The DARPA image understanding benchmark for parallel computers

doi:10.1016/0743-7315(91)90067-J

Journal of Parallel and Distributed Computing

Volume 11, Issue 1, January 1991, Pages 1-24

https://doi.org/10.1016/0743-7315(91)90067-J Get rights and content

Abstract

This paper describes a new effort to evaluate parallel architectures applied to knowledge-based machine vision. Previous vision benchmarks have considered only execution times for isolated vision-related tasks, or a very simple image processing scenario. However, the performance of an image interpretation system depends upon a wide range of operations on different levels of representations, from processing arrays of pixels, through manipulation of extracted image events, to symbolic processing of stored models. Vision is also characterized by both bottom-up (image-based) and top-down (model-directed) processing. Thus, the costs of interactions between tasks, input and output, and system overhead must be taken into consideration. Therefore, this new benchmark addresses the issue of system performance on an integrated set of tasks. The Integrated Image Understanding Benchmark consists of a model-based object recognition problem, given two sources of sensory input, intensity and range data, and a database of candidate models. The models consist of configurations of rectangular surfaces, floating in space, viewed under orthographic projection, with the presence of both noise and spurious nonmodel surfaces. A partially ordered sequence of operations that solves the problem is specified along with a recommended algorithmic method for each step. In addition to reporting the total time and the final solution, timings are requested for each component operation, and intermediate results are output as a check on accuracy. Other factors such as programming time, language, code size, and machine configurations are reported. As a result, the benchmark can be used to gain insight into processor strengths and weaknesses and may thus help to guide the development of the next generation of parallel vision architectures. In addition to discussing the development and specification of the new benchmark, this paper presents the results from running the benchmark on the Connection Machine, Warp, Image Understanding Architecture, Associative String Processor, Alliant FX-80, and Sequent Symmetry. The results are discussed and compared through a measurement of relative effort, which factors out the effects of differing technologies.

References (15)

R.J. Carpenter
Performance measurement instrumentation for multiprocessor computers
A.N. Choudhary
Parallel architectures and parallel algorithms for integrated vision systems
M.J.B. Duff
How not to benchmark image processors
D.W. Hillis
The Connection Machine
(1986)
H.T. Kung et al.
Warp: A programmable systolic array processor
R.M. Lea
ASP: A cost-effective parallel microcomputer
IEEE Micro.
(Oct. 1988)
K. Preston
Benchmark results: The Abingdon Cross

There are more references available in the full text version of this article.

Cited by (52)

The impact of grain size on the efficiency of embedded SIMD image processing architectures
2004, Journal of Parallel and Distributed Computing
Pixel-per-processing element (PPE) ratio—the amount of image data directly mapped to each processing element—has a significant impact on the area and energy efficiency of embedded SIMD architectures for image processing applications. This paper quantitatively evaluates the impact of PPE ratio on system performance and efficiency for focal-plane SIMD image processing architectures by comparing throughput, area efficiency, and energy efficiency for a range of common application kernels using architectural and workload simulation. While the impact of grain size is affected by the mix of executed instructions within an application program, the most efficient PPE ratio often does not occur at PE grain size extremes (i.e., one pixel per processor or one processor per image). In this study, a set of four image processing application tasks is implemented on eight different SIMD configurations. Each configuration has a different PPE ratio and a different amount of local memory. Cycle accurate simulation and analytical technology modeling allows assessment of execution performance, area efficiency, and energy efficiency for each configuration. Results show the highest area and energy efficiency are achieved at PPE ratios between 16 and 256. Using these evaluation techniques (application grain size retargeting combined with area and energy technology modeling), a new class of efficient, embedded SIMD architectures for image processing can be designed.
Array control for high-performance SIMD systems
2004, Journal of Parallel and Distributed Computing
Although arrays of SIMD processing elements can be built with very high operating frequencies, problems exist in keeping the array busy. The inherent mismatch between host and array makes it difficult to maintain high array utilization: either the rate of instruction issue is very low or PE data locality is compromised, having the same effect. Our solution is based on an array control unit (ACU) design that expands macroinstructions in two stages, first by data tile and then into microinstructions. The expansion itself solves the issue problem; decoupling the expansion modalities maintains data locality. Several issues involving host/ACU interaction need to be resolved to effect this solution. We present experimental results showing that our approach delivers substantial improvement in memory hierarchy performance: a cache of only one-fourth the size is sufficient to achieve the same performance as previous approaches. We also describe our implementations which demonstrate that achieving gigaherz operating frequencies with current technologies is plausible.
Linear array processors with multiple access modes memory for real-time image processing
2003, Real-Time Imaging
This paper presents the linear array processors with multiple access modes memory system (LAPMAMM), an efficient mono-dimensional parallel architecture for real-time image processing. This architecture is composed of n processors and n² memory modules. These memory modules have multiple access modes: RAM, FIFO, normal CAM and interactive CAM modes. They are associated to a linear array of VLIW processors which are interconnected using a simple tree network that ensures an O(log(n)) data propagation time. The practical working of the architecture is explained using the example of a labeling algorithm developed for LAPMAMM. A hardware simulation of a LAPMAMM prototype has been carried out to test its performance in low and intermediate level image processing. The simulation results of the VHDL model are presented.
Parallel image processing with one-dimensional DSP arrays
2000, Future Generation Computer Systems
This paper describes a study of parallel image processing algorithms implemented on a one-dimensional DSP array. DSPs are developed for computationally intensive signal processing operations. Recently introduced parallel DSPs can be used for all levels of image processing operations and they provide easy development of a parallel system. In addition, due to the computing power delivered by these processors, we can employ coarse grain parallelism instead of the traditional fine-grain parallelism. Modularity, expandability and easy programming are other advantages of parallel DSPs. In this paper, parallel implementation of some selected image processing algorithms is described and performance results are presented.
A system for evaluating performance and cost of SIMD array designs
2000, Journal of Parallel and Distributed Computing
SIMD arrays are likely to become increasingly important as coprocessors in domain specific systems as architects continue to leverage RAM technology in their design. The problem this work addresses is the efficient evaluation of SIMD arrays with respect to complex applications while accounting for operating frequency and chip area. The underlying issues include the size of the architecture space, the lack of portability of the test programs, and the inherent complexity of simulating up to hundreds of thousands of processing elements. The overall method we use is to combine architecture level and Electronic Design Automation (EDA) level modeling by using an EDA-based tool to calibrate architectural simulations. The resulting system retains much of the high throughput of the architecture level simulator but it also has accuracy similar to that of an early pass EDA synthesis and circuit simulation. The particular problem of computational cost of the architectural level simulation is addressed with a novel approach to trace-based simulation (we call it trace compilation), which we find to be one to two orders of magnitude faster than instruction level simulation while still retaining much of the accuracy of the model. Furthermore, traces must be generated for only a small fraction of the possible parameter combinations. Using trace compilation also addresses program portability by allowing the user to code in a single data parallel language with a single compiler, regardless of the target architecture. We have used our system to evaluate thousands of potential SIMD array designs with respect to real applications and present some sample results.
On the Evolution of Parallel Computers Dedicated to Image Processing through Examples of Some French Computers
1997, Digital Signal Processing: A Review Journal
This paper proposes an overview of the evolution of the architecture of some parallel computers dedicated to image processing. A formal definition of a dedicated computer is proposed and some requirements of different levels of image processing/vision in terms of their algorithmic structures are pointed out. Examples of few French computers show hardware structures useful for efficient (real-time and on-board) implementations of image processing/vision tasks.

View all citing articles on Scopus

^☆: This work was supported in part by the Defense Advanced Research Projects Agency under Contract DACA76-86-C-0015, monitored by the U.S. Army Engineer Topographic Laboratories.

View full text

Invited paperThe DARPA image understanding benchmark for parallel computers☆

Abstract

Performance measurement instrumentation for multiprocessor computers

Parallel architectures and parallel algorithms for integrated vision systems

How not to benchmark image processors

The Connection Machine

Warp: A programmable systolic array processor

ASP: A cost-effective parallel microcomputer

IEEE Micro.

Benchmark results: The Abingdon Cross

Invited paper
The DARPA image understanding benchmark for parallel computers☆