the competition in the market is cruel, and the survival of the fittest is the inevitable development direction. At present, in the desktop CPU market, Intel, AMD and VIA have formed a tripartite confrontation. With Transmeta's involvement in the notebook CPU market, and Intel and AMD also ambitiously extended their reach to the high-end server market, the competition between the server and notebook CPU market became more intense. Server processors are almost all RISC (Reduced Instruction Set) architectures, which are used in high-end workstations or servers. According to the report of IDC, a market analysis organization, Sun ranked first, IBM ranked second and Compaq ranked third in the US server market in 2. With the intervention of Intel and AMD in this profitable market, the high-end server market has formed a new pattern of letting a hundred flowers blossom and a hundred schools of thought contend. Let's get to know these real server CPUs. 1. Compaq Digital Alpha Alpha processor was originally a product of DEC company, and was later classified by Compaq, and Compaq merged with Digital. Alpha first appeared in the market in 1992, leading other RISC processor manufacturers for two or three years. When Alpha was introduced, the personal computer at that time was shifting from the 386 era to the 486. The biggest feature of Alpha is actually its leading position in clock speed. For example, the 3MHz Alpha 21164 introduced in 1995 was the CPU with the fastest clock speed at that time. Alpha 21164 is the second generation processor of Alpha. It has two versions, one is the original 21164, and the other is the 21164PC with partial Cache removed, hoping to attack the lower-end market. In addition, DEC published the third generation of Alpha 21264 in early 1998. Alpha can run in UNIX, OpenVMS and Windows NT, which is the difference between Alpha and several other RISC processors. Compaq Digital Alpha will develop Alpha 21364 and EV8 processors in the future. Second, SGI MIPS MIPS processor is the originator of RISC architecture. After the acquisition of MIPS, SGI divided MIPS processor into two market attacks: on the one hand, it emphasized high performance and continued to develop in the high-end direction; On the other hand, it focuses on high output and turns to the field of embedded and consumer electronic products with huge market. In fact, SGI is only responsible for the R&D and design of processors, and the production and sales of products are authorized by other semiconductor manufacturers. At present, the highest-end MIPS processor is R1, which is used in SGI's full range of products, including O2 workstation with single processor and Origin high-end server with up to 128 processors. SGI specializes in graphics operation, and has strong system I/O and memory bus. MIPS processor does not emphasize the clock frequency alone, but focuses on the improvement of overall performance. SGI MIPS processor mainly runs on its own 64-bit operating system IRIX (operating system of the same family as UNIX); In addition, MIPS is also used as the processor of some handheld computers, so it can also run on Windows CE. In order to continue to gain a foothold in the high-end market, SGI has to go all out to develop more advanced MIPS processors, including R12 and R14. Third, SUN SPARC Sun is the first manufacturer in the world to give RISC architecture to mass production. In order to promote SPARC to become the industry standard and increase the global supply sources, SUN also authorized several semiconductor factories to produce their own SPARC chips. SPARC has superior performance and high price, and is recognized as outstanding on UNIX. The early RISC processor was also 32-bit, and it was not until the birth of Alpha more than six years ago that RISC was pushed to 64-bit. As far as SUN's SPARC is concerned, its 64-bit processor is SPARC-v9 architecture in 1995, and the product is called Ultra SPARC. At present, the highest-end SPARC product is 64-bit Ultra SPARC III, which adopts the technology of Uptime Bus. The working frequencies of Ultra SPARC III are 9MHz, 75MHz and 6MHz. Compared with the previous UltraSPARC II, UltraSPARC III runs programs twice as fast. In recent years, Intel's attempt to enter the high-end market is obvious. Some manufacturers with RISC processors have gradually developed in the direction of Intel's IA-64, while SUN still insists on developing its own Ultra SPARC processor, which has become a mainstay to prevent Intel from invading. Sun will also launch a 1.2GHz Ultra Space 4 processor based on MAJC architecture this year, which will be Sun's hope in the high-end server market competition. Fourth, HP PA-RISC HP also has its own RISC processor, called PA-RISC (precision architecture RISC). PA-RISC appeared in 1986, and HP was also the first computer manufacturer in the world to transfer its system architecture from CISC to RISC, and then HP became the number one in the market. At present, the version of PA-RISC processor is PA-82, which is mainly used in HP enterprise servers (such as the highest-end HP9 series). After PA-82, HP will also introduce PA-85 and PA-87 processors. HP PA-RISC regards Alpha as the main rival of performance, for example, the rival of PA-85 locks Alpha-21264. Hewlett-Packard adopts a two-way parallel strategy in products. In order that both chips can be used in computers, it develops PA-RISC 87 and IA-64 processor with Intel***. IA-64 combines x86 and RISC architecture. x86 comes from Intel's own architecture, and the RISC part is HP's PA-RISC architecture. 5. IBM PowerPC Although the term RISC was first created and used by Professor Patterson of Berkeley University in the early 198s, it became a general term later. In fact, the real pioneer of RISC is the blue giant IBM, which quietly launched its experimental plan in the 197s. IBM started an 81 project in 1975, hoping to design a new computer architecture. However, the 81 project did not succeed in product launch. However, another development line of IBM parallel to the 81 project became the America project in the mid-198s, and this project successfully developed a new architecture product, which was the Power architecture that appeared in 199, and IBM built the RS6 processor and workstation products based on it. In 1991, IBM launched the second generation Power architecture, and formed a "PowerPC" alliance with Motorola and Apple*** to develop a new PowerPC processor architecture. This PowerPC architecture is based on IBM's Power architecture. In 1992, IBM released the first PowerPC processor PowerPC-61, which is a 32-bit RISC architecture processor. Apple immediately changed its architecture and adopted PowerPC-61 as the core processor of a new generation of Mac computers. Subsequently, IBM successively published PowerPC processors of series 63 and 64, and the latest versions are PowerPC 75(G3 is actually the trademark of PowerPC 75) and 74 series. In addition, a new generation of 64-bit Power 4(G4) processor has also been launched. VI. Intel Itanium Intel Corporation announced the software and hardware development status of Intel Itanium (An Teng) with IA-64 structure on March 29th. Itanium's earliest chip (so-called First Silicon) was processed in August 1999, and the start-up demonstration of Windows and Linux was held at IDF at the end of August, but its progress could not be called smooth. After several delays in shipment, Intel finally decided to launch servers and workstations equipped with Itanium before June 3, 21. The key for Intel Itanium processors based on servers and workstations to outperform competitors' RISC processors is to use the innovative merging function epic (Explicitly Parallel Instruction Computing). The architecture of IA-64 is based on EPIC (Explicit Parallel Instruction Computer). The performance of EPIC is better than RISC and CISC, and it is comparable to a large processing source with an intelligent editor, which clearly informs the processor of parallel instructions. The test results show that Itanium has exceeded the speed of a single RISC processor. Intel said that when Itanium was officially put into production, its working calculation frequency could reach 2 billion per second, which was ten times faster than that of Sun Ultra SPARC III. In addition to Itanium processor, a new generation of IA-64 architecture McKinley processor will also be unveiled. Seven, AMD SledgeHammer processor due to the strategic alliance between Intel and IBM, it is stuck in the 64-bit processor and high-end server market in advance. This strategy forces AMD to work harder. AMD is preparing to release the next generation of x86-64 architecture SledgeHammer processor. AMD's 64-bit processor will target servers and high-end applications. AMD is going to release two kinds of SledgeHammer processors: ClawHammer chip for 1~2 CPU servers and SledgeHammer chip for 3~4 CPU servers. AMD is going to use them to compete with Intel's Itanium. AMD plans to launch industrial samples of ClawHammer and SledgeHammer processors in the first quarter of 22, and will put them into mass production in the second quarter of the same year. If the development and production of these two chips of AMD are relatively smooth, AMD will be able to compete with Intel in all markets, from desktop computers to notebook computers to enterprise servers.