Intel Corporation

Nicolas C Galoppo Von Borries

Altug Koker

Abhishek R Appu

Joydeep Ray

Kamal Sinha

Balaji Vembu

Eriko Nurvitadhi

Tsung-Han Lin

Rajkishore Barik

Elmoustapha Ould-Ahmed-Vall

Subramaniam M Maiyuran

Ben J Ashbaugh

Vasanth Ranganathan

Prasoonkumar Surti

Feng Chen

Narayan Srinivasa

Sara S Baghsorkhi

Aravindh V Anantaraman

Mike MacPherson

Anbang Yao

Xiaoming Chen

Valentin Andrei

Varghese George

Justin E Gottschlich

Dukhwan Kim

Nadathur Rajagopalan Satish

Farshad Akhbari

Sanjeev S Jahagirdar

Linda L Hurd

Tatiana Shpeisman

Ping Tak Tang

Michael S Strickland

Travis T Schluessler

Karthik Vaidyanathan

John Charles Weast

Adam T Lake

John H Feit

Jeffery S Boles

Devan Burke

Wenyin Fu

Josh B Mastronarde

Sungye Kim

Murali Ramadoss

Sanu K Mathew

Mark A Anders

Lakshminarayanan Striramassarma

Himanshu Kaul

Vikranth Vemulapalli

Kevin Nealis

Jonathan D Pearce

Scott R Janus

Brent E Insko

Arthur Hunter, Jr

William B Sadler

Chandrasekaran Sakthivel

Jeremy Bottleson

Ankur N Shah

David M Puffer

Marian Alin Petre

James A Valerio

Pattabhiraman K

Liwei Ma

Mike B Macpherson

Barath Lakshmanan

Yoav Harel

Saurabh Tangri

Shailesh Shah

Rajikshore Barik

Selvakumar Panneer

Niranjan L Cooray

William A Hux

Sudhakar Kamma

Jayakrishna P S

Nilay Mistry

Robert M Ioffe

Niran Cooray

K Pattabhiraman

One embodiment provides for a compute apparatus comprising a decode unit to decode a single instruction into a decoded instruction that specifies multiple operands including a multi-bit input value and a one-bit weight associated with a neural network, as well as an arithmetic logic unit including a multiplier, an adder, and an accumulator register. To execute the decoded instruction, the multiplier is to perform a fused operation including an exclusive not OR (XNOR) operation and a population count operation. The adder is configured to add the intermediate product to a value stored in the accumulator register and update the value stored in the accumulator register.

Compute optimizations for neural networks

An apparatus to facilitate optimization of a neural network (NN) is disclosed. The apparatus includes optimization logic to define a NN topology having one or more macro layers, adjust the one or more macro layers to adapt to input and output components of the NN and train the NN based on the one or more macro layers.

Neural network optimization mechanism

Embodiments are generally directed to memory prefetching in multiple GPU environment. An embodiment of an apparatus includes multiple processors including a host processor and multiple graphics processing units (GPUs) to process data, each of the GPUs including a prefetcher and a cache; and a memory for storage of data, the memory including a plurality of memory elements, wherein the prefetcher of each of the GPUs is to prefetch data from the memory to the cache of the GPU; and wherein the prefetcher of a GPU is prohibited from prefetching from a page that is not owned by the GPU or by the host processor.

Prefetch status notification for memory prefetching

In an example, an apparatus comprises a compute engine comprising a high precision component and a low precision component; and logic, at least partially including hardware logic, to receive instructions in the compute engine; select at least one of the high precision component or the low precision component to execute the instructions; and apply a gate to at least one of the high precision component or the low precision component to execute the instructions. Other embodiments are also disclosed and claimed.

Dynamic precision for neural network compute operations

Embodiments described herein provide techniques to facilitate instruction-based control of memory attributes. One embodiment provides a graphics processor comprising a processing resource, a memory device, a cache coupled with the processing resources and the memory, and circuitry to process a memory access message received from the processing resource. The memory access message enables access to data of the memory device. To process the memory access message, the circuitry is configured to determine one or more cache attributes that indicate whether the data should be read from or stored the cache. The cache attributes may be provided by the memory access message or stored in state data associated with the data to be accessed by the access message.

Implicit fence for write messages

Apparatuses including general-purpose graphics processing units having on chip dense memory for temporal buffering are disclosed. In one embodiment, a graphics multiprocessor includes a plurality of compute engines to perform first computations to generate a first set of data, cache for storing data, and a high density memory that is integrated on chip with the plurality of compute engines and the cache. The high density memory to receive the first set of data, to temporarily store the first set of data, and to provide the first set of data to the cache during a first time period that is prior to a second time period when the plurality of compute engines will use the first set of data for second computations.

On chip dense memory for temporal buffering

Methods and apparatus relating to data initialization techniques. In an example, an apparatus comprises a processor to read one or more metadata codes which map to one or more cache lines in a cache memory and invoke a random number generator to generate random numerical data for the one or more cache lines in response to a determination that the one more metadata codes indicate that the cache lines are to contain random numerical data. Other embodiments are also disclosed and claimed.

Data initialization techniques

One embodiment provides for a graphics processing unit to accelerate machine-learning operations, the graphics processing unit comprising a multiprocessor having a single instruction, multiple thread (SIMT) architecture, the multiprocessor to execute at least one single instruction; and a first compute unit included within the multiprocessor, the at least one single instruction to cause the first compute unit to perform a two-dimensional matrix multiply and accumulate operation, wherein to perform the two-dimensional matrix multiply and accumulate operation includes to compute an intermediate product of 16-bit operands and to compute a 32-bit sum based on the intermediate product.

Instructions and logic to perform floating point and integer operations for machine learning

A mechanism is described for facilitating intelligent thread scheduling at autonomous machines. A method of embodiments, as described herein, includes detecting dependency information relating to a plurality of threads corresponding to a plurality of workloads associated with tasks relating to a processor including a graphics processor. The method may further include generating a tree of thread groups based on the dependency information, where each thread group includes multiple threads, and scheduling one or more of the thread groups associated a similar dependency to avoid dependency conflicts.

Efficient thread group scheduling

Systems and methods for improving cache efficiency and utilization are disclosed. In one embodiment, a graphics processor includes processing resources to perform graphics operations and a cache controller of a cache coupled to the processing resources. The cache controller is configured to control cache priority by determining whether default settings or an instruction will control cache operations for the cache.

Systems and methods for improving cache efficiency and utilization

Embodiments are generally directed to graphics processor data access and sharing. An embodiment of an apparatus includes a circuit element to produce a result in processing of an application; a load-store unit to receive the result and generate pre-fetch information for a cache utilizing the result; and a prefetch generator to produce prefetch addresses based at least in part on the pre-fetch information; wherein the load-store unit is to receive software assistance for prefetching, and wherein generation of the pre-fetch information is based at least in part on the software assistance.

Graphics processor data access and sharing

Embodiments provide mechanisms to facilitate compute operations for deep neural networks. One embodiment comprises a graphics processing unit comprising one or more multiprocessors, at least one of the one or more multiprocessors including a register file to store a plurality of different types of operands and a plurality of processing cores. The plurality of processing cores includes a first set of processing cores of a first type and a second set of processing cores of a second type. The first set of processing cores are associated with a first memory channel and the second set of processing cores are associated with a second memory channel.

Compute optimization mechanism for deep neural networks

Methods and apparatus relating to techniques for multi-tile memory management. In an example, a graphics processor includes an interposer, a first chiplet coupled with the interposer, the first chiplet including a graphics processing resource and an interconnect network coupled with the graphics processing resource, cache circuitry coupled with the graphics processing resource via the interconnect network, and a second chiplet coupled with the first chiplet via the interposer, the second chiplet including a memory-side cache and a memory controller coupled with the memory-side cache. The memory controller is configured to enable access to a high-bandwidth memory (HBM) device, the memory-side cache is configured to cache data associated with a memory access performed via the memory controller, and the cache circuitry is logically positioned between the graphics processing resource and a chiplet interface.

Multi-tile memory management

One embodiment provides for a graphics processing unit to accelerate machine-learning operations, the graphics processing unit comprising a multiprocessor having a single instruction, multiple thread (SIMT) architecture, the multiprocessor to execute at least one single instruction; and a first compute unit included within the multiprocessor, the at least one single instruction to cause the first compute unit to perform a two-dimensional matrix multiply and accumulate operation, wherein to perform the two-dimensional matrix multiply and accumulate operation includes to compute a 32-bit intermediate product of 16-bit operands and to compute a 32-bit sum based on the 32-bit intermediate product.

Techniques to improve performance of matrix multiply operations are described in which a compute kernel can specify one or more element-wise operations to perform on output of the compute kernel before the output is transferred to higher levels of a processor memory hierarchy.

Machine learning sparse computation mechanism

Growing community of inventors

Portland, OR, United States of America

Nicolas C Galoppo Von Borries