• No results found

Shared and distributed memory systems

Distributed shared memory on heterogeneous CPUs+GPUs platforms

Distributed shared memory on heterogeneous CPUs+GPUs platforms

... dedicated memory, general purpose computing lan- guages such as CUDA or OpenCL require the programmer to explicitly move data between the (CPU) host memory and the (GPU) device dedicated mem- ory, similarly ...

79

CONSISTENCY MODELS IN DISTRIBUTED SHARED MEMORY SYSTEMS

CONSISTENCY MODELS IN DISTRIBUTED SHARED MEMORY SYSTEMS

... and memory, some holding the address the CPU wants to read or write, some for sending or receiving data, and the rest for controlling the ...most systems, buses are external and are used to connect printed ...

6

Efficient Address Remapping in Distributed Shared-Memory Systems

Efficient Address Remapping in Distributed Shared-Memory Systems

... Active Memory architecture [Chaudhuri et ...Active Memory achieves similarly impressive performance gains as Impulse-MP while ensuring full coherence for remapped data at the cost of extra ...Active ...

21

Checkpointing Distributed Shared Memory

Checkpointing Distributed Shared Memory

... Independent checkpointing and message logging was not a very encouraging option, because DSM systems generate more messages than message-passing programs. Thus, we have chosen a coordinated checkpointing strategy ...

28

An asymmetric distributed shared memory model for heterogeneous parallel systems

An asymmetric distributed shared memory model for heterogeneous parallel systems

... accelerator memory, setting of memory page permission bits, transfer of data between system and accelerator memory, invoca- tion of kernels, and waiting for completion of accelerator execu- tion, I/O ...

12

Logging and Recovery in Adaptive Software Distributed Shared Memory Systems

Logging and Recovery in Adaptive Software Distributed Shared Memory Systems

... a shared memory page under the SW protocol do not create a copy of coherence data, and therefore, it is nec- essary for our AL protocol to produce a volatile log for such data at the sender ...whenever ...

10

Distributed Shared Memory Support in GMACS

Distributed Shared Memory Support in GMACS

... from systems from graphics vendors that expose low-level details of the underlying graphics hardware imple- mentation (CUDA [3], OpenCL [25]) to research and commercial higher-level languages and systems ...

83

Introduction of shared-memory parallelism in a distributed-memory multifrontal solver

Introduction of shared-memory parallelism in a distributed-memory multifrontal solver

... sparse systems of linear equations of the form Ax = b, (1) where A is a sparse matrix, b is the right-hand side vector (or matrix) and x is the unknown ...targeting distributed- memory computers have ...

39

Testing of a Novel Distributed Shared Memory Framework

Testing of a Novel Distributed Shared Memory Framework

... page-based systems by adopting a mode of operation where key variables are stored alone on one page, so that accessing them will not cause any difficulties in accessing other unrelated data ...in ...

98

Concordia: Distributed Shared Memory with In-Network Cache Coherence

Concordia: Distributed Shared Memory with In-Network Cache Coherence

... When a client issues a read request, the switch routes the request to a replica by a load-aware scheduling policy. When a client issues a write request, the switch resets the replica set to include only one server by a ...

17

Software Distributed Shared Memory Scalability and New Applications

Software Distributed Shared Memory Scalability and New Applications

... This project will be done in cooperation with Axis Communications AB. Axis has so far been a network client product company and now also want to expand in the technology sector. As part of this, they are interested in ...

5

REDAC: Distributed, Asynchronous Redundancy in Shared Memory Servers

REDAC: Distributed, Asynchronous Redundancy in Shared Memory Servers

... for memory model speculation using atomic sequence ordering (ASO) ...directory-based shared-memory multiprocessor system running Solaris ...phantom memory requests and synchronizing ...

12

Design of Fault-Tolerant Distributed Shared Memory Systems. Copenhagen Business School and University of California

Design of Fault-Tolerant Distributed Shared Memory Systems. Copenhagen Business School and University of California

... To enable full application recovery in a kernel-oriented DSM system, the system must provide mechanisms to indicate to the application level when checkpoints of non-DSM state should be performed by the application. Such ...

13

Using Memory-Mapped Network Interfaces to Improve the Performance of Distributed Shared Memory

Using Memory-Mapped Network Interfaces to Improve the Performance of Distributed Shared Memory

... local memory, and return to the ...local memory, to which evicted lines are written ...Cashmere systems shown in the figure are identical to those of figure 1, except for the ab- sence of the merge ...

12

System Software Support for Reducing Memory Latency on Distributed Shared Memory Multiprocessors

System Software Support for Reducing Memory Latency on Distributed Shared Memory Multiprocessors

... and memory modules, programs with dynamic and un- predictable memory access patterns, or even programs that migrate between processors due to the operating system scheduling ...remote memory accesses ...

8

Distributed shared memory architectures and global performance state estimation

Distributed shared memory architectures and global performance state estimation

... torus. Systems with more than one LC route packets between them over the Blink to the corrected LC according to a routing ...enables distributed routing of SCI packets between individual SCI rings without ...

199

Figure 1: Network architecture and distributed (shared-nothing) memory.

Figure 1: Network architecture and distributed (shared-nothing) memory.

... evenly distributed across the network, each machine holds (in its local disk) a subcollection whose size (in bytes) is roughly given by ...main memory 1 ...retrieval systems. Thus, we consider that ...

9

Sirocco: cost-effective fine-grain distributed shared memory

Sirocco: cost-effective fine-grain distributed shared memory

... Appbt and em3d significantly benefit from clustering across tag implementations. In appbt, clustering improves performance by reducing the number of remote accesses. The performance impact is more pronounced for ...

10

An Effective Approach of Data Security for
          Distributed Shared Memory Multiprocessors

An Effective Approach of Data Security for Distributed Shared Memory Multiprocessors

... traces. Active attacks that modify coherence messages and alter the behavior of the DSMs are less likely to be attempted, especially if the system is augmented with the ability to detect them. Therefore, we seek to ...

7

Performance Evaluation of the Cray X1 Distributed Shared Memory Architecture

Performance Evaluation of the Cray X1 Distributed Shared Memory Architecture

... consecutive memory locations) to saturate the vector units of the MSP, the Ecache is needed because the bandwidth to main memory is not enough to saturate the vector units without data ...is, memory ...

6

Show all 10000 documents...

Related subjects