工程工作站外文翻译@中英文翻译@外文文献翻译

附件 1：外文资料翻译译文 工程工作站 就原始性能而言 ,工程工作站大体上介于 PC 机和大的小型机之间 ；尽管随着 PC机和工作站两者功能的不断增强 ,这三者之间上的差别越来越难以分清了。但是，工程工作站不论同 PC 机，或是同传统的分时共享技术（或称小型机技术）相比确实有几个优点。 跟 PC 机相比，工作站通常具有更多的功能强的 CPU，而且能够支持更多的主存，尽管 PC 机在功能上同低档工作站有重叠现象。同 PC 机不同的是，工作站能够提供多用户，多任务操作系统，这已成为它的一种标准特点。 OS/2 和 UNIX 可用于 PC 机，尤其是以 Intel80386 为基础的 PC 机。然而， PC 机用得最多的操作系统仍是 MSDOS。 多任务系统同单任务系统相比有几个优点。首先，用户可同时运行多道程序，因此对于应用程序是透明的。虽然 PC 机的台式附件和常驻 RAM 程序可给用户提供某种原始的多任务功能，足以运行后台打印假脱机程序以及诸如此类的程序。但是，他们对应用程序可能是不透明的，而且不能提供像过程间通信和支持多个并行用户这样的重要特点。 对于当今的工程应用来说，也许更为重要的是 PC 机上缺少大容量的物理内存和虚拟内存。对于大型应用程序而言，虚拟存储器是很重要的 ，因为数据组太长，这种大型应用程序简直不能全部在物理存储器内运行。要是没有虚拟内存的话，像编辑大型文件之类的简单任务都会慢的令人头疼，甚至不可能完成。加上，许多应用程序更加复杂，因为它们必须缓冲数据或采用覆盖方式将应用程序的不同部分分页从物理内存中调进调出。 最后，大多数工作站的用户接口要比大多数 PC 机的用户接口高级一个明显的例外情形是 Macintosh 苹果机上的用户接口。计算机的用户接口。计算机的用户接口和连接它的可编程接口决定了应用程序接口的高级程度。强有力的开发手段可让程序员创建直观的用户接口。 虽然 工作站比 PC 机功能强，但跟现代小型机例如数字设备公司（ DEC） VAX8000 系列的小型机相比，情况通常就不是那样了。可是，工作站比起小型机来确有一些优点。 比起小型机来，工程工作站通常可为每个用户提供更多的 CPU 功能，更优越的价格 性能比以及更可预测的响应时间，因为 CPU 通常是为单个用户服务的。工作站的用户接口，常常像它胜过多数 PC 机的用户接口一样，也胜过小型机。不过，随着 X 窗口系统以及支持 X 协议的廉价终端设备的出现，这种优势正在消失。 网络系统的问题 虽然工作站给工程师带来新发现的计算机功能和显示能力， 但并非一切都是完美的。在拥有不少工作站的场所，从用户和系统管理人员的角度看来，问题是显而易见的。工作站操作系统的复杂性接近或超过了仅在几年前小型计算机操作系统的。在几乎没有软件可用来使下述重要任务例如磁盘支持，软件安装和更换，网络管理或帐户管理等实现自动化的情况下，要对这样一种系统进行管理可能前景暗淡。 分布式文件系统能对当地网络上的许多计算机产生从未预料到的依赖性。如果文件服务盘出现故障，许多计算机机器用户都可能受到影响。如果在不同计算机的过程间用文件进行通信的话，隐含目录也可能给两台计算机间生成全局文件 系统造成矛盾。 除了管理分布式系统根深蒂固的问题之外，还有获取软件许可证的费用过高的问题。由于工作站的大多数软件仍要像用在小型机和主机上那样获取许可证，对许多机构来说，这笔费用已经高到难以承受的程度。一些机构已开始提出对大量计算机的管理问题以及软件许可证的问题，但当前的解决方法 还不多。 工作站的未来发展方向 在短暂的发展过程中，工作站技术已经表明它的发展速度可以比其他任何现用计算机技术要快。在不远的将来，这一趋势大概不会改变。预计在将来的一段时间内， CPU 的运行速度会继续大大提高。重要的工作站厂商都在 RISC（缩简指令组计算机）技术上投入了巨额的资金，以便降低处理器芯片的复杂性并缩短其循环周期。大多数 RISC 结构格式都设计成也能用于新的制造工艺例如射极耦合逻辑电路和砷化镓制造技术，这就有希望在将来制造出比现在流行的速度更高得多的芯片来。 正如在有关硬件的那一节中提到的，浮点计算速度的提高一般地讲并没有能够同步赶上 CPU 速度的提高。然而，下一代浮点处理器可能改变这种不合拍的情况。最近一些厂商已经研制出原性能达到浮点运算速度每秒 1000 到 3000 万次的芯片。 开发应用者要做出的最困难决定之一是支持哪些工作站。从 传统上看来，大多数应用的最不可移植部分一直是用户接口，因为每一个厂商都有自己专有的窗口管理系统。随着 X 窗口系统和 Sun 的 NeWS 系统的日益普及，这种情形正在发生变化。这两个系统都是以网络为基础的窗口系统。 因此，能够在合适的计算服务盘上运行计算密集的应用程序。有几种用户接口工具包适用于 X 系统，其中有许多都是不受专利限制的。所有的迹象都表明，应用软件厂商的工作更容易，而且，归根结底，提供给最终用户的应用软件会更好。 对分布式计算提供的支持比对分布式文件系统提供的支持落后得多。随着市面上出售的专用计算机越来越多 ，工程师和计算机科学家将需要开发利用两种或两种以上的计算机的专门能力来求解日益复杂的问题。工作站将成为集成这些异质功能计算机的焦点。有些厂商已经在其系统中增加了远程过程调用程序库。有许多问题，例如平衡装入，容错，应用分区，动态服务盘定位，客户登记以及数据加密等问题，目前还只能部分解决。 附件 2：外文原文（复印件） The Engineering Workstation Workstations fit somewhere between PCs and large minicomputer in terms of raw performance； although as both PCs and workstations get more powerful, the performance distinctions between the there groups are more difficult to make. Engineering Workstations do have several advantages over both PCs and traditional time-sharing or minicomputer technology, however. Workstation generally have more powerful CPUs and can support more main memory than PCs, although the latter overlap with low-end workstations in power. Workstations, unlike PCs, offer multitasking, multitasking operating systems as a standard feature. OS/2 and UNIX are available for PCs, especially for those based on the Intel 80386. However, the dominant operating system for PCs is still MS-DOS. Multitasking systems have several advantages over single-tasking systems. First, the user can run more than a single program at a time , transparent to the application. Desk accessories and RAM-resident programs on PCs give the user some primitive multitasking capability, sufficient to run background print spoolers and the like . However, they may not be transparent to application program, and they do not offer significant features like interprocess communication and support for multiple, simultaneous users. Perhaps more important for todays engineering applications is the lack of large physical and virtual memories on PCs. Virtual memory is important for large applications, many of which simply cannot be run entirely in physical memory because the data sets are too large, Without virtual memory, a simple task like editing a large file becomes tedious or even impossible to perform. In addition, many applications are more complex because they must buffer data or use overlays to page different parts of the application in and out of physical memory. Finally, the user interface of most workstations is more sophisticated than that on most PCs. The one notable exception is that of the Apple Macintosh. The computers user interface and the programming interface to it determine how sophisticated the applications program interfaces can be. Powerful development tools allow programmers to create intuitive user interfaces for complex applications. While workstations are typically more powerful than PCs, that is generally not the case when they are compared with modern minicomputers, such as those of the VAX-8oooseries by Digital Equipment Corporation(DEC). They do have several advantages over minicomputers, however. Engineering workstations generally provide more CPU power per user, a better price-performance ratio, and more predictable response times than minicomputers, because the CPU is usually devoted to a single user. Workstations often have the same user-interface advantages over minicomputers that they have over most PCs. With the advent of the X Window System and inexpensive terminals that support the X protocol, this advantage is disappearing. Problems with Networked Systems Although workstations are giving engineers new-found computer power and display capabilities, all is not perfect. The problems become readily apparent to users and system administrators at sites with more than a handful of workstations. The Complexity of the operating systems on workstations approaches or exceeds that of minicomputer operating systems of just a few years ago. With little or no software available to automate critical tasks such as disk backup, software installation and update, network management, or account management, the administration of such a system can become a nightmare. Distributed file systems can create unexpected dependencies on many machines on a local area network. When a file server goes down, many machines(and their user)may be affected. When files are used to communicate between processes on different machines, directory caching may create inconsistencies in the appearance of the global file system on the two machines as well. On top of the problems inherent in managing distributed systems are software licensing costs, which for many organizations have skyrocketed because most software packages for workstations are still licensed as if they were for minicomputer and mainframes. Some organizations are beginning to address the problems of administering large number of machines and software licensing but few solutions are today. Future Directions for Workstations During its brief history, workstation technology has shown that it can move faster than any other computer technology available. This will probably remain true for the future. Significant increases in CPU speeds are expected to continue for some time in the future. The major workstation vendors have all made a substantial investment in RISC(reduced-instruction-set computer)technology, thereby reducing the complexity and cycle time of processor Chips, Most RISC architectures are designed to be scalable to new fabrication technologies as well, such as emitter-coupled logic and gallium arsenide, promising even higher speed chips in the future than those prevalent today. As mentioned in the section on hardware, the increase in CPU speeds has not, in general, been matched by a parallel increase in floating-point computation speeds. However, the next generation of floating-point processors may performances. Recent developments by several vendors have led to chips with raw performances of between 10 and 30 million floating-point operations per second. One of the most difficult decisions for application developers to make is which workstations to support. Traditionally, the most nonportable part of most applications has been the user interface, since each vendor had a proprietary window-management system. This is starting to change as the X Window System and Suns NeWS gain in popularity. Both are network-based window system, so computer-intensive applications can be run on the appropriate computational se