Search NASA⌕ Search

SEARCH · Search NASA

Results for “PROCESSORS”

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 1,603 records · Page 89

A scalable geometric multigrid solver for nonsymmetric elliptic systems with application to variable-density flows

A geometric multigrid algorithm is introduced for solving nonsymmetric linear systems resulting from the discretization of the variable density Navier–Stokes equations on nonuniform structured rectilinear grids and high-Reynolds number flows. The restriction operation is defined such that the resulting system on the coarser grids is symmetric, thereby allowing for the use of efficient smoother algorithms. To achieve an optimal rate of convergence, the sequence of interpolation and restriction operations are determined through a dynamic procedure. A parallel partitioning strategy is introduced to minimize communication while maintaining the load balance between all processors. To test the proposed algorithm, we consider two cases: 1) homogeneous isotropic turbulence discretized on uniform grids and 2) turbulent duct flow discretized on stretched grids. Testing the algorithm on systems with up to a billion unknowns shows that the cost varies linearly with the number of unknowns. This O (N) behavior confirms the robustness of the proposed multigrid method regarding ill-conditioning of large systems characteristic of multiscale high-Reynolds number turbulent flows. The robustness of our method to density variations is established by considering cases where density varies sharply in space by a factor of up to 104, showing its applicability to two-phase flow problems. Strong and weak scalability studies are carried out, employing up to 30,000 processors, to examine the parallel performance of our implementation. Excellent scalability of our solver is shown for a granularity as low as 104 to 105 unknowns per processor. At its tested peak throughput, it solves approximately 4 billion unknowns per second employing over 16,000 processors with a parallel efficiency higher than 50%.

Computer Science↗

Efficient Heterogeneous Execution on Large Multicore and Accelerator Platforms: Case Study Using a Block Tridiagonal Solver

The algorithmic and implementation principles are explored in gainfully exploiting GPU accelerators in conjunction with multicore processors on high-end systems with large numbers of compute nodes, and evaluated in an implementation of a scalable block tridiagonal solver. The accelerator of each compute node is exploited in combination with multicore processors of that node in performing block-level linear algebra operations in the overall, distributed solver algorithm. Optimizations incorporated include: (1) an efficient memory mapping and synchronization interface to minimize data movement, (2) multi-process sharing of the accelerator within a node to obtain balanced load with multicore processors, and (3) an automatic memory management system to efficiently utilize accelerator memory when sub-matrices spill over the limits of device memory. Results are reported from our novel implementation that uses MAGMA and CUBLAS accelerator software systems simultaneously with ACML for multithreaded execution on processors. Overall, using 940 nVidia Tesla X2090 accelerators and 15,040 cores, the best heterogeneous execution delivers a 10.9-fold reduction in run time relative to an already efficient parallel multicore-only baseline implementation that is highly optimized with intra-node and inter-node concurrency and computation-communication overlap. Detailed quantitative results are presented to explain all critical runtime components contributing to hybrid performance.

Park, Alfred J↗

Effects of ultrasonic processing on phase transition of flame-synthesized anatase TiO{sub 2} nanoparticles

The effect of ultrasonic processing on the phase transformation of flame-synthesized anatase TiO{sub 2} nanoparticles heated to the rutile phase was investigated. TiO{sub 2} nanoparticles of various sizes were prepared using a coflow hydrogen diffusion flame and an ultrasonic processor. Smaller nanoparticles having a similar portion of anatase phase using the ultrasonic processor were produced. On the basis of scanning electron microscopy images and specific surface areas, we observed that smaller nanoparticles tended to be sintered more easily than larger nanoparticles. From X-ray diffraction analysis, we demonstrated that when heated, TiO{sub 2} nanoparticles synthesized using the ultrasonic processor at 60% of its maximum amplitude were transformed from the anatase phase to the rutile phase more easily than those formed without or with the ultrasonic processor operated at 20% of its maximum amplitude.

36 MATERIALS SCIENCE↗

Nonlinear encoding in diffractive information processing using linear optical materials

Nonlinear encoding of optical information can be achieved using various forms of data representation. Here, we analyze the performances of different nonlinear information encoding strategies that can be employed in diffractive optical processors based on linear materials and shed light on their utility and performance gaps compared to the state-of-the-art digital deep neural networks. For a comprehensive evaluation, we used different datasets to compare the statistical inference performance of simpler-to-implement nonlinear encoding strategies that involve, e.g., phase encoding, against data repetition-based nonlinear encoding strategies. We show that data repetition within a diffractive volume (e.g., through an optical cavity or cascaded introduction of the input data) causes the loss of the universal linear transformation capability of a diffractive optical processor. Therefore, data repetition-based diffractive blocks cannot provide optical analogs to fully connected or convolutional layers commonly employed in digital neural networks. However, they can still be effectively trained for specific inference tasks and achieve enhanced accuracy, benefiting from the nonlinear encoding of the input information. Our results also reveal that phase encoding of input information without data repetition provides a simpler nonlinear encoding strategy with comparable statistical inference accuracy to data repetition-based diffractive processors. Our analyses and conclusions would be of broad interest to explore the push-pull relationship between linear material-based diffractive optical systems and nonlinear encoding strategies in visual information processors.

42 ENGINEERING↗

All-optical phase conjugation using diffractive wavefront processing

Abstract Optical phase conjugation (OPC) is a nonlinear technique used for counteracting wavefront distortions, with applications ranging from imaging to beam focusing. Here, we present a diffractive wavefront processor to approximate all-optical phase conjugation. Leveraging deep learning, a set of diffractive layers was optimized to all-optically process an arbitrary phase-aberrated input field, producing an output field with a phase distribution that is the conjugate of the input wave. We experimentally validated this wavefront processor by 3D-fabricating diffractive layers and performing OPC on phase distortions never seen during training. Employing terahertz radiation, our diffractive processor successfully performed OPC through a shallow volume that axially spans tens of wavelengths. We also created a diffractive phase-conjugate mirror by combining deep learning-optimized diffractive layers with a standard mirror. Given its compact, passive and multi-wavelength nature, this diffractive wavefront processor can be used for various applications, e.g., turbidity suppression and aberration correction across different spectral bands.

42 ENGINEERING↗

Comparison of quantum advantage experiments using random circuit sampling

Random circuit sampling, the task of sampling bit strings from a random unitary operator, has been implemented to demonstrate quantum advantage on the Sycamore quantum processor with 53 qubits and on the Zuchongzhi quantum processor with 56 and 61 qubits. Recently, it was claimed that classical computers using tensor network simulation could catch on to current noisy quantum processors for random circuit sampling. While the linear cross-entropy benchmark fidelity was used to certify all these claims, it may not capture statistical properties of outputs in detail. Here, we compare the bit strings sampled from classical computers using tensor network simulation by Pan et al. [F. Pan, K. Chen, and P. Zhang, Phys. Rev. Lett. 129, 090502 (2022)] and by Kalachev et al. [G. Kalachev, P. Panteleev, P. Zhou, and M.-H. Yung, arXiv:2112.15083] with the bit strings from the Sycamore quantum processor. It is shown that all of Kalachev et al.'s samples passed the NIST random number tests. The heat maps of bit strings show that Pan et al.'s and Kalachev et al.'s samples are quite different from the Sycamore or Zuchongzhi samples. The analysis with the Marchenko-Pastur distribution and the Wasssertein distances demonstrates that Kalachev et al.'s samples are statistically closer to the Sycamore samples than Pan et al.'s while the three datasets have similar values for the linear cross-entropy fidelity. In conclusion, our finding implies that further study is needed to certify or beat the claims of quantum advantage using random circuit sampling.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

ROSE::FTTransform - A Source-to-Source Translation Framework for Exascale Fault-Tolerance Research

Exascale computing systems will require sufficient resilience to tolerate numerous types of hardware faults while still assuring correct program execution. Such extreme-scale machines are expected to be dominated by processors driven at lower voltages (near the minimum 0.5 volts for current transistors). At these voltage levels, the rate of transient errors increases dramatically due to the sensitivity to transient and geographically localized voltage drops on parts of the processor chip. To achieve power efficiency, these processors are likely to be streamlined and minimal, and thus they cannot be expected to handle transient errors entirely in hardware. Here we present an open, compiler-based framework to automate the armoring of High Performance Computing (HPC) software to protect it from these types of transient processor errors. We develop an open infrastructure to support research work in this area, and we define tools that, in the future, may provide more complete automated and/or semi-automated solutions to support software resiliency on future exascale architectures. Results demonstrate that our approach is feasible, pragmatic in how it can be separated from the software development process, and reasonably efficient (0% to 30% overhead for the Jacobi iteration on common hardware; and 20%, 40%, 26%, and 2% overhead for a randomly selected subset of benchmarks from the Livermore Loops [1]).

97 MATHEMATICS AND COMPUTING↗

SCORPIO: A scalable two-phase parallel I/O library with application to a large scale subsurface simulator

Inefficient parallel I/O is known to be a major bottleneck among scientific applications employed on supercomputers as the number of processor cores grows into the thousands. Our prior experience indicated that parallel I/O libraries such as HDF5 that rely on MPI-IO do not scale well beyond 10K processor cores, especially on parallel file systems (like Lustre) with single point of resource contention. Our previous optimization efforts for a massively parallel multi-phase and multi-component subsurface simulator (PFLOTRAN) led to a two-phase I/O approach at the application level where a set of designated processes participate in the I/O process by splitting the I/O operation into a communication phase and a disk I/O phase. The designated I/O processes are created by splitting the MPI global communicator into multiple sub-communicators. The root process in each sub-communicator is responsible for performing the I/O operations for the entire group and then distributing the data to rest of the group. This approach resulted in over 25X speedup in HDF I/O read performance and 3X speedup in write performance for PFLOTRAN at over 100K processor cores on the ORNL Jaguar supercomputer. This research describes the design and development of a general purpose parallel I/O library called Scorpio that incorporates our optimized two-phase I/O approach. The library provides a simplified higher level abstraction to the user, sitting atop existing parallel I/O libraries (such as HDF5) and implements optimized I/O access patterns that can scale on larger number of processors. Performance results with standard benchmark problems and PFLOTRAN indicate that our library is able to maintain the same speedups as before with the added flexibility of being applicable to a wider range of I/O intensive applications.

Sreepathi, Sarat↗

Implementation of a level 1 trigger system using high speed serial (VXS) techniques for the 12GeV high luminosity experimental programs at Thomas Jefferson National Accelerator Facility

We will demonstrate a hardware and firmware solution for a complete fully pipelined multi-crate trigger system that takes advantage of the elegant high speed VXS serial extensions for VME. This trigger system includes three sections starting with the front end crate trigger processor (CTP), a global Sub-System Processor (SSP) and a Trigger Supervisor that manages the timing, synchronization and front end event readout. Within a front end crate, trigger information is gathered from each 16 Channel, 12 bit Flash ADC module at 4 nS intervals via the VXS backplane, to a Crate Trigger Processor (CTP). Each Crate Trigger Processor receives these 500 MB/S VXS links from the 16 FADC-250 modules, aligns skewed data inherent of Aurora protocol, and performs real time crate level trigger algorithms. The algorithm results are encoded using a Reed-Solomon technique and transmission of this Level 1 trigger data is sent to the SSP using a multi-fiber link. The multi-fiber link achieves an aggregate trigger data transfer rate to the global trigger at 8 Gb/s. The SSP receives and decodes Reed-Solomon error correcting transmission from each crate, aligns the data, and performs the global level trigger algorithms. The entire trigger system is synchronous and operates at 250 MHz with the Trigger Supervisor managing not only the front end event readout, but also the distribution of the critical timing clocks, synchronization signals, and the global trigger signals to each front end readout crate. These signals are distributed to the front end crates on a separate fiber link and each crate is synchronized using a unique encoding scheme to guarantee that each front end crate is synchronous with a fixed latency, independent of the distance between each crate. The overall trigger signal latency is <3 uS, and the proposed 12GeV experiments at Jefferson Lab require up to 200KHz Level 1 trigger rate.

43 PARTICLE ACCELERATORS↗

An FPGA computing demo core for space charge simulation

In accelerator physics, space charge simulation requires large amount of computing power. In a particle system, each calculation requires time/resource consuming operations such as multiplications, divisions, and square roots. Because of the flexibility of field programmable gate arrays (FPGAs), we implemented this task with efficient use of the available computing resources and completely eliminated non-calculating operations that are indispensable in regular micro-processors (e.g. instruction fetch, instruction decoding, etc.). We designed and tested a 16-bit demo core for computing Coulomb's force in an Altera Cyclone II FPGA device. To save resources, the inverse square-root cube operation in our design is computed using a memory look-up table addressed with nine to ten most significant non-zero bits. At 200 MHz internal clock, our demo core reaches a throughput of 200 M pairs/s/core, faster than a typical 2 GHz micro-processor by about a factor of 10. Temperature and power consumption of FPGAs were also lower than those of micro-processors. Fast and convenient, FPGAs can serve as alternatives to time-consuming micro-processors for space charge simulation.

99 GENERAL AND MISCELLANEOUS↗

Efficient Sorting on the Tilera Manycore Architecture

e present an efficient implementation of the radix sort algo- rithm for the Tilera TILEPro64 processor. The TILEPro64 is one of the first successful commercial manycore processors. It is com- posed of 64 tiles interconnected through multiple fast Networks- on-chip and features a fully coherent, shared distributed cache. The architecture has a large degree of flexibility, and allows various optimization strategies. We describe how we mapped the algorithm to this architecture. We present an in-depth analysis of the optimizations for each phase of the algorithm with respect to the processor’s sustained performance. We discuss the overall throughput reached by our radix sort implementation (up to 132 MK/s) and show that it provides comparable or better performance-per-watt with respect to state-of-the art implemen- tations on x86 processors and graphic processing units.

manycore↗

Reduce Operations: Send Volume Balancing While Minimizing Latency

Communication hypergraph model was introduced in a two-phase setting for encapsulating multiple communication cost metrics (bandwidth and latency), which are proven to be important in parallelizing irregular applications. In the first phase, computational-task-to-processor assignment is performed with the objective of minimizing total volume while maintaining computational load balance. In the second phase, communication-task-to-processor assignment is performed with the objective of minimizing total number of messages while maintaining communication-volume balance. The reduce-communication hypergraph model suffers from failing to correctly encapsulate send-volume balancing. We propose a novel vertex weighting scheme that enables part weights to correctly encode send-volume loads of processors for send-volume balancing. The model also suffers from increasing the total communication volume during partitioning. To decrease this increase, we propose a method that utilizes the recursive bipartitioning framework and refines each bipartition by vertex swaps. For performance evaluation, we consider column-parallel SpMV, which is one of the most widely known applications in which the reduce-task assignment problem arises. Extensive experiments on 313 matrices show that, compared to the existing model, the proposed models achieve considerable improvements in all communication cost metrics. Furthermore, these improvements lead to an average decrease of 30% in parallel SpMV time on 512 processors for 70 matrices with high irregularity.

97 MATHEMATICS AND COMPUTING↗

True Load Balancing for Matricized Tensor Times Khatri-Rao Product

MTTKRP is the bottleneck operation in algorithms used to compute the CP tensor decomposition. For sparse tensors, utilizing the compressed sparse fibers (CSF) storage format and the CSF-oriented MTTKRP algorithms is important for both memory and computational efficiency on distributed-memory architectures. Existing intelligent tensor partitioning models assume the computational cost of MTTKRP to be proportional to the total number of nonzeros in the tensor. However, this is not the case for the CSF-oriented MTTKRP on distributed-memory architectures. We outline two deficiencies of nonzero-based intelligent partitioning models when CSF-oriented MTTKRP operations are performed locally: failure to encode processors' computational loads and increase in total computation due to fiber fragmentation. We focus on existing fine-grain hypergraph model and propose a novel vertex weighting scheme that enables this model encode correct computational loads of processors. We also propose to augment the fine-grain model by fiber nets for reducing the increase in total computational load via minimizing fiber fragmentation. In this way, the proposed model encodes minimizing the load of the bottleneck processor. In conclusion, parallel experiments with real-world sparse tensors on up to 1024 processors prove the validity of the outlined deficiencies and demonstrate the merit of our proposed improvements in terms of parallel runtimes.

97 MATHEMATICS AND COMPUTING↗

Efficient Approximation Algorithms for Weighted $b$-Matching

We describe a half-approximation algorithm, b-Suitor, for computing a b-Matching of maximum weight in a graph with weights on the edges. b-Matching is a generalization of the well-known Matching problem in graphs, where the objective is to choose a subset of M edges in the graph such that at most a speci ed number b(v) of edges in M are incident on each vertex v. Subject to this restriction we maximize the sum of the weights of the edges in M. We prove that the b-Suitor algorithm computes the same b-Matching as the one obtained by the greedy algorithm for the problem. We implement the algorithm on serial and shared-memory parallel processors, and compare its performance against a collection of approximation algorithms that have been proposed earlier. Our results show that the b-Suitor algorithm outperforms the Greedy and Locally Dominant edge algorithms by one to two orders of magnitude on a serial processor. The b-Suitor algorithm has a high degree of concurrency, and it scales well up to 240 threads on a shared memory multiprocessor. The b-Suitor algorithm outperforms the Locally Dominant edge algorithm by a factor of fourteen on 16 cores of an Intel Xeon multiprocessor.

97 MATHEMATICS AND COMPUTING↗

Messiah College Biodiesel Fuel Generation Project (Final Technical Report)

Many obvious and significant concerns arise when considering the concept of small-scale biodiesel production. Does the fuel produced meet the stringent requirements set by the commercial biodiesel industry? Is the process safe? How are small-scale producers collecting and transporting waste vegetable oil? How is waste from the biodiesel production process handled by small-scale producers? These concerns and many others were the focus of the research preformed in the Messiah College Biodiesel Fuel Generation project over the last three years. This project was a unique research program in which undergraduate engineering students at Messiah College set out to research the feasibility of small-biodiesel production for application on a campus of approximately 3000 students. This Department of Energy (DOE) funded research program developed out of almost a decade of small-scale biodiesel research and development work performed by students at Messiah College. Over the course of the last three years the research team focused on four key areas related to small-scale biodiesel production: Quality Testing and Assurance, Process and Processor Research, Process and Processor Development, and Community Education. The objectives for the Messiah College Biodiesel Fuel Generation Project included the following: 1. Preparing a laboratory facility for the development and optimization of processors and processes, ASTM quality assurance, and performance testing of biodiesel fuels. 2. Developing scalable processor and process designs suitable for ASTM certifiable small-scale biodiesel production, with the goals of cost reduction and increased quality. 3. Conduct research into biodiesel process improvement and cost optimization using various biodiesel feedstocks and production ingredients.

09 BIOMASS FUELS↗

Extreme Heterogeneity 2018 - Productive Computational Science in the Era of Extreme Heterogeneity: Report for DOE ASCR Workshop on Extreme Heterogeneity

The 2018 Basic Research Needs Workshop on Extreme Heterogeneity identified five Priority Research Directions for realizing the capabilities needed to address the challenges posed in this era of rapid technological change. This report captures the outcomes of that workshop. In the context of extreme heterogeneity, it defines basic research needs and opportunities in computer science research to develop smart and trainable operating and runtime systems, programming environments, and predictive tools that will make future systems easier to adapt to scientists’ computing needs and easier for facilities to deploy securely.

97 MATHEMATICS AND COMPUTING↗

Data Locality Enhancement of Dynamic Simulations for Exascale Computing (Final Report)

The development of modern processors exhibits two trends that complicate the optimizations of modern software. The first is the increasing sensitivity of processors' throughput to irregularities in computation. With more processors produced through a massive integration of simple cores, future systems will increasingly favor regular data-level parallel computations, but deviate from the needs of applications with complex patterns. Some evidences are already shown on Graphic Processing Units (GPU): Irregular data accesses (e.g., indirect references A[D[i]]) and conditional branches are limiting many GPU applications' performance at a level an order of magnitude lower than the peak of GPU. The second hardware trend is the growing gap between memory bandwidth and the aggregate speed—that is, the sum of all cores' computing power—of a Chip Multiprocessor (CMP). Despite the capped growth of the peak CPU speed, the aggregate speed of a CMP keeps increasing as more cores get into a single chip. It is expected that by 2018, node concurrency in an exascale system will increase by hundreds of times, whereas, memory bandwidth will expand by only 10 to 20 times. Consequently, data movement and storage is expected to consume more than 70% of the total system power. Bridging this gap is difficult; the complexities of modern CMP memory hierarchy make it even harder: Data cache becomes shared among computing units, and the sharing is often non-uniform—whether two computing units share a cache depends on their proximity and the level of the cache. On the recent IBM Power7 architecture, for instance, four hardware contexts (or SMT threads) in a core share the entire memory hierarchy, all cores in one chip share an on-chip L3 cache, and cores across chips share L3 and main memory through off-chip connections. These two trends complicate the translation of computing power into performance, especially for a program with either intensive data accesses or complex patterns in data accesses or control flow paths. Unfortunately, both attributes present and will persist in a class of important applications. For instance, many scientific simulations deal with a large volume of data. And meanwhile, as most real-world processes are non-uniform and evolving (e.g., the evolution of a galaxy or the process of a drug injection), both the computations and data accesses of these programs tend to be irregular and dynamically changing. Currently, the lack of support to these applications on modern CMP severely limits their performance. On GPU, as our recent study shows and other studies echo, performance enhancement of a factor of integers is possible when memory accesses or control flows are streamlined for a set of GPU applications. On multicore CPU, our studies show that traditional locality enhancement, for being oblivious to the new features of multicore memory hierarchy, may even cause large slowdown to data-intensive dynamic applications. The severity of the issues is expected to worsen as the two hardware trends continue. Some recent studies try to match software with the trends, but in a limited scope or manner. For irregularities on GPU, most studies focus on irregularities analyzable through static analysis (e.g., data accesses in regular loops). Dynamic irregularities are harder to address because the needed analysis and transformations typically have to happen at run time. Some other research resorts to hardware extensions, an actual adoption of which is unclear for the entailed space cost and complexity. For data locality, recent years have seen some exploitations of the new memory hierarchy on multicore for performance, but most of them are on process or thread scheduling, rather than program transformations. Our recent study reveals that program-level transformations may magnify the scheduling benefits by a factor of seven, concluding that program-level transformation should play a central role for data locality enhancement on modern CMP. But research in this direction has been sparse, and most have focused on data layout or cache performance modeling, rather than program transformations to match with the new memory hierarchy features. Overall, it is still an open question how to bridge the gap between dynamic computations and the two prominent properties of modern processors. The goal of this project is to develop a set of techniques and software tools to enhance the matching between memory accesses in dynamic simulations and the prominent features of modern and future CMP, alleviating the memory performance issues for petascale and exascale computing. This report summarizes the discoveries and products produced throughout this project. It includes free launch, a new software approach to overcoming the shortcomings of both methods; coherence-free multiview, an approach that allows multiple views of a single data object to co-exist on GPU memory during a GPU kernel execution; algorithmic optimizations to data analytics problems, especially those that involve lots of distance calculations. Supporting dynamic parallelism is important for GPU to benefit a broad range of applications. There are currently two fundamental ways for programs to exploit dynamic parallelism on GPU: a software-based approach with software-managed worklists, and a hardware-based approach through dynamic subkernel launches. Neither is satisfactory. The former is complicated to program and is often subject to some load imbalance; the latter suffers large runtime overhead. In this work, we propose free launch, a new software approach to overcoming the shortcomings of both methods. It allows programmers to use subkernel launches to express dynamic parallelism. It employs a novel compiler-based code transformation named subkernel launch removal to replace the subkernel launches with the reuse of parent threads. Coupled with an adaptive task assignment mechanism, the transformation reassigns the tasks in the subkernels to the parent threads with a good load balance. The technique requires no hardware extensions, immediately deployable on existing GPUs. It keeps the programming convenience of the subkernel launch-based approach while avoiding its large runtime overhead. Meanwhile, its superior load balancing makes it outperform manual worklist-based techniques by 3X on average. The work was published at the 48th Annual IEEE/ACM International Symposium on Microarchitecture (Micro'2015). A Graphic Processing Unit (GPU) system is typically equipped with many types of memory (e.g., global, constant, texture, shared, cache). Data placement determines what data are placed on which type of memory, essential for GPU memory performance. Prior optimizations of data placement always require a single view of a data object on memory, which limits the optimization effectiveness. In this work, we propose coherence-free multiview, an approach that allows multiple views of a single data object to co-exist on GPU memory during a GPU kernel execution. We demonstrate that under certain conditions, the multiple views can remain incoherent while facilitating enhanced data placement. We present a theorem and some compiler support to ensure the soundness of the usage of coherence-free multiview. We further develop reference-discerning data placement, a new way to enhance data placements on GPU. It enables more flexible data placements by using coherence-free multiview to leverage the slack in coherence requirement of some GPU programs. Experiments on three types of GPU systems show that, with less than 200KB space cost, the new data placement technique can provide a 1.6X average (up to 4.27X) speedup. In addition, we have examined algorithmic optimizations to data analytics problems, especially those that involve lots of distance calculations. Computing distances among data points is an essential part of many important algorithms in data analytics, graph analysis, and other domains. In each of these domains, developers have spent significant manual effort optimizing algorithms, often through novel applications of the triangle equality, in order to minimize the number of distance computations in the algorithms. In this work, we observe that many algorithms across these domains can be generalized as an instance of a generic distance-related abstraction. Based on this abstraction, we derive seven principles for correctly applying the triangular inequality to optimize distance-related algorithms. Guided by the findings, we develop Triangular OPtimizer (TOP), the first software framework that is able to automatically produce optimized algorithms that either matches or outperforms manually designed algorithms for solving distance-related problems. TOP achieves up to 237x speedups and 2.5X on average. The work has been published at the 32nd International Conference on Machine Learning (ICML'15) and the 41st International Conference on Very Large Data Bases (VLDB'15).

42 ENGINEERING↗

IceNet for FireBox - A Berkeley Warehouse-Scale Computer

Berkeley’s FireBox is a next-generation warehouse-scale computer (WSC) that utilizes the energy-efficiency and bandwidth density of integrated silicon-photonic interconnects to enable a new high-bandwidth and low-latency network fabric connecting thousands of compute nodes to petabytes of DRAM and Flash storage. The high bandwidth, low latency and high connectivity of FireBox’s WSC network fabric (IceNet) will enable dramatic improvements in the overall system energy efficiency enabling fine-grain power control on system resources (processors, links and memory/storage components). IceNet is a special 3-stage photonic Clos network architected to achieve ultra-low-latency connectivity between processor nodes and memory, drastically cutting down on the energy wasted in resource idling (processors and memory stalled due to pending network requests). This is achieved by integration of the first and last switch stages into processor/memory hub clients and by heavy over-provisioning of the high-radix middle switches (FlareSwitches). A key hardware component developed in this program is an active laser power management photonic integrated circuits called LightSpark. It interacts with the FlareSwitch and provides laser power to a subset of occupied switch ports, increasing the utilization of laser light in the photonic network by an order of magnitude. In addition to guiding the laser power where it is needed, the laser-power management module enables both wavelength and laser redundancy, significantly increasing the robustness of the system. The goal of the IceNet fabric is to enable communication between 1000s of processor nodes and PBs of memory/storage with <100ns latency, <10pJ/b wall-plug energy cost at multiple Pb/s of available connectivity bandwidth. These metrics represent two-orders of magnitude improvement with respect to the status of current data-center technology.

42 ENGINEERING↗