Search NASA⌕ Search

SEARCH · Search NASA

Results for “Compiler techniques”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 127 records · Page 7

A performance model for GPUs with caches

To exploit the abundant computational power of the world's fastest supercomputers, an even workload distribution to the typically heterogeneous compute devices is necessary. While relatively accurate performance models exist for conventional CPUs, accurate performance estimation models for modern GPUs do not exist. This paper presents two accurate models for modern GPUs: a sampling-based linear model, and a model based on machine-learning (ML) techniques which improves the accuracy of the linear model and is applicable to modern GPUs with and without caches. We first construct the sampling-based linear model to predict the runtime of an arbitrary OpenCL kernel. Based on an analysis of NVIDIA GPUs' scheduling policies we determine the earliest sampling points that allow an accurate estimation. The linear model cannot capture well the significant effects that memory coalescing or caching as implemented in modern GPUs have on performance. We therefore propose a model based on ML techniques that takes several compiler-generated statistics about the kernel as well as the GPU's hardware performance counters as additional inputs to obtain a more accurate runtime performance estimation for modern GPUs. We demonstrate the effectiveness and broad applicability of the model by applying it to three different NVIDIA GPU architectures and one AMD GPU architecture. On an extensive set of OpenCL benchmarks, on average, the proposed model estimates the runtime performance with less than 7 percent error for a second-generation GTX 280 with no on-chip caches and less than 5 percent for the Fermi-based GTX 580 with hardware caches. On the Kepler-based GTX 680, the linear model has an error of less than 10 percent. On an AMD GPU architecture, Radeon HD 6970, the model estimates with 8 percent of error rates. As a result, the proposed technique outperforms existing models by a factor of 5 to 6 in terms of accuracy.

97 MATHEMATICS AND COMPUTING↗

Measurement technology: A compilation

Technical information is presented on measurement techniques and instruments, measurement applications for inspection activities, measurement sensors, and data conversion methods. Photographs or diagrams are included for each instrument or method described, and where applicable, patent information is given.

Source record↗

Optics and lasers: A compilation

A number of innovative devices and techniques in optics and related fields were presented. The following areas were covered: advances in laser and holography technology, articles on spectroscopy and general optics, new information in the area of photography.

Source record↗

Domain-specific compilers for dynamic simulations of quantum materials on quantum computers

Abstract Simulation of the dynamics of quantum materials is emerging as a promising scientific application for noisy intermediate-scale quantum (NISQ) computers. Due to their high gate-error rates and short decoherence times, however, NISQ computers can only produce high-fidelity results for those quantum circuits smaller than some given circuit size. Dynamic simulations, therefore, pose a challenge as current algorithms produce circuits that grow in size with each subsequent time-step of the simulation. This underscores the crucial role of quantum circuit compilers to produce executable quantum circuits of minimal size, thereby maximizing the range of physical phenomena that can be studied within the NISQ fidelity budget. Here, we present two domain-specific (DS) quantum circuit compilers for the Rigetti and IBM quantum computers, specifically designed to compile circuits simulating dynamics under a special class of time-dependent Hamiltonians. The compilers outperform state-of-the-art general-purpose compilers in terms of circuit size reduction by around 25%–30% as well as wall-clock compilation time by around 40% (dependent on system size and simulation time-step). Drawing on heuristic techniques commonly used in artificial intelligence, both compilers scale well with simulation time-step and system size. Code for both compilers is open-source and packaged into a full-stack quantum simulation software with tutorials included for ease of use for future researchers wishing to perform dynamic simulations of quantum materials on quantum computers. As our DS compilers provide significant improvements in both compilation time and simulation fidelity, they provide a building block for accelerating progress toward physical quantum supremacy.

Physics↗

Optimization guide for programs compiled under IBM FORTRAN H (OPT=2)

Guidelines are given to provide the programmer with various techniques for optimizing programs when the FORTRAN IV H compiler is used with OPT=2. Subroutines and programs are described in the appendices along with a timing summary of all the examples given in the manual.

Smith, D. M.↗

The application of trend surface analysis to a portion of the Apollo 15 X-ray fluorescence data

X-ray fluorescence data for 8 and 16 second time integrals gathered by Apollo 15 in circum lunar orbit were analyzed to determine the capability for chemical mapping of relatively small lunar features in a portion of Tranquillitatis and Serenitatis basins. Spatial mapping using trend surface analysis demonstrated that a useable signal could be extracted from Al/Si intensity ratios calculated for 8 second time spans. Reliability of the Al/Si ratio was enhanced when 16 second data were compiled using a sliding average technique. Residual anomalies from the trend surface mapping were identified and correlated with relatively small lunar surface features.

Podwysocki, M. H.↗

Avoiding and tolerating latency in large-scale next-generation shared-memory multiprocessors

A scalable solution to the memory-latency problem is necessary to prevent the large latencies of synchronization and memory operations inherent in large-scale shared-memory multiprocessors from reducing high performance. We distinguish latency avoidance and latency tolerance. Latency is avoided when data is brought to nearby locales for future reference. Latency is tolerated when references are overlapped with other computation. Latency-avoiding locales include: processor registers, data caches used temporally, and nearby memory modules. Tolerating communication latency requires parallelism, allowing the overlap of communication and computation. Latency-tolerating techniques include: vector pipelining, data caches used spatially, prefetching in various forms, and multithreading in various forms. Relaxing the consistency model permits increased use of avoidance and tolerance techniques. Each model is a mapping from the program text to sets of partial orders on program operations; it is a convention about which temporal precedences among program operations are necessary. Information about temporal locality and parallelism constrains the use of avoidance and tolerance techniques. Suitable architectural primitives and compiler technology are required to exploit the increased freedom to reorder and overlap operations in relaxed models.

Probst, David K.↗

Automated synthesis and verification of configurable DRAM blocks for ASIC's

A highly flexible embedded DRAM compiler is developed which can generate DRAM blocks in the range of 256 bits to 256 Kbits. The compiler is capable of automatically verifying the functionality of the generated DRAM modules. The fully automated verification capability is a key feature that ensures the reliability of the generated blocks. The compiler's architecture, algorithms, verification techniques and the implementation methodology are presented.

Pakkurti, M.↗

Instrumentation Working Group Summary

The Instrumentation Working Group compiled a summary of measurement techniques applicable to gas turbine engine aerosol precursors and particulates. An assessment was made of the limits, accuracy, applicability, and technology readiness of the various techniques. Despite advances made in emissions characterization of aircraft engines, uncertainties still exist in the mechanisms by which aerosols and particulates are produced in the near-field engine exhaust. To adequately assess current understanding of the formation of sulfuric acid aerosols in the exhaust plumes of gas turbine engines, measurements are required to determine the degree and importance of sulfur oxidation in the turbine and at the engine exit. Ideally, concentrations of all sulfur species would be acquired, with emphasis on SO2 and SO3. Numerous options exist for extractive and non-extractive measurement of SO2 at the engine exit, most of which are well developed. SO2 measurements should be performed first to place an upper bound on the percentage of SO2 oxidation. If extractive and non-extractive techniques indicate that a large amount of the fuel sulfur is not detected as SO2, then efforts are needed to improve techniques for SO3 measurements. Additional work will be required to account for the fuel sulfur in the engine exhaust. Chemical Ionization Mass Spectrometry (CI-MS) measurements need to be pursued, although a careful assessment needs to be made of the sampling line impact on the extracted sample composition. Efforts should also be placed on implementing non-intrusive techniques and extending their capabilities by maximizing exhaust coverage for line-of-sight measurements, as well as development of 2-D techniques, where feasible. Recommendations were made to continue engine exit and combustor measurements of particulates. Particulate measurements should include particle size distribution, mass fraction, hydration properties, and volatile fraction. However, methods to ensure that unaltered samples are obtained need to be developed. Particulate speciation was also assigned a high priority for quantifying the fractions of carbon soot, PAH, refractory materials, metals, sulfates, and nitrates. High priority was also placed on performing a comparison of particle sizing instruments. Concern was expressed by the workshop attendees who routinely make particulate measurements about the variation in number density measured during in-flight tests by different instruments. In some cases, measurements performed by different groups of researchers during the same flight tests showed an order of magnitude variation. Second priority was assigned to measuring concentrations of odd hydrogen and oxidizing species. Since OH, HO2, H2O2, and O are extremely reactive, non-extractive measurements are recommended. A combination of absorption and fluorescence is anticipated to be effective for OH measurements in the combustor and at the engine exit. Extractive measurements of HO2 have been made in the stratosphere, where the ambient level of OH is relatively low. Use of techniques that convert HO2 to OH for combustor and engine exit measurements needs to be evaluated, since the ratio of HO2/OH may be 1% or less at both the combustor and engine exit. CI-MS might be a viable option for H2O2, subject to sampling line conversion issues. However, H2O2 is a low priority oxidizing species in the combustor and at the engine exit. Two candidates for atomic oxygen measurements are Resonance Enhanced Multi-Photon Ionization (REMPI) and Laser-Induced Fluorescence (LIF). Particulate measurement by simultaneous extractive and non-extractive techniques was given equal priority to the oxidizer measurements. Concern was expressed over the ability of typical ground test sampling lines to deliver an unaltered sample to a remotely located instrument. It was suggested that the sampling probe and line losses be checked out by attempting measurements using an optical or non-extractive technique immediately upstream of the sampling probe. This is a possible application for Laser Induced Incandescence (LII) as a check on the volume fraction of soot. Optical measurements of size distribution are not well developed for ultrafine particles less than about 20 nm in diameter, so a non-extractive technique for particulate size distribution cannot be recommended without further development. Carbon dioxide measurements need to be made to complement other extractive measurement techniques. CO2 measurements enable conversion of other species concentrations to emission indices. Carbon monoxide, which acts as a sink for oxidizing species, should be measured using non-extractive techniques. CO can be rapidly converted to CO2 in extractive probes, and a comparison between extractive and non-extractive measurements should be performed. Development of non-extractive techniques would help to assess the degree of CO conversion, and might be needed to improve the concentration measurement accuracy. Measurements of NO(x) will continue to be critical due to the role of NO and NO2 in atmospheric chemistry, and their influence on atmospheric ozone. Time-resolved measurements of temperature, velocity, and species concentrations were included on the list of desired measurement. Thermocouples are typically adequate for engine exit measurements. PIV and LDV are well established for obtaining velocity profiles. The techniques are listed in the accompanying table; are divided into extractive and non-extractive techniques. Efforts were made to include a measurement uncertainty for each technique. An assessment of the technology readiness was included.

Zaller, Michelle↗

Three Proposed Compendia for Genesis Solar Wind Samples: Science Results, Collector Materials Characterization and Cleaning Techniques

Final Paper and not the abstract is attached. Introduction: Planetary material and cosmochemistry research using Genesis solar wind samples (including the development and implementation of cleaning and analytical techniques) has matured sufficiently that compilations on several topics, if made publically accessible, would be beneficial for researchers and reviewers. We propose here three compendia based on content, organization and source of documents (e.g. published peer-reviewed, published, internal memos, archives). For planning purposes, suggestions are solicited from potential users of Genesis solar wind samples for the type of science content and/or organizational style that would be most useful to them. These compendia are proposed as living documents, periodically updated. Similar to the existing compendia described below, the curation compendia are like library or archival finding aids, they are guides to published or archival documents and should not be cited as primary sources.

Allton, J. H.↗

Analysis of memory use for improved design and compile-time allocation of local memory

Trace analysis techniques are used to study memory referencing behavior for the purpose of designing local memories and determining how to allocate them for data and instructions. In an attempt to assess the inherent behavior of the source code, the trace analysis system described here reduced the effects of the compiler and host architecture on the trace by using a technical called flattening. The variables in the trace, their associated single-assignment values, and references are histogrammed on the basis of various parameters describing memory referencing behavior. Bounds are developed specifying the amount of memory space required to store all live values in a particular histogram class. The reduction achieved in main memory traffic by allocating local memory is specified for each class.

Mcniven, Geoffrey D.↗

Compiler-assisted static checkpoint insertion

This paper describes a compiler-assisted approach for static checkpoint insertion. Instead of fixing the checkpoint location before program execution, a compiler enhanced polling mechanism is utilized to maintain both the desired checkpoint intervals and reproducible checkpoint 1ocations. The technique has been implemented in a GNU CC compiler for Sun 3 and Sun 4 (Sparc) processors. Experiments demonstrate that the approach provides for stable checkpoint intervals and reproducible checkpoint placements with performance overhead comparable to a previously presented compiler assisted dynamic scheme (CATCH) utilizing the system clock.

Long, Junsheng↗

Compilation of directly measured nuclear spins of ground states and long-lived isomers

A compilation of the nuclear spins of ground and isomeric states measured by direct methods is presented. The first compilation of direct measurements of nuclear spins and moments was published in 1976 (G. H. Fuller, J. Phys. Chem. Ref. Data 5, 835, (1976)) with literature covered up to 1974. To our knowledge, the present work is the first such compilation since then. It is anticipated that the area of direct spin measurement will continue to expand using the state-of-the-art radioactive ion-beam and laser techniques. Literature cutoff date for the present compilation is February 2013. It is intended that the present compilation will be kept updated in a timely manner.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

Molten Salt Sampling Techniques and Analytical Approaches

Recent global interest in pyroprocessing and molten salt reactors has brought salt sampling methods and techniques back to the forefront of nuclear safeguards concerns. Issues with uranium supplies have also encouraged various countries to pursue advanced nuclear fuel cycles. Tracking nuclear material in molten salt has proven to be a challenge and updating molten salt sampling will greatly help in this endeavor. Molten salt is problematic to sample due to salt stratification, lack of homogeneity, solids, and difficulty with hot cell adaptations. Various salt sampling techniques have been used since before the 1960s including surface, spoon/spatula, and bar solidification. Since then, new types of sampling techniques have been developed to improve sampling results. These include rod/dip, pipet, suction, filtered sampling along with devices such as the Valve Core Sampler and the Multi-Level Sampler. These different approaches are being analyzed and improved upon along with developing requirements for an improved salt sampling device. Work continues to develop salt samplers that are more robust, easier to segment, collect at a specific depth, can work with filters, and can collect fines. Sampling parameters are also being narrowed in terms of stirring, settling time, filtration, depth, etc. In the future, we hope to address deficiencies for process control and nuclear material accountancy control by determining the best way to collect samples that minimizes contaminants and is representative. A compilation of salt sampling approaches, analyses techniques, and an evaluation of findings will be presented.

11 NUCLEAR FUEL CYCLE AND FUEL MATERIALS↗

Centrality determination with a forward detector in the RHIC Beam Energy Scan

Recently, Chatterjee et al [1] used a hadronic transport model to estimate the resolution with which various experimental quantities select the impact parameter of relativistic heavy ion collisions at collision energies relevant to the Beam Energy Scan (BES) program at the Relativistic Heavy Ion Collider (RHIC). Measures based on particle multiplicity at forward rapidity were found to be significantly worse than those based on midrapidity multiplicity. Using the same model, we show that a slightly more sophisticated measure greatly improves the resolution based on forward rapidity particles; this improvement persists even when the model is filtered through a realistic simulation of a recent upgrade detector to the STAR experiment. Furthermore, these results highlight the importance of optimizing centrality measures based on particles detected at forward rapidity, especially for experimental studies that search for a critical point in the QCD phase diagram. Such measurements usually focus on proton multiplicity fluctuations at midrapidity, hence selecting events based on multiplicity at midrapidity raises the possibility of nontrivial autocorrelations.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

The Role of Reformulation in the Automatic Design of Satisfiability Procedures

Recently there has been increasing interest in the problem of knowledge compilation (Selman & Kautz91). This is the problem of identifying tractable techniques for determining the consequences of a knowledge base. We have developed and implemented a technique, called DRAT, that given a theory, i.e., a collection of firstorder clauses, can often produce a type of decision procedure for that theory that can be used in the place of a general-purpose first-order theorem prover for determining many of the consequences of that theory. Hence, DRAT does a type of knowledge compilation. Central to the DRAT technique is a type of reformulation in which a problem's clauses are restated in terms of different nonlogical symbols. The reformulation is isomorphic in the sense that it does not change the semantics of a problem.

VanBaalen, Jeffrey↗