Downloads & Free Reading Options - Results

Declassified in Part - Sanitized Copy Approved for Release 2013/03/18: CIA-RDP88GO0186R000901070024-3 HAT TO: 1 DCI ACTION INFO DATE INITIAL X 2 DDCI 3 EXDIR X 4 D/ICS X 5 DDI X DDA X 7 DDO 8 DDS&T X 9 Chm/NIC 10 GC 11 IG 12 Compt 13 D/Pers 14 D/OLL 15 D/PAO 16 SA/1A 17 AO/DCI 18 C/IPD/OIS 19 20 21 2Z To 4: Year r.co endation please. Executive ecre ark 9 Ma ' Declassified in Part - Sanitized Copy Approved for Release 2013/03/18: CIA-RDP88GO0186R000901070024-3 Declassified in Part - Sanitized Copy Approved for Release 2013/03/18: CIA-RDP88GO0186R000901070024-3 NATIONAL SECURITY AGENCY FORT GEORGE G. MEADE. MARYLAND 20755-6 0 0 0 Serial: N0923 3 May 1985 MEMORANDUM FOR THE DIRECTOR, CENTRAL INTELLIGENCE SUBJECT: Supercomputing Research Center (SRC) (U) - INFORMATION MEMORANDUM 1414/2 1. (U) Please reference your memorandum, subject as above, dated 3 April 1985. As you suggested, we are structuring our FY 1986 programs in such a way that some additional funding for the SRC could be provided through reprogramming procedures. 2. (C) We have also re-examined the facts associated with the $150 million add-on provided by the Defense Resources Board. These funds were not solely for computer security as indicated in your memorandum, but were allocated as follows: Our understanding on this allocation, like yours, was that some of these funds could be diverted to the SRC. Indeed, during the FY 1986 budget build we used a portion of the COMSEC allo- cation for this purpose. The DOD budget decision document severely criticized this attempted diversion and disallowed it. 3. (C) The SRC was not supported by the Department of Defense because it was unclear to some how the DOD would benefit from the effort. The ASD C31, however, was supportive during last year's budget review, is now supportive, and has initiated action to educate the DOD on the benefits of participating in the development of the SRC. The ASD C31 has further indicated that he expects the DOD to become a full Classified By NSAiC::S61 123-2 Declassify On: Ori,,in3 n-, /,- - . Declassified in Part - Sanitized Copy Approved for Release 2013/03/18: CIA-RDP88GO0186R000901070024-3 Declassified in Part - Sanitized Copy Approved for Release 2013/03/18: CIA-RDP88GO0186R000901070024-3 CONFIDENTIAL Serial: N0923 financial partner in FY 1987. It should also be noted that a mid-term objective in the recently received FY 1987-1991 Defense Guidance concerns providing appropriate DoD support for the SRC. We view these efforts favorably and we continue to solicit your support in obtaining at least $10 million of the FY 1986 shortfall from other NFIP programs in the year of execution in order to keep the SRC effort on track. ROBERT E. RICH Deputy Director Copy Furnished: Dir, ICS ASD (C3I ) CCUFIDENTIAL Declassified in Part - Sanitized Copy Approved for Release 2013/03/18: CIA-RDP88GO0186R000901070024-3

The primary objective of this research is to support the design of liquid rocket systems for the Advanced Space Transportation System. Since the space launch systems in the near future are likely to rely on liquid rocket engines, increasing the efficiency and reliability of the engine components is an important task. One of the major problems in the liquid rocket engine is to understand fluid dynamics of fuel and oxidizer flows from the fuel tank to plume. Understanding the flow through the entire turbo-pump geometry through numerical simulation will be of significant value toward design. One of the milestones of this effort is to develop, apply and demonstrate the capability and accuracy of 3D CFD methods as efficient design analysis tools on high performance computer platforms. The development of the Message Passage Interface (MPI) and Multi Level Parallel (MLP) versions of the INS3D code is currently underway. The serial version of INS3D code is a multidimensional incompressible Navier-Stokes solver based on overset grid technology, INS3D-MPI is based on the explicit massage-passing interface across processors and is primarily suited for distributed memory systems. INS3D-MLP is based on multi-level parallel method and is suitable for distributed-shared memory systems. For the entire turbo-pump simulations, moving boundary capability and efficient time-accurate integration methods are built in the flow solver, To handle the geometric complexity and moving boundary problems, an overset grid scheme is incorporated with the solver so that new connectivity data will be obtained at each time step. The Chimera overlapped grid scheme allows subdomains move relative to each other, and provides a great flexibility when the boundary movement creates large displacements. Two numerical procedures, one based on artificial compressibility method and the other pressure projection method, are outlined for obtaining time-accurate solutions of the incompressible Navier-Stokes equations. The performance of the two methods is compared by obtaining unsteady solutions for the evolution of twin vortices behind a flat plate. Calculated results are compared with experimental and other numerical results. For an unsteady flow, which requires small physical time step, the pressure projection method was found to be computationally efficient since it does not require any subiteration procedure. It was observed that the artificial compressibility method requires a fast convergence scheme at each physical time step in order to satisfy the incompressibility condition. This was obtained by using a GMRES-ILU(0) solver in present computations. When a line-relaxation scheme was used, the time accuracy was degraded and time-accurate computations became very expensive.

The primary objective of this research is to support the design of liquid rocket systems for the Advanced Space Transportation System. Since the space launch systems in the near future are likely to rely on liquid rocket engines, increasing the efficiency and reliability of the engine components is an important task. One of the major problems in the liquid rocket engine is to understand fluid dynamics of fuel and oxidizer flows from the fuel tank to plume. Understanding the flow through the entire turbo-pump geometry through numerical simulation will be of significant value toward design. One of the milestones of this effort is to develop, apply and demonstrate the capability and accuracy of 3D CFD methods as efficient design analysis tools on high performance computer platforms. The development of the Message Passage Interface (MPI) and Multi Level Parallel (MLP) versions of the INS3D code is currently underway. The serial version of INS3D code is a multidimensional incompressible Navier-Stokes solver based on overset grid technology, INS3D-MPI is based on the explicit massage-passing interface across processors and is primarily suited for distributed memory systems. INS3D-MLP is based on multi-level parallel method and is suitable for distributed-shared memory systems. For the entire turbo-pump simulations, moving boundary capability and efficient time-accurate integration methods are built in the flow solver, To handle the geometric complexity and moving boundary problems, an overset grid scheme is incorporated with the solver so that new connectivity data will be obtained at each time step. The Chimera overlapped grid scheme allows subdomains move relative to each other, and provides a great flexibility when the boundary movement creates large displacements. Two numerical procedures, one based on artificial compressibility method and the other pressure projection method, are outlined for obtaining time-accurate solutions of the incompressible Navier-Stokes equations. The performance of the two methods is compared by obtaining unsteady solutions for the evolution of twin vortices behind a flat plate. Calculated results are compared with experimental and other numerical results. For an unsteady flow, which requires small physical time step, the pressure projection method was found to be computationally efficient since it does not require any subiteration procedure. It was observed that the artificial compressibility method requires a fast convergence scheme at each physical time step in order to satisfy the incompressibility condition. This was obtained by using a GMRES-ILU(0) solver in present computations. When a line-relaxation scheme was used, the time accuracy was degraded and time-accurate computations became very expensive.

The primary objective of this research is to support the design of liquid rocket systems for the Advanced Space Transportation System. Since the space launch systems in the near future are likely to rely on liquid rocket engines, increasing the efficiency and reliability of the engine components is an important task. One of the major problems in the liquid rocket engine is to understand fluid dynamics of fuel and oxidizer flows from the fuel tank to plume. Understanding the flow through the entire turbo-pump geometry through numerical simulation will be of significant value toward design. One of the milestones of this effort is to develop, apply and demonstrate the capability and accuracy of 3D CFD methods as efficient design analysis tools on high performance computer platforms. The development of the Message Passage Interface (MPI) and Multi Level Parallel (MLP) versions of the INS3D code is currently underway. The serial version of INS3D code is a multidimensional incompressible Navier-Stokes solver based on overset grid technology, INS3D-MPI is based on the explicit massage-passing interface across processors and is primarily suited for distributed memory systems. INS3D-MLP is based on multi-level parallel method and is suitable for distributed-shared memory systems. For the entire turbo-pump simulations, moving boundary capability and efficient time-accurate integration methods are built in the flow solver, To handle the geometric complexity and moving boundary problems, an overset grid scheme is incorporated with the solver so that new connectivity data will be obtained at each time step. The Chimera overlapped grid scheme allows subdomains move relative to each other, and provides a great flexibility when the boundary movement creates large displacements. Two numerical procedures, one based on artificial compressibility method and the other pressure projection method, are outlined for obtaining time-accurate solutions of the incompressible Navier-Stokes equations. The performance of the two methods is compared by obtaining unsteady solutions for the evolution of twin vortices behind a flat plate. Calculated results are compared with experimental and other numerical results. For an unsteady flow, which requires small physical time step, the pressure projection method was found to be computationally efficient since it does not require any subiteration procedure. It was observed that the artificial compressibility method requires a fast convergence scheme at each physical time step in order to satisfy the incompressibility condition. This was obtained by using a GMRES-ILU(0) solver in present computations. When a line-relaxation scheme was used, the time accuracy was degraded and time-accurate computations became very expensive.

NASA's Kepler Mission was launched in March 2009 as NASA's first mission capable of finding Earth-size planets orbiting in the habitable zone of Sun-like stars, that range of distances for which liquid water would pool on the surface of a rocky planet. Kepler has discovered over 1000 planets and over 4600 candidates, many of them as small as the Earth. Today, Kepler's amazing success seems to be a fait accompli to those unfamiliar with her history. But twenty years ago, there were no planets known outside our solar system, and few people believed it was possible to detect tiny Earth-size planets orbiting other stars. Motivating NASA to select Kepler for launch required a confluence of the right detector technology, advances in signal processing and algorithms, and the power of supercomputing.

The super computer center at the National Center for Atmospheric Research (NCAR) migrates large numbers of files to and from its mass storage system (MSS) because there is insufficient space to store them on the Cray supercomputer's local disks. This paper presents an analysis of file migration data collected over two years. The analysis shows that requests to the MSS are periodic, with one day and one week periods. Read requests to the MSS account for the majority of the periodicity; as write requests are relatively constant over the course of a week. Additionally, reads show a far greater fluctuation than writes over a day and week since reads are driven by human users while writes are machine-driven.

We present our experiences using cloud computing to support data-intensive analytics on satellite imagery for commercial applications. Drawing from our background in high-performance computing, we draw parallels between the early days of clustered computing systems and the current state of cloud computing and its potential to disrupt the HPC market. Using our own virtual file system layer on top of cloud remote object storage, we demonstrate aggregate read bandwidth of 230 gigabytes per second using 512 Google Compute Engine (GCE) nodes accessing a USA multi-region standard storage bucket. This figure is comparable to the best HPC storage systems in existence. We also present several of our application results, including the identification of field boundaries in Ukraine, and the generation of a global cloud-free base layer from Landsat imagery.

We present our experiences using cloud computing to support data-intensive analytics on satellite imagery for commercial applications. Drawing from our background in high-performance computing, we draw parallels between the early days of clustered computing systems and the current state of cloud computing and its potential to disrupt the HPC market. Using our own virtual file system layer on top of cloud remote object storage, we demonstrate aggregate read bandwidth of 230 gigabytes per second using 512 Google Compute Engine (GCE) nodes accessing a USA multi-region standard storage bucket. This figure is comparable to the best HPC storage systems in existence. We also present several of our application results, including the identification of field boundaries in Ukraine, and the generation of a global cloud-free base layer from Landsat imagery.

Research on the Goodyear Massively Parallel Processor (MPP) suggests that high-level parallel languages are practical and can be designed with powerful new semantics that allow algorithms to be efficiently mapped to the real machines. For the MPP these semantics include parallel/associative array selection for both dense and sparse matrices, variable precision arithmetic to trade accuracy for speed, micro-pipelined train broadcast, and conditional branching at the processing element (PE) control unit level. The preliminary design of a FORTRAN-like parallel language for the MPP has been completed and is being used to write programs to perform sparse matrix array selection, min/max search, matrix multiplication, Gaussian elimination on single bit arrays and other generic algorithms. A description is given of the MPP design. Features of the system and its operation are illustrated in the form of charts and diagrams.

Samsung's HBM Flashbolt (HBM2E) is optimized to enable quick and accurate tasks in AI and Big Data. Accelerate the journey of supercomputing with Samsung's HBM Flashbolt. Source: https://www.youtube.com/watch?v=XqrZmmMGHV8 Uploader: Samsung

Attached is the report of the Defense Science Board Task Force on DoD Super Computing Needs. The Task Force validated the need for high performance computers that provide extremely rapid access to extremely large global memories. This capability would support not only cryptanalysis but several other important DoD needs as well (e.g. calculation of weapons effects, weapon design and analysis, acoustic analysis, computational fluid dynamics, radar cross sectional modeling, and synthetic materials design) The Task Force recommends a three part strategy to meet the DoD's Super Computing Needs. First, the DoD should continue short-term support of the CRAY SV2 development. This is a risky development, but the modest expenditures are worth the potential payoff in performance improvement. Secondly, the DoD should develop a high bandwidth memory system using Commercial-off-the-Shelf microprocessors for the medium term. This strategy mitigates any potential failure of the SV2 development. Finally, DoD should invest in long-term research to address unique Defense computing needs. Such research is essential to refill the Research and Development pipeline with new technologies that will enable tomorrow's high performance computers.

The DURIP '99 initiative award was used primarily far a major upgrade of a dual rack ONYX2 Silicon Graphics system. The upgrade consisted in purchasing 8 of the newest generation of 400 MHz CPUs, converting one of ONYX2 racks into a fully loaded 16-processor Origin 2000/2400 system and moving both high performance graphics heads into the top half of the remaining ONYX2 rack. The existing 14 processor system consisted of a mix of 6 195 MHz and 8250 MHz CPUs running in a system with 6 Gigabytes of memory. In addition, Silicon Graphics replaced our existing mix of 14 195 and 250 MHz CPUs by 14300 MHz CPUs at no added cost. The new system provides a state-of-the-art graphical visualization and computing capability. It also offers cost-effective upgrade route. The reconfigured in-house supercomputing engine provides a significant boost to our past and ongoing AFOSR-funded research projects.

This is the tenth annual edition of Maui High Performance Computing Center s (MHPCC) Application Briefs which highlights some of the successes our customers have achieved this year. MHPCC, established in September 1993, is an Air Force Research Laboratory (AFRL) Center managed by the University of Hawaii. A leader in scalable parallel computing technologies, MHPCC is primarily chartered to support the Department of Defense (DoD) and other government organizations. MHPCC offers an innovative environment for High Performance Computing (HPC) applications. The user application briefs in this publication represent selected research efforts that have taken place at MHPCC during 2004. Each application brief was written by an individual researcher or research team, and reflects their first-hand experiences using MHPCC resources. These articles reflect the diverse nature of our users and projects.

A new class of very powerful workstations has recently become available which integrate near supercomputer computational performance with very powerful and high quality graphics capability. These graphics super-workstations are expected to play an increasingly important role in providing an enhanced environment for supercomputer users. Their potential uses include: off-loading the supercomputer (by serving as stand-alone processors, by post-processing of the output of supercomputer calculations, and by distributed or shared processing), scientific visualization (understanding of results, communication of results), and by real time interaction with the supercomputer (to steer an iterative computation, to abort a bad run, or to explore and develop new algorithms).

In a previous humorous note, I outlined twelve ways in which performance figures for scientific supercomputers can be distorted. In this paper, the problem of potentially misleading performance reporting is discussed in detail. Included are some examples that have appeared in recent published scientific papers. This paper also includes some proposed guidelines for reporting performance, the adoption of which would raise the level of professionalism and reduce the level of confusion in the field of supercomputing.

In a previous humorous note, I outlined twelve ways in which performance figures for scientific supercomputers can be distorted. In this paper, the problem of potentially misleading performance reporting is discussed in detail. Included are some examples that have appeared in recent published scientific papers. This paper also includes some proposed guidelines for reporting performance, the adoption of which would raise the level of professionalism and reduce the level of confusion in the field of supercomputing.

In a previous humorous note, I outlined twelve ways in which performance figures for scientific supercomputers can be distorted. In this paper, the problem of potentially misleading performance reporting is discussed in detail. Included are some examples that have appeared in recent published scientific papers. This paper also includes some proposed guidelines for reporting performance, the adoption of which would raise the level of professionalism and reduce the level of confusion in the field of supercomputing.

artificial intelligence; big data; cloud computing; communication; computer architecture; computer science; computer systems; data management; databases; hardware; High-Performance Computing (HPC); information management; map-reduce; processors; programming languages; semantics; wireless telecommunication systems

Includes bibliographical references and index

From the bitsavers.org collection, a scanned-in computer-related document. intel :: dataBooks :: Intel Multimedia and Supercomputing Processors Jan92

The latest generation of radio astronomy interferometers will conduct all sky surveys with data products consisting of petabytes of spectral line data. Traditional approaches to identifying and parameterising the astrophysical sources within this data will not scale to datasets of this magnitude, since the performance of workstations will not keep up with the real-time generation of data. For this reason, it is necessary to employ high performance computing systems consisting of a large number of processors connected by a high-bandwidth network. In order to make use of such supercomputers substantial modifications must be made to serial source finding code. To ease the transition, this work presents the Scalable Source Finder Framework, a framework providing storage access, networking communication and data composition functionality, which can support a wide range of source finding algorithms provided they can be applied to subsets of the entire image. Additionally, the Parallel Gaussian Source Finder was implemented using SSoFF, utilising Gaussian filters, thresholding, and local statistics. PGSF was able to search on a 256GB simulated dataset in under 24 minutes, significantly less than the 8 to 12 hour observation that would generate such a dataset.

The experiment was designed to assist the Numerical Aerodynamic Simulation (NAS) Project Office in the testing and evaluation of long haul communications for remote users. The objectives of this work were to: (1) use foreign workstations to remotely access the NAS system; (2) provide NAS with a link to a large university-based computing facility which can serve as a model for a regional node of the Long-Haul Communications Subsystem (LHCS); and (3) provide a tail circuit to the University of Colorado a Boulder thereby simulating the complete communications path from NAS through a regional node to an end-user.

Contents of the papers include: Visualization techniques- The next generation, Small-machine visualizing very big data sets, Public display of Lunar Prospector data, Steering molecules via Feel, High-dimensional data browser, Multi-source visualization, Interactive surface flow visualization, and the Virtual Windtunnel.

1 online resource (270 pages)

1 online resource (270 pages)

From the bitsavers.org collectio n, a scanned-in computer-related document. intel :: dataBooks :: 1992 Intel Multimedia and Supercomputing Processors

Declassified and Approved For Release 2013/03/18: C IA-RDP88GO0186R000901070026-1 DDA SUBJECT ~ FILE COP- ?~ Executive R egtstiy NATIONAL SECURITY AGENCY 885- FORT GEORGE G MEADE. MARYLAND 20755 1414 -6000 I X11 J Serial: N0675 26 March 1985 DD/A Registry SUBJECT: The Supercomput-}-ng Research Center (SRC) - formerly called the National Security Supercomputer Research Laboratory (NSSRL) (U) - ACTION MEMORANDUM MEMORANDUM FOR THE DIRECTOR, CENTRAL INTELLIGENCE 1. (C) As you know, we had expected considerable SRC support from the Department of Defense, based on their FY 1986 Program Build Guidance issued in March 1984. During the DoD budget review, however, the $16M that was requested for this project was not supported by the Defense Resources Board. While I believe the Defense action on this issue was short-sighted, we are nevertheless faced with a significant shortfall in the creation of this important capability. The Center is off to a very good start and we should collectively make certain that none of the momentum is lost due to insufficient funding. 2. (C) The benefits to be derived from the SRC justify our full support and should not be delayed or scaled back from the original estimate. Parallel processing is a necessity if we are to continue to protect our edge in cryptanalysis and communi- cations security. But, I believe that our research into supercomputing will be of meaningful benefit to other members of the DoD and intelligence community as well. Our goals are very ambitious (several orders of magnitude more performance than commercially available), but enough work has been done already to conclude that these goals are realistic. Although I have not explored in detail the areas outside of cryptology where this level of performance is applicable, the following applications appear to be ideal users of such high speed systems as we plan to design for the 1990s: pattern recognition, imagery processing and enhancement (NPIC); massive data comparison and inferencing as might be used in all-source processing (CIA and DIA); generalized signals processing for the foreign instrumentation signals intelligence program (CIA, FTD, Air Force) or for the ACOUSTINT problem (NAVY); natural language processing (universal); digitization, storage and processing of world mapping (DMA); and document analysis and processing of open source materials (CIA, DIA). Declassified and Approved For Release 2013/03/18: CIA-RDP88GO0186R000901070026-1 Declassified and Approved For Release 2013/03/18: CIA-RDP88GO0186R000901070026-1 Serial: N0675 3. (S) As you know the CCP has already made a significant commitment to this project ($17.7M in FY 1986), but identifying from the CCP an additional $16M, required in FY 1986 to proceed as planned, would have a significant negative impact on our program. The SRC can be of ben9Jif;tO` +'Tp programs, and r believe its support should be shared. USA is prepared to cover $6M of the FY 1986 shortage if the other programs can cover the remaining $10M. This cooperative action is similar to last year's cooperative effort in developing the Community measurements program under DIA's direction. .I request your favorable consideration in providing $10M from other NFIP programs to fund the SRC. 4. (U) We fully anticipate that the DoD will become a financial partner beginning in FY 1987 and we will work closely with the ASD(C31) to assist him in assuring that DoD support is forthcoming. If we fail in this, I would recommend the entire SRC amount be programmed within the NFIP starting in FY 1987. LI Lieut Copy Furnished: ASD (C3I ) v N D. FAURER ant General USAF rector, NSA Declassified and Approved For Release 2013/03/18: CIA-RDP88GO0186R000901070026-1

The Journal of Supercomputing 1992: Volume 6 , Issue Index. Digitized from IA1518713-03 . Previous issue: sim_journal-of-supercomputing_1992-02_5_4 . Next issue: sim_journal-of-supercomputing_1992-03_6_1 .

The Journal of Supercomputing 1992: Volume 6 , Issue Index. Digitized from IA1518713-03 . Previous issue: sim_journal-of-supercomputing_1992-02_5_4 . Next issue: sim_journal-of-supercomputing_1992-03_6_1 .

The Journal of Supercomputing 1992: Volume 6 , Issue Index. Digitized from IA1518713-03 . Previous issue: sim_journal-of-supercomputing_1992-02_5_4 . Next issue: sim_journal-of-supercomputing_1992-03_6_1 .

The hyperwall is a new concept in visual supercomputing, conceived and developed by the NAS Exploratory Computing Group. The hyperwall will allow simultaneous and coordinated visualization and interaction of an array of processes, such as a the computations of a parameter study or the parallel evolutions of a genetic algorithm population. Making over 65 million pixels available to the user, the hyperwall will enable and elicit qualitatively new ways of leveraging computers to accomplish science. It is currently still unclear whether we will be able to transport the hyperwall to SC02. The crucial display frame still has not been completed by the metal fabrication shop, although they promised an August delivery. Also, we are still working the fragile node issue, which may require transplantation of the compute nodes from the present 2U cases into 3U cases. This modification will increase the present 3-rack configuration to 5 racks.

Approved For Release 2009/09/14 :CIA-RDP87M00539R000901170010-0 L~~~CUTIV]E SECRETARIAT ROUTING SLIP 33 ~ Q%E~O '.~a J,ao1~c: ~ i Cl"IPD/C ~~ y e Approved For Release 2009/09/14 :CIA-RDP87M00539R000901170010-0 Approved For Release 2009/09/14 :CIA-RDP87M00539R000901170010-0 ~T~l~3[~fl~L..S;ECURITY AGENCY I~ITRAt_..~E~CURtTY .SERVECE L~'~ICOLN D . FAURER LBUtenant General.. IISAF` Director , NSA~!Ch ief , CSS ~_~'_ ~ Approved For Release 2009/09/14 :CIA-RDP87M00539R000901170010-0 ~.ur_t~y ,bupercv~pu~~?~y s~coc...v.. .r.,..,,_~~___s ,. .. _. - - - .._ _ . ._.y --- _ _.~. ~ ~ ~te..ear ~~ae to your question Zs this bung done th~~-~trategfc DefeRSe Initiative in mind, ;t~e;work_done by the Laboratory in learning ~``T ~ - T licable to some of the very em~l wi II be app ~ _ . c?;D'~'~en~e Initiative 1SDZI computational -~: s , , ; t up to address basic protil~e~as that ~~P arallel supercomputers on' cr~i-tical ~~ ' ~~ , .y. p ~;~r,........ p~Q~blems ? We ,need to begin with well-d,Qf ined ~He~potential for .concurrent processing. The :-str~s~ture. The Defense Advance ,~~~ ~ . ,, A~!~ncy~ -tDARPAy gazalle3 processing research ms l l ~ ~ . :~ pro ~ e lad :ma the starting point for theg problems are formulated, they ~~~n onal t# : : t~aa1 Defense problems the Laboratory ie ,R;. ,..~.... . .the.results o? the Laboratory will it~ahat . ~pp.aieati~on in Defense and this, secure the technology transfer of these critical ~~~~,~~~ ILLEGIB Approved For Release 2009/09/14 :CIA-RDP87M00539R000901170010-0 co .. :; i i . Approved For Release 2009/09/14 :CIA-RDP87M00539R000901170010-0 l ~G~33 /~

A novel metric is introduced to compare the supercomputing resources available to academic researchers on a national basis. Data from the supercomputing Top 500 and the top 500 universities in the Academic Ranking of World Universities (ARWU) are combined to form the proposed "500/500" score for a given country. Australia scores poorly in the 500/500 metric when compared with other countries with a similar ARWU ranking, an indication that HPC-based researchers in Australia are at a relative disadvantage with respect to their overseas competitors. For HPC problems where single precision is sufficient, commodity GPUs provide a cost-effective means of quenching the computational thirst of otherwise parched Lattice practitioners traversing the Australian supercomputing desert. We explore some of the more difficult terrain in single precision territory, finding that BiCGStab is unreliable in single precision at large lattice sizes. We test the CGNE and CGNR forms of the conjugate gradient method on the normal equations. Both CGNE and a modified form of CGNR (with restarts) provide reliable convergence for quark propagator calculations in single precision.

A novel metric is introduced to compare the supercomputing resources available to academic researchers on a national basis. Data from the supercomputing Top 500 and the top 500 universities in the Academic Ranking of World Universities (ARWU) are combined to form the proposed "500/500" score for a given country. Australia scores poorly in the 500/500 metric when compared with other countries with a similar ARWU ranking, an indication that HPC-based researchers in Australia are at a relative disadvantage with respect to their overseas competitors. For HPC problems where single precision is sufficient, commodity GPUs provide a cost-effective means of quenching the computational thirst of otherwise parched Lattice practitioners traversing the Australian supercomputing desert. We explore some of the more difficult terrain in single precision territory, finding that BiCGStab is unreliable in single precision at large lattice sizes. We test the CGNE and CGNR forms of the conjugate gradient method on the normal equations. Both CGNE and a modified form of CGNR (with restarts) provide reliable convergence for quark propagator calculations in single precision.

From the bitsavers.org collection, a scanned-in computer-related document. components :: intel :: dataBooks :: 1992 Intel Multimedia and Supercomputing Processors Jan92

Since its inception, VLSI theory has expanded in many fruitful and interesting directions. One major branch is layout theory which studies the efficiency with which graphs can be embedded in the plane according to VLSI design rules. In this survey paper, I review some of the major accomplishments of VLSI layout theory and discuss how layout theory engendered the notion of area and volume-universal networks, such as fat-trees. These scalable networks offer a flexible alternative to the more common hypercube-based networks for inter-connecting the processors of large parallel supercomputers. Keywords: Integrated circuits; Interconnection networks; Parallel computing; Super-computing; Universality; Thompson's model; Tree of meshes.

Since its inception, VLSI theory has expanded in many fruitful and interesting directions. One major branch is layout theory which studies the efficiency with which graphs can be embedded in the plane according to VLSI design rules. In this survey paper, I review some of the major accomplishments of VLSI layout theory and discuss how layout theory engendered the notion of area and volume-universal networks, such as fat-trees. These scalable networks offer a flexible alternative to the more common hypercube-based networks for inter-connecting the processors of large parallel supercomputers. Keywords: Integrated circuits; Interconnection networks; Parallel computing; Super-computing; Universality; Thompson's model; Tree of meshes.

From the bitsavers.org collection, a scanned-in computer-related document. floatingPointSystems :: FPS-164 :: Charlesworth - Introducing Replicated VLSI to Supercomputing the FPS-164 MAX Scientific Computer 1986

We describe the political and technical complications encountered during the astronomical CosmoGrid project. CosmoGrid is a numerical study on the formation of large scale structure in the universe. The simulations are challenging due to the enormous dynamic range in spatial and temporal coordinates, as well as the enormous computer resources required. In CosmoGrid we dealt with the computational requirements by connecting up to four supercomputers via an optical network and make them operate as a single machine. This was challenging, if only for the fact that the supercomputers of our choice are separated by half the planet, as three of them are located scattered across Europe and fourth one is in Tokyo. The co-scheduling of multiple computers and the 'gridification' of the code enabled us to achieve an efficiency of up to $93\%$ for this distributed intercontinental supercomputer. In this work, we find that high-performance computing on a grid can be done much more effectively if the sites involved are willing to be flexible about their user policies, and that having facilities to provide such flexibility could be key to strengthening the position of the HPC community in an increasingly Cloud-dominated computing landscape. Given that smaller computer clusters owned by research groups or university departments usually have flexible user policies, we argue that it could be easier to instead realize distributed supercomputing by combining tens, hundreds or even thousands of these resources.

We describe the construction of a high performance parallel computer composed of PC components, present some physical results for light hadron and hybrid meson masses from lattice QCD. We also show that the smearing technique is very useful for improving the spectrum calculations.

Astronomers have come to rely on the increasing performance of computers to reduce, analyze, simulate and visualize their data. In this environment, faster computation can mean more science outcomes or the opening up of new parameter spaces for investigation. If we are to avoid major issues when implementing codes on advanced architectures, it is important that we have a solid understanding of our algorithms. A recent addition to the high-performance computing scene that highlights this point is the graphics processing unit (GPU). The hardware originally designed for speeding-up graphics rendering in video games is now achieving speed-ups of $O(100\times)$ in general-purpose computation -- performance that cannot be ignored. We are using a generalized approach, based on the analysis of astronomy algorithms, to identify the optimal problem-types and techniques for taking advantage of both current GPU hardware and future developments in computing architectures.

Astronomers have come to rely on the increasing performance of computers to reduce, analyze, simulate and visualize their data. In this environment, faster computation can mean more science outcomes or the opening up of new parameter spaces for investigation. If we are to avoid major issues when implementing codes on advanced architectures, it is important that we have a solid understanding of our algorithms. A recent addition to the high-performance computing scene that highlights this point is the graphics processing unit (GPU). The hardware originally designed for speeding-up graphics rendering in video games is now achieving speed-ups of $O(100\times)$ in general-purpose computation -- performance that cannot be ignored. We are using a generalized approach, based on the analysis of astronomy algorithms, to identify the optimal problem-types and techniques for taking advantage of both current GPU hardware and future developments in computing architectures.

Internet supercomputing is an approach to solving partitionable, computation-intensive problems by harnessing the power of a vast number of interconnected computers. This paper presents a new algorithm for the problem of using network supercomputing to perform a large collection of independent tasks, while dealing with undependable processors. The adversary may cause the processors to return bogus results for tasks with certain probabilities, and may cause a subset $F$ of the initial set of processors $P$ to crash. The adversary is constrained in two ways. First, for the set of non-crashed processors $P-F$, the \emph{average} probability of a processor returning a bogus result is inferior to $\frac{1}{2}$. Second, the adversary may crash a subset of processors $F$, provided the size of $P-F$ is bounded from below. We consider two models: the first bounds the size of $P-F$ by a fractional polynomial, the second bounds this size by a poly-logarithm. Both models yield adversaries that are much stronger than previously studied. Our randomized synchronous algorithm is formulated for $n$ processors and $t$ tasks, with $n\le t$, where depending on the number of crashes each live processor is able to terminate dynamically with the knowledge that the problem is solved with high probability. For the adversary constrained by a fractional polynomial, the round complexity of the algorithm is $O(\frac{t}{n^\varepsilon}\log{n}\log{\log{n}})$, its work is $O(t\log{n} \log{\log{n}})$ and message complexity is $O(n\log{n}\log{\log{n}})$. For the poly-log constrained adversary, the round complexity is $O(t)$, work is $O(t n^{\varepsilon})$, %$O(t \, poly \log{n})$, and message complexity is $O(n^{1+\varepsilon})$ %$O(n \, poly \log{n})$. All bounds are shown to hold with high probability.

Internet supercomputing is an approach to solving partitionable, computation-intensive problems by harnessing the power of a vast number of interconnected computers. This paper presents a new algorithm for the problem of using network supercomputing to perform a large collection of independent tasks, while dealing with undependable processors. The adversary may cause the processors to return bogus results for tasks with certain probabilities, and may cause a subset $F$ of the initial set of processors $P$ to crash. The adversary is constrained in two ways. First, for the set of non-crashed processors $P-F$, the \emph{average} probability of a processor returning a bogus result is inferior to $\frac{1}{2}$. Second, the adversary may crash a subset of processors $F$, provided the size of $P-F$ is bounded from below. We consider two models: the first bounds the size of $P-F$ by a fractional polynomial, the second bounds this size by a poly-logarithm. Both models yield adversaries that are much stronger than previously studied. Our randomized synchronous algorithm is formulated for $n$ processors and $t$ tasks, with $n\le t$, where depending on the number of crashes each live processor is able to terminate dynamically with the knowledge that the problem is solved with high probability. For the adversary constrained by a fractional polynomial, the round complexity of the algorithm is $O(\frac{t}{n^\varepsilon}\log{n}\log{\log{n}})$, its work is $O(t\log{n} \log{\log{n}})$ and message complexity is $O(n\log{n}\log{\log{n}})$. For the poly-log constrained adversary, the round complexity is $O(t)$, work is $O(t n^{\varepsilon})$, %$O(t \, poly \log{n})$, and message complexity is $O(n^{1+\varepsilon})$ %$O(n \, poly \log{n})$. All bounds are shown to hold with high probability.

Current and future surveys of large-scale cosmic structure are associated with a massive and complex datastream to study, characterize, and ultimately understand the physics behind the two major components of the 'Dark Universe', dark energy and dark matter. In addition, the surveys also probe primordial perturbations and carry out fundamental measurements, such as determining the sum of neutrino masses. Large-scale simulations of structure formation in the Universe play a critical role in the interpretation of the data and extraction of the physics of interest. Just as survey instruments continue to grow in size and complexity, so do the supercomputers that enable these simulations. Here we report on HACC (Hardware/Hybrid Accelerated Cosmology Code), a recently developed and evolving cosmology N-body code framework, designed to run efficiently on diverse computing architectures and to scale to millions of cores and beyond. HACC can run on all current supercomputer architectures and supports a variety of programming models and algorithms. It has been demonstrated at scale on Cell- and GPU-accelerated systems, standard multi-core node clusters, and Blue Gene systems. HACC's design allows for ease of portability, and at the same time, high levels of sustained performance on the fastest supercomputers available. We present a description of the design philosophy of HACC, the underlying algorithms and code structure, and outline implementation details for several specific architectures. We show selected accuracy and performance results from some of the largest high resolution cosmological simulations so far performed, including benchmarks evolving more than 3.6 trillion particles.

Current and future surveys of large-scale cosmic structure are associated with a massive and complex datastream to study, characterize, and ultimately understand the physics behind the two major components of the 'Dark Universe', dark energy and dark matter. In addition, the surveys also probe primordial perturbations and carry out fundamental measurements, such as determining the sum of neutrino masses. Large-scale simulations of structure formation in the Universe play a critical role in the interpretation of the data and extraction of the physics of interest. Just as survey instruments continue to grow in size and complexity, so do the supercomputers that enable these simulations. Here we report on HACC (Hardware/Hybrid Accelerated Cosmology Code), a recently developed and evolving cosmology N-body code framework, designed to run efficiently on diverse computing architectures and to scale to millions of cores and beyond. HACC can run on all current supercomputer architectures and supports a variety of programming models and algorithms. It has been demonstrated at scale on Cell- and GPU-accelerated systems, standard multi-core node clusters, and Blue Gene systems. HACC's design allows for ease of portability, and at the same time, high levels of sustained performance on the fastest supercomputers available. We present a description of the design philosophy of HACC, the underlying algorithms and code structure, and outline implementation details for several specific architectures. We show selected accuracy and performance results from some of the largest high resolution cosmological simulations so far performed, including benchmarks evolving more than 3.6 trillion particles.

No Description

Volume 2-Computational Fluid dynamics and Wave Propagation; Parallel Computing; Concurrent and Supercomputing; Computational Physics/Computational Chemistry and Evolutionary Systems.

Larry Smarr speaks at the inaugural Graduate Studies Program in 2009 at the Singularity University

This instrumentation grant supported the purchase of a small supercomputer for cognitive brain imaging and other large 3D dataspaces. The machine that was purchased was an SGI (Silicon Graphics) Origin 2000 parallel supercomputer with 14 CPU's, 7 gigabytes of main memory, 750 gigabytes of disk storage, and 5 terabytes of fast tape storage. In addition, 6 SGI Unix workstations were purchased for graphics display and processing, and 9 NT/Intel workstations for local data analysis. The machines are networked in a 100mbit LAN. The area of computing in functional brain imaging and other massive 3-D dataspaces has extraordinary scientific, medical, and commercial potential. The potential application to DoD is the ability to measure brain function during the performance of high workload tasks, such as decision making, communication, and visualization. The measurement of the brain function provides a new way to measure cognitive workload, its limitations, and its efficiencies. These new methods have implications for the selection and training of personnel, and for the design of human computer interfaces. The research develops methods for measuring brain activation in the cortex during the performance of high level cognitive tasks and in dynamic decision making, in order to evaluate the cognitive workload imposed on an operator by various tasks, situations, interfaces, or instruments. The fMRI data are acquired at a high rate (one brain slice about 5 mm thick consisting of 64x128 voxels approximately every 200 msec), in tasks taking a total of about 15 minutes each. The data are then subjected to signal processing and statistical modeling techniques, to discover the spatial and temporal patterns of cortical activation in a network of brain areas that underlie higher level cognition.