1. PStack summary Summarizes the PStack approach, and includes measurements of proto-runtime versus Linux threads.

1. Summarizes the PStack approach, and includes measurements of proto-runtime versus Linux threads.

1. Position: It Is Better to Support Runtime Systems than Individual Parallelism Constructs Gives reasons why runtime support is better than hardwiring individual parallelism constructs, and introduces architecture concepts for communications, memory models, and address spaces driven by results from the computation model and the language research.

Position: It Is Better to Support Runtime Systems than Individual Parallelism Constructs Gives reasons why runtime support is better than hardwiring individual parallelism constructs, and introduces architecture concepts for communications, memory models, and address spaces driven by results from the computation model and the language research.

1. Integrated Performance Tuning Using Semantic Information Collected by Instrumenting the Language Runtime The first paper that mentions the basic parallel computation model, in the context of using it as a tool to gain insight while performance tuning applications. It takes a long time to load in the browser because the figures use real data collected during a run, and the pdf is 10 MB, so patience is needed ; )

1. The DKU Pattern for Performance Portable Parallelism The first paper submitted for publication on the DKU pattern, submitted to EuroPar09. It provides an introduction to the DKU pattern and relevant background. It also describes BLIS, the automated infrastructure that specializes a single source bundle to multiple hardware platforms.
2. A 5-page description of the project proposal submitted to the EU commission to develop the portability stack for parallel software.

1. Dissertation: A Study of Frameworks for Collectively Meeting the Productivity, Portability, and Adoptability Goals for Parallel Software This PhD Dissertation was published in 2011 by Sean Halle. It explains the problem being solved by parallelism research, and introduces new computation models as a basis for understanding parallel computation. The next-to-last 2 chapters cover the VMS proto-runtime approach. Chapters 6 and 7 are related to the model, and chapter 8 is the first publication of the basic elements of the model. 2, 3, and 4 are useful for understanding the base for the portability infrastructure.
2. Support for the Collective Effort Towards Portable Performance The first paper published that mentions VMS, in HotPar 2011. It is a position paper arguing for a segment-wide infrastructure to support portability. VMS is mentioned as a candidate to base this on.

1. A Mutable Hardware Abstraction to Replace Threads The first paper to give a detailed specification of VMS, in LCPC 2011. It gives technical depth and early measurements of time to create a plugin, VMS overhead, and performance comparisons.
2. Integrated Performance Tuning Using Semantic Information Collected by Instrumenting the Language Runtime The first paper that mentions the model, in the context of using it as a tool to gain insight while performance tuning applications. It takes a long time to load in the browser because the figures use real data collected during a run, and the pdf is 10 MB, so patience is needed ; )

1. HWSim reference manual Gives an introduction, the philosophy (structure) of the language, and a list of constructs and how to use them

1. Dissertation: A Study of Frameworks for Collectively Meeting the Productivity, Portability, and Adoptability Goals for Parallel Software

This PhD Dissertation was published in 2011 by Sean Halle. It explains the problem being solved by parallelism research, and introduces new computation models as a basis for understanding parallel computation. The next-to-last 2 chapters cover the VMS proto-runtime approach. Chapters 6 and 7 are related to the model, and chapter 8 is the first publication of the basic elements of the model. 2, 3, and 4 are useful for understanding the base for the portability infrastructure.

1. Support for the Collective Effort Towards Portable Performance

1. A Mutable Hardware Abstraction to Replace Threads

1. Integrated Performance Tuning Using Semantic Information Collected by Instrumenting the Language Runtime This is the first paper that mentions the model, in the context of using it as a tool to gain insight while performance tuning applications. It takes a long time to load in the browser because the figures use real data collected during a run, and the pdf is 10 MB, so patience is needed ; )

1. HWSim reference manual Gives an introduction, the philosophy (structure) of the language, and a list of constructs and how to use them
2. The DKU Pattern for Performance Portable Parallelism Δ The first paper submitted for publication on the DKU pattern, submitted to EuroPar09. It provides an introduction to the DKU pattern and relevant background. It also describes BLIS, the automated infrastructure that specializes a single source bundle to multiple hardware platforms.
3. A 5-page description of the project proposal Δ submitted to the EU commission to develop the portability stack for parallel software.

This is the first paper published that mentions VMS, in HotPar 2011. It is a position paper arguing for a segment-wide infrastructure to support portability. VMS is mentioned as a candidate to base this on.

A Mutable Hardware Abstraction to Replace Threads The first paper to give a detailed specification of VMS, in LCPC 2011. It gives technical depth and early measurements of time to create a plugin, VMS overhead, and performance comparisons.

Dissertation: A Study of Frameworks for Collectively Meeting the Productivity, Portability, and Adoptability Goals for Parallel Software This PhD Dissertation was published in 2011. It explains the problem being solved by parallelism research, and introduces new computation models as a basis for understanding parallel computation. The next-to-last 2 chapters cover VMS. Chapters 6, 7, and 8 are useful background for understanding VMS, and 2, 3, and 4 are useful for understanding the use of VMS as a base for portability infrastructure..

Integrated Performance Tuning Using Semantic Information Collected by Instrumenting the Language Runtime This is the first paper that mentions the model, in the context of using it as a tool to gain insight while performance tuning applications. It takes a long time to load in the browser because the figures use real data collected during a run, and the pdf is 10 MB, so patience is needed ; )

Dissertation: A Study of Frameworks for Collectively Meeting the Productivity, Portability, and Adoptability Goals for Parallel Software

HWSim reference manual Gives an introduction, the philosophy (structure) of the language, and a list of constructs and how to use them

HWSim reference manual

Support for the Collective Effort Towards Portable Performance

Dissertation:

Integrated Performance Tuning Using Semantic Information Collected by Instrumenting the Language Runtime

Dissertation:

• The DKU Pattern for Performance Portable Parallelism Δ

This is the first paper submitted for publication on the DKU pattern, submitted to EuroPar09. It provides an introduction to the DKU pattern and relevant background. It also describes the automated infrastructure that specializes a single source bundle to multiple hardware platforms.

• 5-page description of a project proposal Δ submitted to the EU commission.

List of all OSRI related papers