The last few years have seen considerable research activity in performance prediction for parallel programs. For example, V. Sarkar estimates the runtime of parallel programs at compile time using execution profile information. This approach targets single-assignment languages, which simplifies analysis significantly. V. Balasundaram et al introduced a performance estimator to evaluate different data partitioning strategies by using premeasured kernel codes. They achieved good results for the loosely synchronous programming model. K.Y. Wang characterizes parallel programs by an analytical performance model that uses symbolic analysis to obtain runtime information.

NOTES:

1. tedious (adj.) — скучный, утомительный.

2. kernel (code) (n.) — код ядер.

3. to validate (v.) — делать достоверным.

II. Make up 5 key questions. III. Sum up the main information (see p. 5).

TEXT № 2

Dynamic Behavior or Parallel Programs

Although petri nets," queueingnetworks, and Markov chains can be valuable in understanding the dynamic behavior of parallel programs, the associated analysis costs in terms of both runtime and memory requirements prohibit their use in compilers. F.

Hertleb and V. Mertsiotakis derive upper and lower execution time bounds for parallel programs with the aim of tuning implementation and mapping alternatives. In their scheme, parallel programs are modeled as a stochastic graph, and random variables are used to describe the runtime behavior of a specific processor. Most of the above approaches have serious drawbacks. Some, for example, apply only to programs that can be modeled as multiple nested loops with constant loop bounds, no procedure calls, and no conditional statements. Others are too expensive in terms of runtime and memory requirements. Purely theoretic approaches are difficult to validate against actual program behavior on real architectures.

Unit 11

I. Read and translate text №1 in written form using a dictionary if necessary.

ТЕХТ № 1

Chapter 1: Introduction to Computer Graphics Systems.

This book concerns itself with recent hardware developments and the impact these developments have on the discipline of computer-generated graphics. Here then, is a brief introduction to graphics system architectures.

Computer graphics had its start with X-Y plotters. The technology was soon extended to include calligraphic (linedrawing) CRT systems based on refresh-vector hardware and, at a later date, direct-view storage tubes followed by higher performance refresh-vector systems.

Meanwhile, a separate image processing technology had been following its own evolutionary path. Instead of dealing with lines and points randomly positioned anywhere on the display surface, image processing was based on a rectangular array of picture elements or pixels. Digital information defining the state of each pixel was stored in a random-access memory and was used to generate a television-type, raster-scan CRT display in monochrome or full color.

The division between vector graphic (also known as calligraphic) computer graphics and raster-scan image processing has now been bridged by the raster display technology. Low-cost memories have made it economically feasible to assemble high-resolution, fine-detail raster images that all but eliminate the objectionable stair stepping of vector graphic lines when displayed on earlier raster-CRT screens. The ability to fill areas with solid color (and shading) makes raster-CRT screens useful in certain applications. The result is raster scan graphics, which combines the full-color, pixel-by-pixel control potentials of image processing and the line-drawing capabilities of vector graphics in a single display system.

Unfortunately, the move from calligraphic to raster displays has brought new problems. For calligraphic, only the endpoint coordinates of lines needed to be stored, so memory requirements were held to a minimum. Refresh-vector writing speed made it possible to animate the display and to create interactive systems that allowed the operator to control the display in real-time through such input devices as light pens, joysticks, and digitizer tablets. Again, however, only the endpoints needed to be recalculated with each refresh cycle, minimizing the need for high-speed computational capabilities,

However, in raster-scan graphics, each pixel must be computed explicitly. To generate a raster of image, millions of pixels must be computed. The proper value at each pixel is a function of the database (the simulated environment), the viewing position and orientation of the simulated viewer, and the location(s) of the light source(s) in the simulated environment. A long-standing goal of researchers in computer graphics systems has been the development of real-time three-dimensional modeling systems. These systems, which produce a realistic image of a simulated three dimensional environment. have a wide variety of potential uses, from flight simulators for pilot training to interactive computer-aided design (CAD) systems. The most sophisticated of these systems produce, in real-time, images of startling reality.

II. Answer the questions:

1. Say a few words about Computer Graphics early history -

2. What is the raster display technology ?

3. What are the new problems brought by the move from calligraphic to raster displays?

III. Make up a close summary of text №!( see p.5 ).

IV. Read and translate text №2 orally without a dictionary.

TEXT № 2

Functional Model of Computer Graphics Systems

A graphics system can be regarded as an implementation of a sequence of processes that transforms pictorial information into a perceptible form and presents it as an image on a graphics display device. Along the sequence a number of operations may be performed. This sequence of operations can be organized and abstracted into a functional model of a computer graphics system [CARL80, FOLE82].

The functional model consists of a sequence of logical processors (also known as the display pipeline) operating on the representations of the scene (see Figure I.I). A logical processor in the functional model corresponds to one or more physical processors, and two logical processors in the model may share a physical processor. Similarly, representations may exist in one or more different memories.

NOTES:

1. feasible (adj,) — допустимый, возможный.

2. explicity (n.) — формальный, четкий.

3. perceptible (adj.) — заметный, ощутимый-

Chapter 2

In this chapter you will gain practice in reading, understanding and oral translating the computer language in context. Before starting to work on this chapter read the recommendations and advice on pp 4-5 very attentively.

Unit 1.

I. Read and translate the text orally at sight.

Optical Computers.

"You can stilt say it's impractical, but you can't say it's impossible anymore." Alan Huang, head of a year-old research effort at AT&T Bell Laboratories, is talking about his passion: optical computers. The idea of building blindingly fast machines that compute with light pulses instead of electrical signals has been controversial for two decades — and the prevailing view is still one of skepticism.

Codes and Bombs The reason for the excitement is that optical computers promise speeds thousands, even millions of times faster than today's speediest supercomputers. The most optimistic estimate for the switching time of familiar silicon transistors is $0 picoseconds, or close to a trillion switches a second. But a conventional supercomputer will probably never run that fast. Since electrons travel through copper wires at only a fraction of the speed of light, the system has to wait for signals to arrive from other chips. Bell Labs thinks it already has an idea of how to make optical transistors with switching times as fast as 50 femtoseconds, or quadrillionths of a second, and 1 femtosecond is the ultimate goal. And light rips through optical fibers at sirsost the speed of light.

At such fantastic speeds, computers could open the doors to new scientific insights that would give users a commanding edge in engineering everything from new materials to aircraft that would fly circles around existing planes. These computers could be crucial to cracking now-unbreakable codes and designing superior nuclear weapons —the purpose for which supercomputers were originally conceived. And they could be the cornerstone of President Reagan's Star Wars missile Optical computers, says Air Force Lieutenant Colonel David R. Audley, program manager for battle management systems at the Strategic Defense Initiative, "could revolutionize the way we think about computing, much as the semiconductor chip revolutionized electronics 30 years ago."

П. Try to summarize the main idea (see pp. 5-8 ).

NOTES:

1. Blind (ingly) (adj/adv) — слепой, слепо; (v) блокировать данные.

Unit 2

I. Read and translate the text orally at sight.

Systems are never deliberately designed to fail, but sometimes it seems as though they were. Instead of experiencing the anticipated boons of efficiency, costs savings, and new potential that the system was designed to offer, the organization loses hundreds of thousands of dollars invested in system development and hardware. One way to make this happen is to keep the users from understanding the system, and thus make them resist it.

Take the actual case of a new word-processing system installed at the headquarters of a Fortune-500 corporation. User training about the purpose of the system - - and the career futures of the users — was conducted primarily by rumor. When the system was introduced, secretaries feared that it would eventually put an end to their jobs. This was wrong: The system was meant to produce only a single change in their jobs — it automated their repetitive typing chores.

Some of the secretaries were chosen to learn to operate the new system and asked to report to the corporation's Word Processing Center The secretaries who stayed in the offices were unsure exactly what they were supposed to do. They opened and routed mail, answered the phones, filed documents, and scheduled appointments. But without typewriters — or typing — they suspected that it was time to update their resumes and look for new jobs; they felt sure that their present jobs would soon be eliminated. Accordingly, some secretaries found new jobs and left the corporation. Others actually smuggled in their own typewriters and began typing for their bosses as they had in the past.

As for the managers, there was no way they could regard the new system as an improvement. Instead of having an individual secretary, several managers now had to share one among them. And getting something typed required the cumbersome extra step of routing it to and from the Word Processing Center.

Was the system benefiting the secretaries who now worked full time in the Word Processing Center? Hardly: Without any administrative tasks to do, they felt they were in a slightly glorified typing pool. They hated having to type all day. Removed from direct interaction with managers, they felt they were also removed from the functioning of the organization.

In other words, the system not only failed to please everyone; it failed to please anyone. After six months of this all-around agony, the system was cropped The hardware was discarded; the Word Processing Center closed down, and those secretaries who remained were returned to their old jobs and old duties. All the time and money invested in designing the new system and trying to make it work were lost. Moreover, even returning to the old setup involved the expense of hiring and training new people to replace those who had left in despair,

II. Try to summarize the main idea ( see pp. 5-8).

NOTES:

1. deliberately (adv.) — намеренно.

2. anticipated (p.p.) - ожидаемые преимущества.

3. chore (n.) — повседневная работа.

4. routed (p.p.) (mail.) — маршрутизация сообщений E-mail

5. to smuggle (v.) — делать что-либо тайком (контрабандой)

6. cumbersome (adj.) — громоздкий, затруднительный.

7. to scrap (v.) — ломать, разламывать, превращать в обломки.

Unit 3

I. Read and translate the text orally at sight.