an error, it triggers the replay unit to reissue the selected instructions to the protected execution unit. One embodiment of the replay unit provides an instruction buffer that includes pointers to track issue and retirement status of in-flight instructions. When the check unit indicates an error, the replay unit resets a pointer to reissue the instruction for which the error was detected.

Growing community of inventors

Saratoga, CA, United States of America

William Rash