Current location - Quotes Website - Collection of slogans - What is the history of Intel's CPU?
What is the history of Intel's CPU?
Intel is one of the greatest companies in IT history, especially in the field of processors. It can even be said that the development history of Intel for more than 30 years is the evolution process of PC processors. From the earliest memory (on which Intel does business) to the latest "terascale" terascale chip technology, Intel has launched countless processors, some of which shine brilliantly in the market, while others leave the market suddenly. Combing Intel's processor product line, we have extracted 15 classic X86 processors (regardless of embedded processors for the time being). Reviewing the history of these processors is not only to review the past, but also to learn something new. We see that the performance of the processor is getting stronger and stronger, but the power consumption is gradually decreasing. Future processors are expected to continue this path, which can be said to be "better and more powerful". Let's take a look at the glorious moment of 15 processor!

1, 8086: the first PC processor

8086 is the first x86 CPU-before that, Intel has released 4004, 8008, 8080, 8085 and other CPUs. 8086 can use the external 20-bit address bus to manage the memory of 1MB. However, the speed of 4.77MHz chosen by IBM is indeed a bit low, and the speed can reach 10MHz before delisting.

The processor used in the world's first PC is a derivative of 8086-8088, with only 8-bit (external) data bus. Interestingly, the control system on the American space shuttle uses 8086 processing. In 2002, NASA bought several 8086s in Yi Bei because Intel no longer supplied them.

2.80286: Support 16MB memory, or 16 bits.

The performance of 80286 released by 1982 is three times that of the same frequency 8086 processor. It can support 16MB memory, but it is still a 16-bit processor. It is the first processor with MMU (Memory Management Unit) module, which enables it to manage virtual memory. Like 8086, there is no floating-point arithmetic unit (FPU), but X87 coprocessor can be used. Its highest frequency is 12.5MHz, in contrast, the speed of competitors has reached 25MHz.

3,80386: 32-bit cache

Intel's 80386 is the first 32-bit X86 processor. There are several versions, the most famous of which are/kloc-386 SX (single word external connection) for 0/6 bit data bus and 386 DX (double word external connection) for 32 bit data bus. The other two versions are not worth mentioning: 386 SL provides (external) cache management function for the first time, and 386SL.

4.486: It is the first time to have APU (floating point arithmetic unit) and multiplier.

The emergence of 486 is a sign of the times. For a long time, 486 DX2/66 is the minimum configuration for gamers. This CPU released by 1989 brings several interesting new functions: onboard APU, data cache and first clock multiplier. The combination of onboard APU and x87 coprocessor constitutes the 486 DX (non-SX) series. There is an 8KB L 1 cache inside the processor (the write-back speed is slightly faster than the write-in speed), which also makes it possible to integrate L2 cache on the motherboard (running at bus frequency).

The second generation 486 began to have CPU multipliers. With the release of DX2(2 multiplier) and DX4(3 multiplier) series, the frequency of processors began to be higher than FSB (Front Side Bus). There is also a little story. The 487SX sold as the APU of the 486SX is actually a 486DX that shields some cores.

5. Pentium: a BUG that brings trouble

Pentium, which came out of 1993, is remarkable for many reasons: it abandons the traditional naming method of numbers, because Intel is forbidden to use numbers as trademarks, and the most famous one is a BUG. Some division operations of the first generation Pentium processors will produce incorrect results. Although Intel quickly replaced these processors, the adverse effects have been caused. This rare BUG once made IT media coverage overwhelming.

Pentium * * * has three different versions. The first version didn't have a CPU multiplier, and the second version had a multiplier (including the famous Pentium 166). Finally, it began to support the SIMD instruction set of X86 architecture-MMX. Pentium MMX also increased the size of L 1 cache and made a slight improvement. This is Intel's first X86 CPU that can execute two instructions at the same time. Its L2 cache is integrated on the motherboard, and its running frequency is equal to FSB frequency.

Here we explain the BUG of Pentium: some calculations using FPU will lead to incorrect results. The probability of this error is very small, and Inel quickly replaced the problem product for free. The following is an example of Pentium error:

4195835.0/3145727.0 =1.333 820 449136 241002 (correct result) 4195835.

6. Pentium Pro: Support more than 4GB of memory for the first time.

Pentium Pro released by 1995 is the first processor to support more than 4GB of memory. By using 36-bit physical address extension (PAE) technology, it can support up to 64GB of memory. This CPU is also the first processor with P6 architecture (from which Core 2 core is derived), and it is also the first time to integrate L2 cache inside the CPU. In fact, the cache from 256KB to 1MB is placed next to the CPU core, which is at the same speed as the CPU and is no longer on the board.

This CPU also has a performance problem. The performance of running 32-bit programs is very good, but running 16-bit programs (such as Windows 95 system) is much slower, because there may be some problems in managing 32-bit registers with 16-bit registers, which offsets the advantages of Pentium Pro out-of-order execution architecture.

7. Pentium II and III: Same Brothers

The Pentium II released by 1997 is the product of Pentium Pro's beginning to move towards the general public (Pentium Pro is not popular). On the whole, it is similar to Pentium Pro, but there are some differences in caching. L2 cache is no longer at the same speed as the CPU core (which is costly). P II's 5 12KB cache works at half the speed of CPU. In addition, Intel abandoned the traditional packaging method and began to package L2 cache inside the shell, instead of integrating it on the motherboard or processor as before.

Compared with Pentium Pro, Pentium II natively supports MMX(SIMD) instruction and has dual L2 cache. Pentium III(Katmai core) released by 1999 is the same as Pentium II except that it supports SSE(SIMD) instruction.

Pentium II and III both have 5 12KB L2 cache, but the mobile version of Pentium II manufactured by 180nm technology only has 256KB L2 cache, but this processor runs much faster than the desktop version.

8. Celeron and Xeon: aim at the low end/high end.

In the late 1990s, Intel launched two famous brands: Celeron and Xeon. The former aims at the entry-level market, while the latter intends to set foot in the field of servers and workstations. The first generation Celeron actually castrated the Pentium II with L2 cache. At that time, compared with Xeon processors with larger L2 cache, its performance was very poor. Up to now, these two brands still exist: entry-level Celeron (usually reducing L2 cache and FSB speed) and Xeon in the server field (high frequency, high FSB speed and large-capacity cache).

Intel later added L2 cache (only 128KB) to Celeron, among which Celeron 300A has been the hottest star product in the market for a long time with 50% overclocking.

Like P Ⅱ, there is an external L2 cache in the Xeon shell, with the capacity ranging from 5 12KB to 2MB, and the number of transistors ranging from 3 1M to 124M.

9. Impact of 1GHz Pentium III.

Pentium, the core of Coppermine, is the first X86 processor in Intel's history to reach 1Ghz, and even introduced the model of 1. 13GHz, but it quickly withdrew from the market due to instability. The new version of Pentium III increases the L2 cache capacity in the kernel, which is faster than the earlier model with external 5 12KB L2 cache. Intel claims that it can also speed up Internet surfing. * * * There are three versions of the release: server-level (Xeon), entry-level (Celeron) and mobile version (SpeedStep skill technology is introduced for the first time).

In 2002, an improved version of Tualatin (tualatin) Pentium III was released, which has a 5 12KB L2 cache and is manufactured by a more advanced 130nm process. Initially, it was used by Intel in the server and mobile markets, so it was only a flash in the pan in the consumer market and was unknown.

10, Pentium 4: synonymous with high noise and low performance.

In 2000, Intel announced the launch of a new generation of processors-Pentium 4. Although its clock frequency is higher (the lowest speed reaches 1.4GHz), its performance at the same frequency is far inferior to that of its competitors. ADM's Athlon (even its own Pentium III) runs faster than it at the same frequency. Most importantly, Intel decided to abandon the mainstream memory specifications and only support RAMBUS's RDRAM memory (the only memory that could meet the bandwidth requirements of Pentium 4 at that time), but it failed in the end. Despite its high price and high calorific value, Pentium 4 survived in the market for several years with many technical improvements (such as adding L3 cache and supporting hyper-threading technology).

There are three P4 processors in the market: Mobile (adding a set of variable multipliers), Clerlon (simplifying L2 cache) and Xeon (adding L3 cache). Hyper-threading technology and L3 first appeared in the server market, and then were introduced into general processors (L3 cache only appeared in EE Extreme models).

This is FSB, supported by a technology called QDR (Quadruple Speed Data Transmission), which is four times faster than the rated clock frequency. The bus speed of 400MHz is actually only 100MHz, and the real speed of 533MHz is only 133MHz. In 2005, Intel also released a 64-bit P4 processor, which we will talk about later.

1 1, Pentium m: started to exert its strength in the notebook computer market.

In 2003, the portable PC market began to explode. At this time, Intel has only two CPUs to choose from: P3 and P4 in backward tualatin, but P4' s huge calorific value determines that it cannot be suitable for portable computer processing. Just then, a savior came to the Israeli laboratory: Banas (also known as Pentium M). This processor based on P6 architecture (such as Pentium Pro) has high performance and ultra-low power consumption beyond P4. It became the processor of Intel Centrino platform and was replaced by the faster Dothan core in 2004. Pentium M has left a deep imprint on the mobile platform, and Stealey(A 100) still uses Dothan architecture (only with lower frequency and lower power consumption).

Like the desktop P4, its FSB is four times the rated frequency (QDR), and the socket is 479. In fact, there are only 478 pins, but the definition of each pin is different from the 478 sockets of the desktop P4.

12, Pentium 4: Starting to support 64-bit and transforming to dual-core.

In 2005, Intel improved the P4 processor twice: first, it brought Prescott -2M, and then it released the core product of Smithfield. The former is a 64-bit processor based on Proscott, and the latter is a dual-core processor. They are very similar to P4, but they face the same problems: low IPC (instruction per cycle) calculation, and it is difficult to increase the frequency. These two processors are no longer the focus of Intel (their focus is on the future Core 2). In addition, Pentium D says it is a dual-core processor, but in fact it just encapsulates two Proscott cores in one shell.

Interestingly, P4 for the consumer market does not support PAE technology (using 36-bit instead of 32-bit management memory), so the maximum memory supported is limited to 4GB, but it can break through this limit. In fact, the address bus is still limited to 36 bits (40 bits on Xeon), but PAE technology has become history-64-bit programs can make full use of all memory.

Hyper-threading technology (Xeon and EE Extreme Edition) can be supported on some specific models. Intel later released 9x0 series P4 at 65nm, but there was no obvious improvement.

13, the first dual-core mobile version

In 2006, Intel released the Core Duo processor. This is the first dual-core processor designed for portable computers, with excellent performance, at least much faster than P4. This is also the first true dual-core X86 processor, enjoying the cache design. The previous Pentium D dual-core was more like packaging two processors in one shell. Core processor is an important part of Intel Centrino platform and has achieved great success in the market. The only drawback is that it is still a 32-bit processor and does not support 64-bit technology like P 4.

Single-core Core Solo also appears on the market. The FSB speed of this product, which pursues low power consumption, is reduced from 667 MHz to 533MHz and applied to the server (code name Sossaman). This is also the first time that a CPU specially designed for mobile has been used in the server field. In fact, the Core processor does not use the architecture of the Core 2 processor, and it was quickly replaced by the Core 2 of Merom Core in the portable PC market. In addition, Socket 479 of Yonah core is different from Socket 479 of Pentium M (although the name is the same).

14, today's mainstay: Core 2

In 2006, Intel released the Core 2 processor, and then it became a hot commodity in the market. This Pentium M-based processor has a brand-new core architecture. Previously, Intel had two product lines: P4 focusing on the desktop market and Pentium M focusing on the mobile market, and they also jointly built a server product line. Now, Intel only needs a micro-architecture to meet all product lines, and a 64-bit Core 2 can be used in all fields from low-end to high-end, from desktop to portable to server.

There are many models of Core 2 architecture in the market, which are classified according to different configurations, including different number of cores (from 1 to 4, single core to quad core), cache size (from 5 12KB to 12MB) and FSB speed (from 400MHz to 1600MHz).

The following table shows the original data of Core 2, but the latest 45nm version is also applicable.

The specifications of the mobile version of Merom are basically the same, but the FSB is slightly reduced, while the EE Extreme Edition is faster. Core 2 is also quad-core, in fact, only two core cores are packaged together. The 45nm Core 2(Penryn) cache is larger and has lower calorific value, but the basic architecture root is similar to the above.

15, future: Nehalem, Atom, etc.

Of course, these are only part of this article. The second part about AMD processors (including AMD-ATI graphics cards) will come soon. The story of Intel X86 processor will not end with Core 2 Duo, and the part about Intel's future processor has been planned, because Nehalem and Atom are also X86 processors. Moreover, it is revealed that Larrabee, Intel's entry into the graphics card market, is also based on X86 processor cores.

From: /837/article_836305.html