Current location - Music Encyclopedia - Chinese History - The history of arm company
The history of arm company
1995- the joint production base of fuji AMD semiconductor co., ltd (FASL) started construction.

1995-fab 25 completed.

1996-amd acquires NexGen.

1996-AMD started to build Fab 30 in Dresden.

1997-AMD launches AMD-K6 processor.

1998-AMD released the AMD Athlon processor (previously code-named K7) at the microprocessor forum.

1998-AMD and Motorola announced the establishment of a long-term cooperative partnership on the development of copper interconnection technology.

1999-AMD celebrates its 30th anniversary.

1999-AMD introduces AMD Athlon processor, which is the first seventh-generation processor in the industry to support Microsoft Windows computing.

2000-AMD announced the appointment of Hector Ruiz as president and CEO of the company.

2000-AMD Japan Branch celebrated its 25th anniversary.

2000-AMD's sales in the first quarter exceeded $654.38 billion for the first time, breaking the company's sales record.

2000-AMD's Dresden Fab 30 began to be supplied for the first time.

2001-AMD launches AMD Athlon XP processor.

2001-AMD introduces AMD Athlon MP dual processors for servers and workstations.

2002-AMD and UMC announced the establishment of a comprehensive partnership to jointly own and manage the 300 mm wafer manufacturing center in Singapore and jointly develop advanced processing technology and equipment.

2002-AMD acquired Alchemy Semiconductor and established Personal Connection Solutions Division.

2002-Hector Ruiz succeeds Jerry Sanders as CEO of AMD.

2002-AMD introduced the first flash memory device based on MirrorBit(TM) architecture.

2003-AMD introduced AMD Opteron(TM) processor for servers and workstations.

2003-AMD introduced AMD Athlon (TM)64 processor for desktop and notebook computers.

2003-AMD introduced AMD Athlon (TM)64FX processor, which enabled the system based on AMD Athlon (TM)64FX processor to provide theater-level computing performance.

198 1 year, AMD 287FPU, using Intel80287 kernel. The market positioning and performance of the product are basically the same as Intel80287. It is also the only FPU product produced by AMD so far, which is very rare.

■AMD 8080( 1974), 8085( 1976), 8086( 1978), 8088( 1979), 80186 (.

■AMD 386( 199 1 year) microprocessor with core code P9, which is divided into SX and DX, and is compatible with Intel80386SX and DX respectively. AMD 386DX and Intel 386DX are both 32-bit processors. The difference is that AMD 386SX is a complete 16-bit processor, while Intel 386SX is a quasi-32-bit processor (internal bus 32 bits, external 16 bits). The performance of AMD 386DX is similar to that of Intel80386DX, and it was one of the mainstream products at that time. AMD has also developed 386 DE and other embedded products based on 386 cores.

■AMD 486DX( 1993) microprocessor, core code P4, the first generation of 486 products designed and produced by AMD. Then other 486-level products were launched one after another. Common models are: 486DX2, core code P24;; ; 486DX4, core code P24C;; 486SX2, core code P23, etc. Other derivative models are 486DE and 486DXL2, which are relatively rare. The highest frequency of AMD 486 is 120MHz(DX4- 120), which is the first time that AMD has surpassed its powerful competitor Intel in frequency.

■AMD 5X86( 1995) microprocessor, core code X5, AMD's sharp weapon in the 486 market. In the late 486' s, TI (Texas Instruments) launched the TI486DX2-80 with high cost performance, which quickly occupied the low-end market, and Intel also launched the high-end Pentium series. In order to seize the market vacancy, AMD introduced 5x86 series CPU (almost at the same time as Cyrix 5x86). It is a 486-class product, with the highest frequency of-33 * 4, 133MHz, 0.35 micron manufacturing process, built-in 16KB first-level write-back cache, and its performance is directed at Pentium 75, with lower power consumption.

Product drawings before K6 era (12) ■AMD K5( 1997) microprocessor, 1997 released. Because of research and development problems, its listing time is much later than that of competitor Intel Pentium, and its performance is not very good. This unsuccessful product once made AMD lose a lot of market share. The performance of K5 is very general, and its integer operation ability is not as good as Cyrix x86, but slightly better than Pentium. Floating-point budgeting ability is far less than Pentium, but slightly better than Cyrix 6x86. On the whole, K5 is a product with average strength, and its low price at the initial stage of listing is more attractive to consumers than its performance. In addition, the production of the top-end K5-RP200 is very small, and it is not sold in Chinese mainland.

■AMD K6( 1997) processor is at the same level as Intel PentiumMMX. It is the representative work after AMD acquired NexGen and incorporated the advanced NexGen 686 technology at that time. It also contains MMX instruction set and 64KB L 1 cache, which is twice as big as Pentium MMX! Overall, K6 is a successful work, but in terms of performance, the floating-point computing ability is still lower than Pentium MMX.

■K6-2( 1998) series microprocessors used to be AMD's fist products, and now we call them classics. In order to beat competitor Intel, AMD K6-2 series microprocessors have been greatly improved on the basis of K6, the most important of which is the addition of "3DNow!" Command support. “3DNow!” Instruction is a breakthrough of X86 system, and the advantage of this technology is that it greatly enhances the 3D processing ability of computers and brings us truly excellent 3D.

K6 era (14). When you use the special "3DNow!" When optimizing the software, we can find out how much potential K6-2 has. Moreover, K6-2 is mostly frequency-unlocked, and it is easy to overclock due to the low calorific value brought by the 0.25 micron manufacturing process. That is, from K6-2, overclocking is no longer a proper term of Intel. At the same time, K6-2 also inherited the tradition of AMD. The price of the same frequency model is about 25% lower than that of Intel products, and the market sales are amazing. At the beginning of its launch, the K6-2 series used the name "K63D" ("3D" means "3DNow!" ), it was renamed as "K6-2" until it was officially listed. Because of this, K6 3D is mostly ES (a few official versions, after all, there is no mass production). K6 3D once had a non-standard 250MHz product, but it did not appear in the official K6-2 series. The lowest frequency of K6-2 is 200MHz and the highest frequency is 550MHz.

■AMD launched the K6-3( 1998) series microprocessor code-named "Sharptooth" in February, 1999, which is the last CPU supported by AMD in super architecture and CPGA package. K6-3 adopts 0.25 micron manufacturing process, integrates 256KB L2 cache (competitor Intel's new Celeron is 128KB), and runs at the main frequency speed of CPU. L2 on Socket 7 motherboard is automatically recognized as L3 by K6-3 at this time, which is undoubtedly very advantageous for high-frequency CPU, although K6-3' s floating-point operation is still unsatisfactory. For various reasons, it is hard to get a ticket after K6-3 is put on the market, and the price is not approachable, even after the emergence of the more advanced K6-3+.

Product drawings after K6 era (20) ■AMD introduced K8 architecture at 200 1 and 10. Although K8 and K7 use the same number of floating-point scheduling windows, the integer unit is extended from 18 of K7 to 24. In addition, AMD also improved the branch prediction unit of K7. Compared with Athlon, the global history counter buffer (used to record the CPU's access to data in a certain period of time, called the total history counter buffer) is four times larger, and the pipeline can accommodate more instructions before branch debugging. AMD's improvement in integer scheduling makes K8' s pipeline depth two levels more than Athlon's. The purpose of increasing the depth of two-stage conduit is to increase the core frequency of K8. In K8, AMD added a backup conversion buffer to meet Opteron's huge memory requirements in server applications.

■AMD launched the K 10 architecture in the second half of 2007.

Barcelona with K 10 architecture is quad-core with 463 million transistors. Barcelona is AMD's first quad-core processor, and its native architecture is based on 65nm process technology. Different from Intel Kentsfield quad-core, Barcelona does not package two dual-cores, but a real single-chip quad-core.

■ Introduce SSE 128 technology.

An important improvement of Barcelona is what AMD calls "SSE 128". In K8 architecture, the processor can process two SSE instructions in parallel, but the SSE execution unit generally has only 64-bit bandwidth. For SSE operation with 128 bits, the K8 processor needs to treat it as two 64-bit instructions. That is to say, when taking an SSE instruction with 128 bits, it needs to be decoded into two micro-operations, so a single instruction also occupies an extra decoding port, which reduces the execution efficiency.

■ The memory controller is strengthened again.

When AMD integrated the memory controller into the CPU, we saw a new and powerful K8 architecture. Today, Barcelona's memory controller will be designed to greatly improve memory performance again.

■ Innovation-L3 cache

Affected by technology, the cache capacity of AMD processors has been lagging behind Intel. AMD itself knows that it is impossible to add more transistors to the precious die to realize large-capacity cache, but the innovative AMD has found a better way-integrated memory controller.

■ Leading performance meets today's most pressing business needs.

Managers of data centers are facing increasing pressure, such as network services.

In recent years, AMD's main product LOGO( 18), database applications and other enterprise workloads have higher and higher demand for computing; In the current IT expenditure environment, it is necessary to achieve higher output with lower input. Fast-growing new computing technologies such as cloud computing and virtualization achieved a year-on-year growth rate of 60% in the second quarter of this year. These technologies are applied rapidly, and a balanced system solution is urgently needed. The latest quad-core AMD Opteron processor further enhances the advantages of AMD's unique direct-attached architecture, and can provide excellent stability and scalability solutions for the expanding heterogeneous computing environment, including cloud computing and virtualization.