Applications: All application run in their own respective activity and communicate
with the runtime system module over messages.
12.2 Concurrency Support for Resilient Computing
Two monitors as above FT Interrupt Handler and Main Monitor deal with
interruptions and resource handling, including reconfiguration of architecture and
software structure when required.
12.2.1 Problem of Concurrency
Real world is represented in our imagination and later developments either algorithmically or heuristically. The algorithmic representation used to use one or
several logics of Aristotle, Boolean, sequential, concurrent, k-logic.
Technological support of the one that was chosen manifests itself via hardware
implementations.
From the other side, flexibility implementation by the software of world problems is required. These requirements define the design of system software: languages and operating systems.
Elements of hardware might be specified in terms of required performance and
reliability. At the same time, every hardware technology has its own limits: either
maximum possible frequency is approaching or, actually, implementation problems
for brand new approaches such as quantum—computing trick with Shreddinger’
paradox. Here, we have to be very careful what is achievable and what is for
funding and hardly possible: nature does not accept cheating;
Thus one can see a semantic gap between physical possibility and limitations of
hardware and required flexibility of software to make problems represented by
algorithms doable.
Clear, someone has to pay for this gap, and pay dearly: physical limits of calculations are not even discussed seriously, our personal observations show that modern
home computers easily take 0.5 KW to do the simplest work such as typing!
Current situation looks even more surprising and confusing: 1–5 GHz hardware
nowadays is required to cope with simple problems and demands from concurrency
approach is flourishing by reinventing the wheel: concurrent programming and
concurrent architectures are “blue chips” now in research domain for scientists and
engineers: more than 10 Mln entries one might find using simple Google search.
We lost the sense of reality and historical perspective: Cray has never achieved
any industrially sensitive results, but it was one of the most expensive and one of
the best among supercomputer systems for several decades… If best engineers
failed then, why we naively believe that the new generation of scientists and
engineers can reincarnate an idea and concept?
12.1 Runtime System Structure
179
with the runtime system module over messages.
12.2 Concurrency Support for Resilient Computing
Two monitors as above FT Interrupt Handler and Main Monitor deal with
interruptions and resource handling, including reconfiguration of architecture and
software structure when required.
12.2.1 Problem of Concurrency
Real world is represented in our imagination and later developments either algorithmically or heuristically. The algorithmic representation used to use one or
several logics of Aristotle, Boolean, sequential, concurrent, k-logic.
Technological support of the one that was chosen manifests itself via hardware
implementations.
From the other side, flexibility implementation by the software of world problems is required. These requirements define the design of system software: languages and operating systems.
Elements of hardware might be specified in terms of required performance and
reliability. At the same time, every hardware technology has its own limits: either
maximum possible frequency is approaching or, actually, implementation problems
for brand new approaches such as quantum—computing trick with Shreddinger’
paradox. Here, we have to be very careful what is achievable and what is for
funding and hardly possible: nature does not accept cheating;
Thus one can see a semantic gap between physical possibility and limitations of
hardware and required flexibility of software to make problems represented by
algorithms doable.
Clear, someone has to pay for this gap, and pay dearly: physical limits of calculations are not even discussed seriously, our personal observations show that modern
home computers easily take 0.5 KW to do the simplest work such as typing!
Current situation looks even more surprising and confusing: 1–5 GHz hardware
nowadays is required to cope with simple problems and demands from concurrency
approach is flourishing by reinventing the wheel: concurrent programming and
concurrent architectures are “blue chips” now in research domain for scientists and
engineers: more than 10 Mln entries one might find using simple Google search.
We lost the sense of reality and historical perspective: Cray has never achieved
any industrially sensitive results, but it was one of the most expensive and one of
the best among supercomputer systems for several decades… If best engineers
failed then, why we naively believe that the new generation of scientists and
engineers can reincarnate an idea and concept?
12.1 Runtime System Structure
179
