Xputers
|
wrongroadmap
|
anti-machine
| configware
| morphware |
flowware
|
data.streams
|
KressArray
Xputers
(in German Language) |
Auto-sequencing memory (asM) |
Generic Address Generator (GAG) |
Reinvent Computing |
The Xputer supporting the Mead-&-Conway Microchip Design Revolution:
DRC speed-up by a factor of 15.000 (see the PISA project).
Note the
amateurish introduction of the Systolic Array. The Xputer results from the
generalization of the sytolic array.
When we developped the anti-machine
paradigm as the counterpart of the von Neumann computer, we have been looking
for a prefix to replace "com" within the word "computer". Not
yet having a good
idea (do you have one?) we replaced "com" by "x" yielding "xputer".
The term "Xputer" does not mean "transputer",
which was a von Neumann machine, whereas the Xputer is
not - it is an anti-machine.
Why a different computing machine paradigm? It is the massive inefficiency of
the von Neumann paradigm. UC Berkeley professor C. V. "RAM" Ramamoorthy coined
the term "von Neumann
Syndrome" (also see
here) when
opening the discussion after listening to a keynote speech by Prof. Reiner Hartenstein.
Well known about this is "
Nathan's
Law" saying, that software is a gas which completely fills all available
storage space. Also see
here and
here.
Why we urgently must go heterogeneous:
Keywords, abstracted by Reiner from:
“Cooling Off Accelerated Computing —
Intel/Altera Attacks Memory Bottleneck — Abstract”
by Kevin Morris, May 25, 2016 - see:
http://www.eejournal.com/archives/articles/20160525-coolingoff/<>
A battle is on to claim supremacy in the next generation of computing.
Alliances are forming, battle plans are being forged, and armies are amassing.
Energy consumption: the enemy is power. The single limiting factor is energy consumption – for fastest processors increasing exponentially.
Efficiency is Everything. Von Neumann (vN) maximizes computational complexity. Silicon has become almost free.
However, Because of a lack of flexibility ASICs do not really solve these kinds of efficiency problems.
We urgently need a game changer providing incredible bandwidth and power efficiency.
We need multiple orders of magnitude in computational energy efficiency.
We must go heterogeneous by using von Neumann/FPGA machines.
Von-Neumann-only systems are inefficiently shuffling data back and force to and from DRAM.
We need low-latency connections between FPGA fabric and memory.
The problems are far from being solved. (Remark by Reiner: not quite that far by the twin paradigm approach using (vN) computers together with (non-vN-)
Xputers.)
The next years will be interesting to watch.
Literature
Reiner Hartenstein: Xputers and their relations to H/S Codesign; internal report,
Univ. Kaiserslautern, Sep 1994 -
http://www.fpl.uni-kl.de/papers/paper_Xp.pdf (achievements survey)
R. W. Hartenstein, M. Riedmuller, K. Schmitt, M.
Weber: A Novel ASIC Design Approach Based on a New Machine Paradigm; Special
Issue of IEEE Journal of Solid State Circuits, July 1991 -
http://www.fpl.uni-kl.de/xputer-pages/IEEE_JSSC91.pdf
R. W. Hartenstein, K. Schmidt, H. Reinig, M.
Weber: A Novel Compilation Technique for a Machine Paradigm Based on
Field-Programmable Logic; Proc. FPL 1991, Oxford, UK -
http://www.fpl.uni-kl.de/xputer-pages/Xputer-paper-039.pdf
R.W. Hartenstein, A.G. Hirschbiel, M.Weber: A Novel
Paradigm of Parallel Computation and its Use to Implement Simple High
Performance Hardware; InfoJapan'90- International Conference memorating the
30th Anniversary of the Computer Society of Japan, Tokyo, Japan, 1990 -
http://www.fpl.uni-kl.de/xputer-pages/Xputer_InfoJapan_1990_31ps.pdf
R. Hartenstein, A.
Hirschbiel, K. Schmidt, M. Weber: A Novel Paradigm of Parallel Computation and
its Use to Implement Simple High-Performance-HW; Future Generation Computer
Systems 7 91/92, p. 181-198, North Holland (invited reprint of Paper 31) -
http://www.fpl.uni-kl.de/xputer-pages/Xputer-paper-042.pdf
R.W. Hartenstein, A.G. Hirschbiel, M.
Riedmueller, K. Schmidt, M.Weber: A High Performance Machine Paradigm Based on
Auto-Sequencing Data Memory; HICSS-24, Hawaii Int. Conference on System
Sciences, Koloa Hawaii, 1991 - Second Best Paper Award (Honorable Mention)
-
http://www.fpl.uni-kl.de/xputer-pages/Xputer-paper-036.pdf |
http://www.fpl.uni-kl.de/anti-machine/asM-Koloa-1991.pdf
R.W. Hartenstein, A.G. Hirschbiel, M.
Riedmuller, K. Schmidt, M.Weber: Automatic Synthesis of Cheap Hardware
Accelerators for Signal Processing and Image Preprocessing; 12. DAGM-Symposium
Mustererkennung, Oberkochen-Aalen, 1990
Best
Paper and Best Presentation Award (DM 1000.--) - Speaker: Michael
Weber -
http://www.fpl.uni-kl.de/xputer-pages/Xputer-paper-029.pdf
R.W. Hartenstein,
A.G. Hirschbiel, M.Weber: The Machine Paradigm of Xputers and its Application
to Digital Signal Processing Acceleration; 1990 Int. Conference on Parallel
Processing, St. Charles, Illinois , 1990 -
http://www.fpl.uni-kl.de/xputer-pages/Xputer-paper-027.ps |
http://www.fpl.uni-kl.de/anti-machine/Reiner-StCharles1990.pdf

Reiner W. Hartenstein: A decade of reconfigurable computing: a visionary
retrospective;
Conference: Design, Automation, and Test in Europe - DATE, 2001
downloadable
Christophe Bobda: Introduction to Reconfigurable Computing: Architectures;
Springer Verlag, 2007
More Xputer-related Literature
|