46 Hits in 7.1 sec

Serializing instructions in system-intensive workloads: Amdahl's Law strikes again

Philip M. Wells, Gurindar S. Sohi
2008 High-Performance Computer Architecture  
As explained by Amdahl's Law, such frequent SIs, which create serial regions within the instruction-level parallel execution of a single thread, can have a significant impact on performance.  ...  Maintaining sequential semantics may force SIs to serialize the pipeline and execute as the only instruction in the window.  ...  Conclusions In this paper, we identify serializing instructions (SIs) as a limiting performance factor for system-intensive workloads.  ... 
doi:10.1109/hpca.2008.4658645 dblp:conf/hpca/WellsS08 fatcat:ybk32t37svbqtnosyfdte7tzai

Increasing Capacity In Telemedicine Using Flow-Based Programming.pdf

Jeremy Thornton
2016 Figshare  
and subject their inherently stochastic scaling to statistical analysis in order to categorize their scaling as either bound by Amdahl's Law or approaching the linear scaling of Gustafson-Barsis' Law  ...  Develop Communicating Sequential Process and Graph Theoretic mathematical semantics for Flow-Based Programming(FBP) in order to define 3 fundamental FBP digraph types, build matching FBP software artifacts  ...  in, again as near as possible, the same operating system environment.  ... 
doi:10.6084/m9.figshare.3580755.v1 fatcat:s5up62llevadfkstav45t7esge

Software challenges in extreme scale systems

Vivek Sarkar, William Harrod, Allan E Snavely
2009 Journal of Physics, Conference Series  
them, and scaling multiple chips to complete systems, for a range of real system applications, from highly scalable deep space exploration to trans-petaflops level supercomputing.  ...  More recent work is investigating how PIM-like ideas may port into quantum cellular array (QCA) and other nanotechnology logic, where in-stead of "Processing-In-Memory" we have opportunities for "Processing-In-Wire  ...  B.5.2 Efficiency and Amdahl's Law If Amdahl's Law is a reasonable view of how some particular program behaves on some parallel computer system, then we can use the last of the above approaches to compute  ... 
doi:10.1088/1742-6596/180/1/012045 fatcat:iukutry2dvbitfdh6ng7kgz564

Beyond programmable shading

Aaron Lefohn, Mike Houston, Chas Boyd, Kayvon Fatahalian, Tom Forsyth, David Luebke, John Owens
2008 ACM SIGGRAPH 2008 classes on - SIGGRAPH '08  
Amdahl's law states that the maximum speedup attainable by parallelism is the reciprocal of the proportion of code that is not parallelizable.  ...  bandwidth-and compute-intensive phases of execution.  ...  computer hardware architecture-the arrival of multi-core CPUs, the generalization of graphics processing units (GPUs), and the imminent increase in bandwidth available between CPU and GPU cores-make a  ... 
doi:10.1145/1401132.1401145 dblp:conf/siggraph/LefohnHBFFLO08 fatcat:jrt5e5373zairmf4fvn2tehaxa

Designing Computational Clusters for Performance and Power [chapter]

Kirk W. Cameron, Rong Ge, Xizhou Feng
2007 Advances in Computers  
Power consumption in computational clusters has reached critical levels.  ...  In this chapter, we motivate the need to reconsider the traditional performance-at-any-cost cluster design approach.  ...  At some later point all of these codes will reach the limits of either the input data set size (Amdahl's Law) or the interconnect technology (saturation) where performance will drop drastically again.  ... 
doi:10.1016/s0065-2458(06)69002-5 fatcat:6piim42jtzcttnogp5gf45sz5e

A Berkeley View of Systems Challenges for AI [article]

Ion Stoica, Dawn Song, Raluca Ada Popa, David Patterson, Michael W. Mahoney, Randy Katz, Anthony D. Joseph, Michael Jordan, Joseph M. Hellerstein, Joseph E. Gonzalez, Ken Goldberg, Ali Ghodsi (+1 others)
2017 arXiv   pre-print
In this paper, we propose several open research directions in systems, architectures, and security that can address these challenges and help unlock AI's potential to improve lives and society.  ...  These changes have been made possible by unprecedented levels of data and computation, by methodological advances in machine learning, by innovations in systems software and architectures, and by the broad  ...  one ine cient processor/chip to about a dozen e cient processors per chip, but there are limits to parallelism due to Amdahl's Law. e one path le to continue the improvements in performanceenergy-cost  ... 
arXiv:1712.05855v1 fatcat:mbg3m2ltqncmxe3z35gnawrgnu

Efficient Master/Worker Parallel Discrete Event Simulation

Alfred Park, Ric Fujimoto
2009 2009 ACM/IEEE/SCS 23rd Workshop on Principles of Advanced and Distributed Simulation  
It has truly been an honor and a blessing to perform research under the supervision of one of the pioneers in the parallel and distributed simulation field.  ...  I feel I have learned a great deal not only from his advisement and expertise but also through the exposure to a variety of projects that I was able to participate in.  ...  ) Amdahl's law is expressed in equation (1.1) .  ... 
doi:10.1109/pads.2009.9 dblp:conf/pads/ParkF09 fatcat:6aiga6uc6jdy5adfzhujiqab3a

The Datacenter as a Computer: An Introduction to the Design of Warehouse-Scale Machines, Second edition

Luiz André Barroso, Jimmy Clidaras, Urs Hölzle
2013 Synthesis Lectures on Computer Architecture  
Thanks in advance for taking the time to contribute.  ...  Acknowledgments While we draw from our direct involvement in Google's infrastructure design and operation over the past several years, most of what we have learned and now report here is the result of  ...  Having said that, even highly parallel systems abide by Amdahl's law, and there may be a point where Amdahl's effects become dominant even in this domain.  ... 
doi:10.2200/s00516ed2v01y201306cac024 fatcat:435o455inbcmrakl6l7jp4gope

Chemistry with ADF

G. te Velde, F. M. Bickelhaupt, E. J. Baerends, C. Fonseca Guerra, S. J. A. van Gisbergen, J. G. Snijders, T. Ziegler
2001 Journal of Computational Chemistry  
MO model in conjunction with the ADF-typical fragment approach to quantitatively understand and predict chemical phenomena.  ...  In the Applications section we discuss the physical model of the electronic structure and the chemical bond, i.e., the Kohn-Sham molecular orbital (MO) theory, and illustrate the power of the Kohn-Sham  ...  parallel mode (Amdahl's law).  ... 
doi:10.1002/jcc.1056 fatcat:scmti53mhjahrfkcbv5wz7rp54

Memory leads the way to better computing

H.-S. Philip Wong, Sayeef Salahuddin
2015 Nature Nanotechnology  
The goal of the study was to assay the state of the art, and not to either propose a potential system or prepare and propose a detailed roadmap for its development.  ...  Further, the report itself was assembled in just a few months at the beginning of 2008 from input by the participants.  ...  an absolute sense, but just as a fraction of total run-time due to Amdahl's Law.  ... 
doi:10.1038/nnano.2015.29 pmid:25740127 fatcat:d6iiuuwcozbxlgn4kxxzdzwd4m

Addressing Application Bottlenecks: Distributed Memory [chapter]

Alexander Supalov, Andrey Semin, Michael Klemm, Christopher Dahnken
2014 Optimizing HPC Applications with Intel® Cluster Tools  
Following this pragmatic approach, in this chapter we will show how to detect and exploit optimization opportunities in the realm of communication patterns.  ...  Indeed, by moving data around in the right manner, you hope to get more computational power in return. The main point, then, is to optimize this investment so that your returns are maximized.  ...  From this observation, as well as the MPI communication percentages shown here and Amdahl's Law explained earlier, we can deduce that there is possibly 2-at most 3-percent overall performance upside in  ... 
doi:10.1007/978-1-4302-6497-2_5 fatcat:aozzgrkxmrasnm54ufvvkwr4g4

State-of-the-art in Heterogeneous Computing

Andre R. Brodtkorb, Christopher Dyken, Trond R. Hagen, Jon M. Hjelmervik, Olaf O. Storaasli
2010 Scientific Programming  
With the increase of fine-grained parallelism in high-performance computing, as well as the introduction of parallelism in workstations, there is an acute need for a good overview and understanding of  ...  We give an overview of the state-of-the-art in heterogeneous computing, focusing on three commonly found architectures: the Cell Broadband Engine Architecture, graphics processing units (GPUs), and field  ...  This has the drawback captured in Amdahl's law, where the serial part of the code quickly becomes the bottleneck.  ... 
doi:10.1155/2010/540159 fatcat:xu4n5ubgfzh3bobd445cmg7qyu

Is Parallel Programming Hard, And, If So, What Can You Do About It? (Release v2021.12.22a) [article]

Paul E. McKenney
2021 arXiv   pre-print
The purpose of this book is to help you program shared-memory parallel systems without risking your sanity.  ...  In some surprisingly common cases, these tasks can be automated.  ...  workload, thus incurring severe retribution from the laws of physics.  ... 
arXiv:1701.00854v4 fatcat:pxiajyczebd5pm76htwnrczhm4

The Vesta parallel file system

Peter F. Corbett, Dror G. Feitelson
1996 ACM Transactions on Computer Systems  
The system is fully implemented and forms the basis for the AIX Parallel I/O File System on the IBM SP2. The implementation does not compromise scalability or parallelism.  ...  In fact, all data accesses are done directly to the I/O node that contains the requested data, without any indirection or access to shared metadata.  ...  If this component is small, parallel I/O will not help, as a result of Amdahl's law.  ... 
doi:10.1145/233557.233558 fatcat:gctsk35cavayhau5fqpzs2kuc4

The Vesta Parallel File System [chapter]

2009 High Performance Mass Storage and Parallel I/O  
The system is fully implemented and forms the basis for the AIX Parallel I/O File System on the IBM SP2. The implementation does not compromise scalability or parallelism.  ...  In fact, all data accesses are done directly to the I/O node that contains the requested data, without any indirection or access to shared metadata.  ...  If this component is small, parallel I/O will not help, as a result of Amdahl's law.  ... 
doi:10.1109/9780470544839.ch20 fatcat:eepeqszzgrhfnp5gvbbmifqzmi
« Previous Showing results 1 — 15 out of 46 results