Benedict Gaster's Outputs (26)

Allocating memory and using the allocated memory in a workgroup in a dispatched data parallel kernel (2016)
Patent

In a computing system, memory may be managed by using a distributed array, which is a global set of local memory regions. A segment in the distributed array is allocated and is bound to a physical memory region. The segment is used by a workgroup in... Read More about Allocating memory and using the allocated memory in a workgroup in a dispatched data parallel kernel.

HSA queuing model (2015)
Book Chapter

HSA is a system architecture design for modern computing, integrating CPUs, GPUs, and other accelerators into a single shared memory system. To enable low-latency dispatch between different agents (devices) HSA provides architected queues, which enab... Read More about HSA queuing model.

HSA memory model (2015)
Book Chapter

The HSA architecture has adopted a variant of the HRF-Relaxed memory model, which allows memory accesses to be controlled via fine grain scopes. This chapter introduces the HSA memory consistency model, how programmers can control memory visibility,... Read More about HSA memory model.

HRF-relaxed: Adapting HRF to the complexities of industrial heterogeneous memory models (2015)
Journal Article

© 2015 ACM. Memory consistency models, or memory models, allow both programmers and program language imple-menters to reason about concurrent accesses to one or more memory locations. Memory model specifications balance the often conflicting needs fo... Read More about HRF-relaxed: Adapting HRF to the complexities of industrial heterogeneous memory models.

Method for memory consistency among heterogeneous computer components (2014)
Patent

A method, computer program product, and system is described that determines the correctness of using memory operations in a computing device with heterogeneous computer components. Embodiments include an optimizer based on the characteristics of a Se... Read More about Method for memory consistency among heterogeneous computer components.

Heterogeneous-race-free memory models (2014)
Presentation / Conference Contribution

Commodity heterogeneous systems (e.g., integrated CPUs and GPUs), now support a unified, shared memory address space for all components. Because the latency of global communication in a heterogeneous system can be prohibitively high, heterogeneous sy... Read More about Heterogeneous-race-free memory models.