Intel Corporation

Microsoft Technology Licensing, LLC

Xiangyao Yu

Christopher J Hughes

Thomas Moscibroda

Nadathur Rajagopalan Satish

Feng Zhao

Mark Eugene Russinovich

Jiaxing Zhang

Zhengping Qian

Two techniques address bottlenecking in processors. The first is indirect prefetching. The technique can be especially useful for graph analytics and sparse matrix applications. For graph analytics and sparse matrix applications, the addresses of most random memory accesses come from an index array B which is sequentially scanned by an application. The random accesses are actually indirect accesses in the form A[B[i]]. A hardware component is introduced to detect this pattern. The hardware can then read B a certain distance ahead, and prefetch the corresponding element in A. For example, if the 'prefetch distance' is k, when B[i] is accessed, the hardware reads B[i+k], and then A[B[i+k]. For partial cacheline accessing, the indirect accesses are usually accessing random memory locations and only accessing a small portion of a cacheline. Instead of loading the whole cacheline into L1 cache, the second technique only loads a part of the cacheline.

Hardware prefetcher for indirect access patterns

The techniques and arrangements described herein provide for updating services, host operating systems and other applications while satisfying update domain constraints. In some examples, one or more controller modules may maintain a data structure including a plurality of server update domains, each server update domain including a set of machines of a plurality of machines of a distributed computing system which may be concurrently updated. The one or more controller modules may allocate the plurality of instances to the plurality of machines such that a number of server update domains is minimized.

Service allocation in a distributed computing platform

Growing community of inventors

Cambridge, MA, United States of America

Xiangyao Yu