Exploit Instruction Level Parallelism Research Articles

Consistency Constraints for Mapping Dataflow Graphs to Hybrid Dataflow/von Neumann Architectures

Dataflow process networks (DPNs) provide a convenient model of computation that is often used to model system behavior in model-based designs. With fixed sets of nodes, they are also used as dataflow graphs as an intermediate program representation by compilers to uncover instruction-level parallelism of sequential programs. Many recent processor architectures, which are still von Neumann architectures, also use dataflow computing to increase their exploitation of instruction-level parallelism by exposing their datapaths so that the compiler can take care of the allocation of processing units (PUs), the execution schedules of instructions on the PUs, and the communication of intermediate values between PUs. If the communication paths are buffered, these architectures can be abstracted into a DPN architecture whose PUs and interconnection network are DPN nodes. In this article, we introduce a DPN abstraction of hybrid dataflow/von Neumann architectures and consider the mapping of the nodes of a given dataflow graph to the PUs of such a DPN architecture such that there are no conflicts due to the mapping of different nodes to the same PU. We express the allocation and scheduling constraints in terms of propositional logic for the original dataflow graph and for a modified version of the dataflow graph that simplifies the constraints by introducing levels using copy nodes, such that all nodes receive inputs only from nodes of the previous level. We also formulate equisatisfiable SMT constraints using integer variables to reason directly about the parallel runtime. On this basis, we further present alternative SAT constraints that explicitly encode concurrency, and discuss variants of the constraints for a better understanding of the same.

An Application Specific Vector Processor for Efficient Massive MIMO Processing

This paper presents an implementation for a baseband massive multiple-input multiple-output (MIMO) application-specific instruction set processor (ASIP). The ASIP is geared with vector processing capabilities in the form of single instruction multiple data (SIMD), and furthermore exploits instruction level parallelism by employing a very large instruction word (VLIW) architecture. Additionally, a systolic array is built into the pipeline which is tuned to speed up matrix calculations. A parallel memory subsystem and stand-alone accelerators are integrated into the ASIP architecture in order to meet the processing requirement. The processor is synthesized in 22 nm FD-SOI technology running at a clock frequency of 800 MHz. The system achieves a maximum detection throughput of 0.75 Gb/s/mm <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> for a <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$128\times 8$ </tex-math></inline-formula> massive MIMO system.

UltraSynth: Insights of a CGRA Integration into a Control Engineering Environment

Coarse Grained Reconfigurable Arrays (CGRAs) or Architectures are a concept for hardware accelerators based on the idea of distributing workload over Processing Elements. These processors exploit instruction level parallelism, while being energy efficient due to their simplistic internal structure. However, the incorporation into a complete computing system raises severe challenges at the hardware and software level. This article evaluates a CGRA integrated into a control engineering environment targeting a Xilinx Zynq System on Chip (SoC) in detail. Besides the actual application execution performance, the practicability of the configuration toolchain is validated. Challenges of the real-world integration are discussed and practical insights are highlighted.

Energy Efficient Low Latency Multi-issue Cores for Intelligent Always-On IoT Applications

Advanced Internet-of-Things applications require control-oriented codes to be executed with low latency for fast responsivity while their advanced signal processing and decision making tasks require computational capabilities. For this context, we propose three multi-issue core designs featuring an exposed datapath architecture with high performance, while retaining energy-efficiency. These features are achieved with exploitation of instruction-level parallelism, fast branching and the use of an instruction register file. With benchmarks in control-flow and signal processing application domains we measured in the best case 64% reduced energy consumption compared to a state-of-the-art RISC core, while consuming less silicon area. A high-performance design point reaches nearly 2.6 GHz operating frequency in the best case, over 2× improvement, while simultaneously achieving a 14% improvement in system energy-delay product.

A Comparative Analysis of HPL-PD and MIPS Architectures by Using Integrated Approach for IS and RA for Exploiting ILP

In this article, the authors have proposed an integrated algorithm for instruction scheduling and register allocation, and implemented it for compiler optimization in a machine description in a trimaran infrastructure for exploitation of instruction-level parallelism. For this experimental work, the authors added machine descriptions (MDES) targeted to HL-PD architecture. As a result, only a few spills were needed and the quality of the code generated was improved. The application of abstract base graph has been applied for experimental purposes. For these experiments, this article used 20 benchmarks available with trimaran infrastructure for HPL-PD architecture. This article compares some of these results with results obtained performed on LLVM compiler on an MIPS architecture. The implemented algorithm is based on subgraph isomorphism. The input program is represented in the form of directed acyclic graph (DAG). The vertices of the DAG represent the instructions, input, and output operands of the program, while the edges represent dependencies among the instructions.

Optimal Scheduling for Exposed Datapath Architectures with Buffered Processing Units by ASP

AbstractConventional processor architectures are restricted in exploiting instruction level parallelism (ILP) due to the relatively low number of programmer-visible registers. Therefore, more recent processor architectures expose their datapaths so that the compiler (1) can schedule parallel instructions to different processing units and (2) can make effective use of local storage of the processing units. Among these architectures, the Synchronous Control Asynchronous Dataflow (SCAD) architecture is a new exposed datapath architecture whose processing units are equipped with first-in first-out (FIFO) buffers at their input and output ports.In contrast to register-based machines, the optimal code generation for SCAD is still a matter of research. In particular, SAT and SMT solvers were used to generate optimal resource constrained and optimal time constrained schedules for SCAD, respectively. As Answer Set Programming (ASP) offers better flexibility in handling such scheduling problems, we focus in this paper on using an answer set solver for both resource and time constrained optimal SCAD code generation. As a major benefit of using ASP, we are able to generatealloptimal schedules for a given program which allows one to study their properties. Furthermore, the experimental results of this paper demonstrate that the answer set solver can compete with SAT solvers and outperforms SMT solvers.This paper is under consideration for acceptance in TPLP.

고급 언어에서 ASIP을 위한 전용 부호 생성 기술 연구

In this paper, we discuss a code generation technique for custom transport triggered architecture (TTA) from a high-level language structure. This methodology is implemented by using TTA-based Co-design Environment (TCE) tool. The results show how the scheduler exploits instruction level parallelism in the custom target architecture and source program. Thus, the scheduler generates parallel TTA instructions using lower cycle counts than the sequential scheduling algorithm. Moreover, we take Tensilica tool to make a comparison with TCE. Because of the efficiency of TTA, TCE takes less execution cycles compared to Tensilica configurations. Finally, this paper shows that it requires only 7 cycles to generate the parallel TTA instruction set for implementing Cyclic Redundancy Check (CRC) applications as an input design, and presents the code generation technique to move complexity from the processor software to hardware architecture. This method can be applicable lots of channel Codecs like CRC and source Codecs like High Efficiency Video Coding (HEVC).

Design and Development of Stream Processor Architecture for GPU Application Using Reconfigurable Computing

Graphical Processing Units (GPUs) have become an integral part of today’s mainstream computing systems. They are also being used as reprogrammable General Purpose GPUs (GP-GPUs) to perform complex scientific computations. Reconfigurability is an attractive approach to embedded systems allowing hardware level modification. Hence, there is a high demand for GPU designs based on reconfigurable hardware. Stream processor consists of clusters of functional units which provide a bandwidth hierarchy, supporting hundreds of arithmetic units. The arithmetic cluster units are designed to exploit instruction level parallelism and subword parallelism within a cluster and data parallelism across the clusters.For decreasing the area and power, a single controller is used to control data flow between clusters and between host processor and GPU. The designed of stream processor unit has been carried out in Verilog on Altera Quartus II and simulated using ModelSim tools. The functionality of the modelled blocks is verified using test inputs in the simulator.The simulated execution time of 8-bit pipelined multiplier is 60 ps and 100 ns for 8-bit pipelined adder while operating at 90 MHz.

Disjoint out-of-order execution processor

High-performance superscalar architectures used to exploit instruction level parallelism in single-thread applications have become too complex and power hungry for the multicore processors era. We propose a new architecture that uses multiple small latency-tolerant out-of-order cores to improve single-thread performance. Improving single-thread performance with multiple small out-of-order cores allows designers to place more of these cores on the same die. Consequently, emerging highly parallel applications can take full advantage of the multicore parallel hardware without sacrificing performance of inherently serial and hard to parallelize applications. Our architecture combines speculative multithreading (SpMT) with checkpoint recovery and continual flow pipeline architectures. It splits single-thread program execution into disjoint control and data threads that execute concurrently on multiple cooperating small and latency-tolerant out-of-order cores. Hence we call this style of execution Disjoint Out-of-Order Execution (DOE). DOE uses latency tolerance to overcome performance issues of SpMT caused by interthread data dependences. To evaluate this architecture, we have developed a microarchitecture performance model of DOE based on PTLSim, a simulation infrastructure of the x86 instruction set architecture. We evaluate the potential performance of DOE processor architecture using a simple heuristic to fork control independent threads in hardware at the target addresses of future procedure return instructions. Using applications from SpecInt 2000, we study DOE under ideal as well as realistic architectural constraints. We discuss the performance impact of key DOE architecture and application variables such as number of cores, interthread data dependences, intercore data communication delay, buffers capacity, and branch mispredictions. Without any DOE specific compiler optimizations, our results show that DOE outperforms conventional SpMT architectures by 15%, on average. We also show that DOE with four small cores can perform on average equally well to a large superscalar core, consuming about the same power. Most importantly, DOE improves throughput performance by a significant amount over a large superscalar core, up to 2.5 times, when running multitasking applications.

Time-Predictable Out-of-Order Execution for Hard Real-Time Systems

Superscalar out-of-order CPU designs can achieve higher performance than simpler in-order designs through exploitation of instruction-level parallelism in software. However, these CPU designs are often considered to be unsuitable for hard real-time systems because of the difficulty of guaranteeing the worst-case execution time (WCET) of software. This paper proposes and evaluates modifications for a superscalar out-of-order CPU core to allow instruction-level parallelism to be exploited without sacrificing time predictability and support for WCET analysis. Experiments using the M5 O3 CPU simulator show that WCETs can be two-four times smaller than those obtained using an idealized in-order CPU design, as instruction-level parallelism is exploited without compromising timing safety.

Exploiting an abstract-machine-based framework in the design of a Java ILP processor

Abstract machines bridge the gap between the high-level of programming languages and the low-level mechanisms of a real machine. The paper proposed a general abstract-machine-based framework (AMBF) to build instruction level parallelism processors using the instruction tagging technique. The constructed processor may accept code written in any (abstract or real) machine instruction set, and produce tagged machine code after data conflicts are resolved. This requires the construction of a tagging unit which emulates the sequential execution of the program using tags rather than actual values. The paper presents a Java ILP processor by using the proposed framework. The Java processor takes advantage of the tagging unit to dynamically translate Java bytecode instructions to RISC-like tag-based instructions to facilitate the use of a general-purpose RISC core and enable the exploitation of instruction level parallelism. We detailed the Java ILP processor architecture and the design issues. Benchmarking of the Java processor using SpecJVM98 and Linpack has shown the overall ILP speedup improvement between 78% and 173%.

A predecoding technique for ILP exploitation in Java processors

Java processors have been introduced to offer hardware acceleration for Java applications. They execute Java bytecodes directly in hardware. However, the stack nature of the Java virtual machine instruction set imposes a limitation on the achievable execution performance. In order to exploit instruction level parallelism and allow out of order execution, we must remove the stack completely. This can be achieved by recursive stack folding algorithms, such as OPEX, which dynamically transform groups of Java bytecodes to RISC like instructions. However, the decoding throughputs that are obtained are limited. In this paper, we explore microarchitectural techniques to improve the decoding throughput of Java processors. Our techniques are based on the use of a predecoded cache to store the folding results, so that it could be reused. The ultimate goal is to exploit every possible instruction level parallelism in Java programs by having a superscalar out of order core in the backend being fed at a sustainable rate. With the use of a predecoded cache of 2x2048 entries and a 4-way superscalar core we have from 4.8 to 18.3 times better performance than an architecture employing pattern based folding.

Early Register Deallocation Mechanisms Using Checkpointed Register Files

Modern superscalar microprocessors need sizable register files to support a large number of in-flight instructions for exploiting instruction level parallelism (ILP). An alternative to building large register files is to use a smaller number of registers, but manage them more effectively. More efficient management of registers can also result in higher performance if the reduction of the register file size is not the goal. Traditional register file management mechanisms deallocate a physical register only when the next instruction writing to the same destination architectural register commits. In this paper, we propose several techniques for deallocating physical registers much earlier. Our designs rely on the use of a checkpointed register file (CRF), where a local shadow copy of each bitcell is used to temporarily save the values of the early deallocated registers should they be needed to recover from branch mispredictions or to reconstruct the precise state after exceptions or interrupts. The proposed techniques try to release registers as soon as possible and are more aggressive than the previously proposed schemes for early deallocation of registers

A RISC architecture extended by an efficient tightly coupled reconfigurable unit

In this paper, the architecture of an embedded processor extended with a tightly-coupled coarse-grain reconfigurable functional unit (RFU) is proposed. The efficient integration of the RFU with the control unit and the datapath of the processor eliminate the communication overhead between them. To speed up execution, the RFU exploits instruction level parallelism (ILP) and spatial computation. Also, the proposed integration of the RFU efficiently exploits the pipeline structure of the processor, leading to further performance improvements. Furthermore, a development framework for the introduced architecture is presented. The framework is fully automated, hiding all reconfigurable hardware related issues from the user. The hardware model of the architecture was synthesized in a 0.13 µm process and all information regarding area and delay were estimated and presented. A set of benchmarks is used to evaluate the architecture and the development framework. Experimental results prove performance improvements in addition to potential energy reduction.

A Simulation and Exploration Technology for Multimedia-Application-Driven Architectures

The increasing of computational power requirements for DSP and Multimedia application and the needs of easy-to-program development environment has driven recent programmable devices toward Very Long Instruction Word (VLIW) [1] architectures and Hw-Sw co-design environments [2]. VLIW architecture allows generating optimized machine code from high-level languages exploiting Instruction Level Parallelism (ILP) [3]. Furthermore, applications requirements and time to market constraints are growing dramatically moving functionalities toward System on Chip (SoC) direction. This paper presents VLIW-SIM, an Application-Driven Architecture-design approach based on Instruction Set simulation. VLIW architectures and Instruction Set simulation were chosen to fulfill multimedia domain requirements and to implement an efficient Hw-Sw co-design environment. The VLIW-SIM simulation technology is based on pipeline status modeling, Simulation cache and Simulation Oriented Hw description. An effective support for Hw-Sw co-design requires high simulation performance (in terms of Simulated Instruction per Second--SIPS), flexibility (the ability to represent a number of different architectures) and cycle accuracy. There is a strong trade-off between these features: cycle accurate or close to cycle accurate simulation have usually low performance [4, 5]. Good simulation performance can be obtained loosing the simulator flexibility. Moreover SoC simulation requires a further degree of flexibility in simulating different components (core, co-processors, memories, buses). The proposed approach is focused on interpretative (not compiled [6]) re-configurable Instruction Set Simulator (ISS) in order to support both application design and architecture exploration. VLIW-SIM main features are: efficient host resource allocation, Instruction Set and Architecture description Flexibility (Instruction Set Dynamic Generation and Simulation Oriented Hardware Description), Step by step pipeline status tracking, Simulation Speed and Accuracy. Performance of simulation test for three validation case studies (TI TMS320C62x, TI TMS320C64x and ST200) are reported.

Software and Hardware Techniques to Optimize Register File Utilization in VLIW Architectures

High-performance microprocessors are currently designed with the purpose of exploiting instruction level parallelism (ILP). The techniques used in their design and the aggressive scheduling techniques used to exploit this ILP tend to increase the register requirements of the loops. This paper reviews hardware and software techniques that alleviate the high register demands of aggressive scheduling heuristics on VLIW cores. From the software point of view, instruction scheduling can stretch lifetimes and reduce the register pressure. If more registers than those available in the architecture are required, some actions (such as the injection of spill code) have to be applied to reduce this pressure, at the expense of some performance degradation. From the hardware point of view, this degradation could be reduced if a high-capacity register file were included without causing a negative impact on the design of the processor (cycle time, area and power dissipation). Novel organizations for the register file based on clustering and hierarchical organization are necessary to meet the technology constraints. This paper proposes the used of a clustered organization and proposes an aggressive instruction scheduling technique that minimizes the negative effect of the limitations imposed by the register file organization.

The JAFARDD processor: a Java architecture based on a folding algorithm, with reservation stations, dynamic translation, and dual processing

This paper presents the JAFARDD (a Java Architecture based on a Folding Algorithm, with Reservation stations, Dynamic translation, and Dual processing) processor. JAFARDD dynamically translates Java stack-dependent bytecodes to RISC-style stack-independent instructions to facilitate the use of a general-purpose RISC core. JAFARDD enables the exploitation of instruction level parallelism among the translated instructions by the use of bytecode folding coupled with Tomasulo's algorithm. We detail the JAFARDD architecture and the global architecture design principles observed while designing each pipeline module. We also illustrate the flow of the Java bytecodes through each of the processing phases. Benchmarking of JAFARDD using SPECjvm98 has shown a performance improvement between 1.10 and 2.25.

Evaluation and choice of various brånch predictors for low-power embedded processor

Power is an important design constraint in embedded computing systems. To meet the power constraint, microarchitecture and hardware designed to achieve high performance need to be revisited, from both performance and power angles. This paper studies one of them: branch predictor. As well known, branch prediction is critical to exploit instruction level parallelism effectively, but may incur additional power consumption due to the hardware resource dedicated for branch prediction and the extra power consumed on mispredicted branches. This paper explores the design space of branch prediction mechanisms and tries to find the most beneficial one to realize Iow-power embedded processor. The sample processor studied is Godson-like processor, which is a dual-issue, out-of-order processor with deep pipeline, supporting MIPS instruction set.

MediaBreeze

Decoupled architectures are fine-grain processors that partition the memory access and execute functions in a computer program and exploit the parallelism between the two functions. Although some concepts from the traditional decoupled access execute paradigm made its way into commercial processors, they encountered resistance in general-purpose applications because these applications are not very structured and regular. However, multimedia applications have recently become dominant workload on desktops and workstations. Media applications are very structured and regular and lend themselves well to the decoupling concept. In this paper, we present an architecture that decouples the useful/true computations from the overhead/supporting instructions in media applications. The proposed scheme is incorporated into an out-of-order general-purpose processor enhanced with SIMD extensions. Explicit hardware support is provided to exploit instruction level parallelism in the overhead component. Performance evaluation shows that such hardware can significantly improve performance over conventional SIMD enhanced general-purpose processors. Results on nine multimedia benchmarks show that the proposed MediaBreeze architecture provides a 1.05x to 16.7x performance improvement over a 2-way out-of-order SIMD machine. On introducing slip-based data prefetching, a performance improvement up to 28x is observed.

Characterization and Parallelization of Decision-Tree Induction

This paper examines the performance and memory-access behavior of the C4.5 decision-tree induction program, a representative example of data mining applications, for both uniprocessor and parallel implementations. The goals of this paper are to characterize C4.5, in particular its memory hierarchy usage, and to decrease the run-time of C4.5 via algorithmic improvement and parallelization. Performance is studied via RSIM, an execution driven simulator, for three uniprocessor models that exploit instruction level parallelism to varying degrees. This paper makes the following four contributions. The first contribution is presenting a complete characterization of the C4.5 decision-tree induction program. The results show that with the exception of the input data set, the working set fits into an 8-Kbyte data cache; the instruction working set also fits into an 8-Kbyte instruction cache. For data sets larger than the L2 cache, performance is limited by accesses to main memory. The results further establish that four-way issue can provide up to a factor of two performance improvement over single-issue for larger L2 caches; for smaller L2 caches, out-of-order dispatch provides a large performance improvement over in-order dispatch. The second contribution made by this paper is examining the effect on the memory hierarchy of changing the layout of the input dataset in memory, showing again that the performance is limited by memory accesses. One proposed data layout decreases the dynamic instruction count by up to 24%, but usually results in lower performance due to worse cache behavior. Another proposed data layout does not improve the dynamic instruction count over the original layout, but has better cache behavior and decreases the run-time by up to a factor of two. Third, this paper presents the first decision-tree induction program parallelized for a ccNUMA architecture. A method for splitting the decision tree hash table is discussed that allows the hash table to be updated and accessed simultaneously without the use of locks. The performance of the parallel version is compared to the original version of C4.5 and a uniprocessor version of C4.5 using the best data layout found. Speedup curves from a six-processor Sun E4000 SMP system show a speedup on the induction step of 3.99, and simulation results show that the performance is mostly unaffected by increasing the remote memory access time until it is over a factor of ten greater than the local memory access time. Last, this paper characterizes the parallelized decision-tree induction program. Compared to the uniprocessor version, the parallel version exerts significantly less pressure on the memory hierarchy, with the exception of having a much larger first level data working set.