Where is time spent in message-passing and shared-memory programs?
Satish Chandra, James R. Larus, Anne E. Rogers
Abstract
Satish Chandra, James R. Larus, Anne E. Rogers
Abstract
Message passing and shared memory are two techniques parallel programs use for coordination and communication. This paper studies the strengths and weaknesses of these two mechanisms by comparing equivalent, well-written message-passing and shared-memory programs running on similar hardware. To ensure that our measurements are comparable, we produced two carefully tuned versions of each program and measured them on closely-related simulators of a message-passing and a shared-memory machine, both of which are based on same underlying hardware assumptions.
OpenAlex reports 47 citations for this work. Citation counts describe recorded attention and do not establish research quality.
A contribution statement is not available in the OpenAlex record.
Method details are not available in the OpenAlex metadata.
Findings are not separately available in the OpenAlex metadata.
Limitations are not available in the OpenAlex metadata.
Application details are not available in the OpenAlex metadata.
Message passing and shared memory are two techniques parallel programs use for coordination and communication. This paper studies the strengths and weaknesses of these two mechanisms by comparing equivalent, well-written message-passing and shared-memory programs running on similar hardware. To ensure that our measurements are comparable, we produced two carefully tuned versions of each program and measured them on closely-related simulators of a message-passing and a shared-memory machine, both of which are based on same underlying hardware assumptions.
Key concepts: Message passing, Computer science, Shared memory, Distributed memory, Message broker, Parallel computing, Distributed shared memory, Memory management