The two sides started to contact from 1995. 1996, the top management of DEC agreed to Compaq's acquisition proposal, only three years before 1998 was finally delisted. According to the company report of 1989, DEC has about130,000 employees, and its market value exceeds1400 million US dollars. It is the largest computer manufacturing company in the United States after IBM, and also has excellent R&D departments and large-scale production plants. People have reason to ask: why did such a huge company collapse in an instant? The reasons are naturally complicated. Let's describe the seven major faults of DEC.
One of the failures is backward thinking.
For a long time, Kenneth Olsen, the founder and CEO of DEC, has a great deviation in the sales concept. As an engineer, he doesn't know much about sales. He once said that "an excellent product can be sold by itself". This sentence profoundly reflects DEC's consistent attitude towards product promotion and market development. He also believes that "it is impossible for every household to have a computer."
Perhaps these ideas may be correct in the past, because the output of computers at that time was very small and the price was very high. However, at the end of the 20th century, millions of computers were sold every year, and computer outlets were all over the United States, so consumers could easily buy their own computers. At the same time, the consumers who buy computers have also changed, and the main consumers have turned from technicians/professionals to ordinary people. These consumers have no professional background knowledge, and even can't tell the spelling of "transistor" and "resistor" (note: the spelling of transistor and resistor is similar, that is, transistor and resistor). So we don't pay enough attention to market promotion and are insensitive to market changes. We still sell our products to professionals and ordinary consumers, and we get twice the result with half the effort. This is a failure.
The second failure is to miss the opportunity.
199 1 February, DEC launched the EV4 processor. At the same time, Apple engineers are looking for processors with better performance for the company's products, and the launch of EV4 left a deep impression on them. Therefore, Apple CEO john sculley met with Kenneth Olson in June of the same year, hoping to use DEC's new processor in future Apple computers. However, Olsen thought that the time for EV4 to be put on the market was not mature, and the potential of VAX architecture had not been fully tapped, so he rejected Apple's request.
A few months later, Apple introduced a MAC computer based on PowerPC developed by IBM and Motorola. 1On April 28th, 997, Willian Dehmel, a DEC engineer who had participated in the development of VAX and Alpha, pointed out in an interview with Businessweek that the company's top management didn't want to bet on Apple in the future. Therefore, indecision and missing the opportunity are the second failures.
The third reason for failure is that insufficient attention is paid to the production of parts.
DEC produces all accessories and peripherals related to the Alpha processor, but the motherboard developed for desktop computers does not support SMP. At that time, almost all companies using Alpha processors used multiprocessor systems, so the desktop models launched by DEC were not competitive. However, these motherboards are very good in electrical design. Because the layout circuits of these motherboards are open, many companies have been attracted to imitate and transform, and a large number of cloned versions have been produced.
Meanwhile, only one company has developed its own motherboard desktop workstation for the desktop market. Although more and more companies sell their motherboards to users who use Alpha processors, DEC doesn't care about these. They think it is most important to sell their workstations or servers first, and the market for some fragmentary computer accessories is not important. So some people commented that DEC occupied the market, but did not conquer it.
The fourth reason for failure is the high price.
DEC has never thought about reducing the price of their products (processors, chipsets, motherboards) to the price that most potential consumers can afford. For example, at the beginning of 1995, the price of 266MHz and 300MHz EV5 was as high as $2,052 and $2,937 per thousand units. Of course, this is still the wholesale price. If the actual retail price is considered, the pricing of EV5 is more than twice as high as that of RISC design competitors in the same period.
Although DEC has released a cheap product Alcor, the price of this motherboard is $295 per 5000 chips, which is far lower than the price of the processor, but it bundles the processor (EB 164, 1MB L2 cache) with 16MB main memory. Because the main memory is too small, it is also stretched for the program at that time. The price of this mainframe is 7500 dollars.
The fifth reason for failure is to collect expensive patent fees;
Although DEC advocates the concept of open architecture, the Alpha project has been open from the beginning, but all the research and development work has always been done by DEC's own engineers, only in the production stage was entrusted to Mitsubishi. Since I made it myself, I only disclosed the general product framework, but the most important hardware design part is confidential and needs to pay a high patent fee to get it.
Although DEC has extended an olive branch to Intel, Motorola, NEC, Texas Instruments and other companies since EV4, the patent fee is too high. These companies did not accept DEC's "kindness" and developed their own products in a blink of an eye. Shortsightedness and greed are the fifth reason for failure.
No matter how good the computer is, it cannot be separated from the support of the operating system, otherwise it is just an expensive heater. Therefore, DEC attaches great importance to the operating system. Windows NT, Digital UNIX and OpenVMS have all become the choice objects of the company's top management, but ...
The sixth reason is to choose NT architecture as the preferred operating system.
The first thing to know is that WINNT is designed for users, not programmers. There is no integrated software development tool in the operating system, so the running software needs to be precompiled. At that time, there were quite a few softwares developed on the market based on Alpha and i386, which could not run on the other platform and needed to be reformed first.
FX is only released on 1996! The 32 developed by Anton Chernoff's team can simulate x86 well and convert it into Alpha, but the result is about 40% performance loss. Any driver and FX! There is nothing we can do about it, and everyone is puzzled about it. Then a few programmers found that WINNT is a 32-bit operating system. Even if it can work on the 64-bit Alpha platform, it is difficult to give full play to the potential of the 64-bit architecture. In fact, NT should not be the preferred operating system of Alpha architecture at all, but at most it exists as an alternative.
The seventh reason is that the operating system is expensive.
In fact, there are two operating systems on the market that are very suitable for Alpha architecture, namely OpenVMS and Digital UNIX. However, these two commercial operating systems are overpriced, resulting in their low market share and no open source code. Moreover, the support of these two operating systems for peripherals is not as rich as NT, and the problems of operating systems have been plaguing the popularity of Alpha, and the selected operating systems are not satisfactory.
The eighth reason for the failure is that the open source operating system is not used.
Although all commercial operating systems have some problems and have not been successfully upgraded to the Alpha platform, DEC has never supported free and open source operating systems. As early as 1995, NetBSD was transplanted to the Alpha platform, followed by Linux, OpenBSD and FreeBSD. The performance of these systems is not worse than that of Digital UNIX and OpenVMS, and their compatibility with hardware is better than that of WINNT, and they can also provide a large number of open source programs for users to use. Therefore, these systems are popular on the Alpha platform.
Of course, we can continue to make a long list of DEC's strategic mistakes, including their insufficient attention to the changes in the mainstream market and PC market. But these are not directly related to the Alpha architecture itself, so we skipped it. In short, DEC did put a lot of effort into Alpha, but after the product was launched, DEC only thought about how to exchange money for Alpha architecture, not how to promote it.
abstract
In the late 1980s and early 1990s, DEC made many bad decisions, including 1992 a series of reorganizations made by Robert Palmer, the newly appointed chairman of the board of directors. Parr's default is that the existing matrix model (dividing different departments according to different functions and coordinating each project by multiple departments) is not suitable for the company, and it needs to return to the traditional vertical model (from top to bottom, everyone's posts and tasks are allocated very specifically).
From 19 1 to 1994, DEC lost more than 4 billion dollars, and only 1993-94 lost 2 billion yuan. In order to make up for the huge fiscal deficit, Palmer plans to sell all other separable parts of DEC. So a global sales began. 1In July, 1994, the storage department of DEC hard disk was sold to Quantum for 400 million dollars. Soon after, the R&D department of database software was sold to Oracle Bone Inscriptions for $6,543.8 billion. 1In June, 1997, DEC sold the network products division to Cabletron again for $430 million.
1997, DEC, who was seriously ill, took Intel to court, accusing it of using 65,438+00 patents of Alpha processor in Pentium, Pentium Pro and Pentium II processors. 1September 1997, the two sides went to court and refused to give way to each other. However, in the same year1October 27th 10, both parties reached an out-of-court understanding. DEC granted Intel the production rights of all hardware (except Alpha) and agreed to support its IA-64 architecture development plan in the future. Intel also bought DEC's manufacturing plant in Hudson and design centers in Israel and Texas for $625 million, and agreed to produce DEC's Alpha processor in the future, and obtained the right to use all DEC's patents for 65,438+00 years.
Finally, I have to talk about the whereabouts of talented engineers who have worked in DEC for many years. Derek Meyer joined AMD to design K7; James Keller also went to AMD, but he was an architect of K8. Daniel Leibholz went to Sun to develop ultrasparc v, and Intel was far from as lucky as expected: although DEC benefited a lot at the last minute, Stronarm architecture could only watch it die. Because the chief architect who originally designed Stronarm-110-has no one-Daniel Dobberpuhl, Richard Witek, Gregory Hoeppner and Liam Madden. Richard Sites, the first person who proposed the Alpha architecture, has never found a decent job.
By1May 1998, 18, Compaq, with only 32,000 employees, acquired DEC, with 38,000 employees, ending the last scene of DEC on the historical stage.