Odin is a new high performance single address space multiprocessor design. The contribution of this investigation is the synthesis of three important new methods into a unified system which maximises data locality and significantly reduces data access latencies. To achieve high performance Odin uses a segmented stack to maintain data locality after thread migration, and a memory mapping that discriminates between shared and local data structures. When these techniques fail to provide optimal locality high speed access to data is provided by a new hardware based fine grain consistency protocol. This approach reduces false sharing and network usage while improving data locality. Keywords: Single address-space, multiprocessor design, simulati...
We present design details and some initial performance results of a novel scalable shared memory mul...
The programming of parallel and distributed applications is difficult. The proliferation of net wor...
Effective memory hierarchy utilization is critical to the performance of modern multiprocessor archi...
Due to their excellent price-performance ratio, clusters built from commodity nodes have become broa...
grantor: University of TorontoThis dissertation presents novel operating system structurin...
The DASH research project is addressing the general problem of achiev-ing high-performance network c...
Seven distinct configurations of shared-memory multiprocessors are defined and parameterized in term...
Computing drives a lot of developments all around us, and leads to innovation in many fields of scie...
The memory system is a major bottleneck in achieving high performance and energy efficiency for vari...
{In a recent issue of Operating System Review, Hayter and McAuley [1991] argue that future high-perf...
The ideal memory system assumed by most programmers is one which has high capacity, yet allows any w...
To design effective large-scale multiprocessors, designers need to understand the characteristics of...
Distributed-memory multiprocessing systems (DMS), such as Intel’s hypercubes, the Paragon, Thinking ...
The Partitioned Global Address Space (PGAS) model is a parallel programming model that aims to im-pr...
The performance evaluation of multiprocessor interconnects cannot be divorced from issues of traffic...
We present design details and some initial performance results of a novel scalable shared memory mul...
The programming of parallel and distributed applications is difficult. The proliferation of net wor...
Effective memory hierarchy utilization is critical to the performance of modern multiprocessor archi...
Due to their excellent price-performance ratio, clusters built from commodity nodes have become broa...
grantor: University of TorontoThis dissertation presents novel operating system structurin...
The DASH research project is addressing the general problem of achiev-ing high-performance network c...
Seven distinct configurations of shared-memory multiprocessors are defined and parameterized in term...
Computing drives a lot of developments all around us, and leads to innovation in many fields of scie...
The memory system is a major bottleneck in achieving high performance and energy efficiency for vari...
{In a recent issue of Operating System Review, Hayter and McAuley [1991] argue that future high-perf...
The ideal memory system assumed by most programmers is one which has high capacity, yet allows any w...
To design effective large-scale multiprocessors, designers need to understand the characteristics of...
Distributed-memory multiprocessing systems (DMS), such as Intel’s hypercubes, the Paragon, Thinking ...
The Partitioned Global Address Space (PGAS) model is a parallel programming model that aims to im-pr...
The performance evaluation of multiprocessor interconnects cannot be divorced from issues of traffic...
We present design details and some initial performance results of a novel scalable shared memory mul...
The programming of parallel and distributed applications is difficult. The proliferation of net wor...
Effective memory hierarchy utilization is critical to the performance of modern multiprocessor archi...