PowerPC Architecture and X86 architecture

PowerPC Architecture
PowerPC is aStreamlined instruction Set(RISC) Architecture of the central Processing Unit (CPU), whose basic design derives from the power (performanceoptimized with enhanced RISC) of IBM (International Business Machine Company), the Ibmconnect newsletter August 2007 "Enhanced RISC Performance optimization") architecture. Power is the microprocessor architecture exhibited by the Aim Alliance of Apple (Apple), IBM, Motorola (Moto), which was made in 1991. PowerPC is part of the entire AIM Alliance platform and is by far the only part. But Apple has switched its computer products to INTELCPU since 2005.

PowerPC's history dates back to the Ibmpower architecture that was introduced along with RISC system/6000 as early as 1990. The design was inspired by early RISC architectures such as IBM 801 and MIPS architecture processors.

In the 1990 's, IBM, Apple and Motorola developed PowerPC chips that were successful and produced PowerPC-based multiprocessor computers. PowerPC architecture is characterized by a good scalability, convenient and flexible. The first generation of PowerPC uses a 0.6 micron production process, with transistors having a single chip integration of 3 million. The advent of copper chips in the 1998 ushered in a new historical era. In 2000, IBM launched a large number of products using copper chips, such as Rs/6000 's X80 series. Copper technology replaces aluminum technology, which has been in use for more than 30 years, Silicon chip multi-CPU production process reached the level of 0.20 microns, single-chip integration of 200 million transistors, greatly improve the computing performance, and 1.8V of Low voltage operation (formerly 2.5V) greatly reduce the power consumption of the chip, easy to heat dissipation, thereby greatly improving the stability of the system.


X86 Architecture
30 years ago, June 8, 1978, Intel released the new 16-bit microprocessor "8086", but also created a new era: the x86 architecture was born.x86 refers to a set of computer language instructions executed by a particular microprocessor, which defines the basic rules of use of the chip ., as today's x64, IA64 and so on.

In fact, the launch of the 8086 processor has not received much attention, and has not been widely used in the beginning, but its position in the PC industry can not be described, this is because it brings x86. Not only does it make Intel's heyday, it's also an industry standard, and even on today's powerful multi-core processors you can see x86. In the history of the 30, the x86 family continued to grow, moving from the desktop to notebooks, servers, supercomputers, writing equipment, while also frustrating or limiting the development of many competitors, so many processor manufacturers and their architecture technology has become a historical name, even if some closed development is unsustainable, Apple, for example, has given up PowerPC.

Of course, we must not forget the struggles of x86-64 and EM64T. In 2003, AMD introduced the industry's first 64-bit processor Athlon64, which also brought the x86-64, a 64-bit extended superset of the x86 instruction set, with backward-compatible features. Intel was also implementing 64-bit technology at the time, but its IA64 architecture was not compatible with x86, but only on the server processor Itanium. To compete with AMD, Intel launched its own 64-bit version of x86 in 2004, also known as EM64T. AMD and Intel are blaming each other for this, but in any case at least the development and popularization of 64-bit technology has allowed x86 technology to flourish. "This proves that the flexibility of the x86 instruction set can be used against Intel, so even if Intel dominates the market, other companies can still change the direction of x86," says Davidpatterson, a professor of computer science at the University of California, Berkeley, and one of RISC inventors. ”

The x86 is a standard number abbreviation for an Intel general-purpose computer family, and also identifies a set of generic sets of machine instructions, X has nothing to do with the processor, and is a simple wildcard definition for all *86 systems, such as i386,586, Pentium (Pentium). Since earlier Intel's CPU numbers are numbered like 8086,80286, since this entire series of CPUs are directive compatible, they are used X86 to identify the set of instructions used today's Pentium, P2,P4, Celeron series are supported X86 instruction system, so all belong to the X86 family. The X86 instruction set was developed by Intel Corporation for its first 16-bit CPU (i8086), and the cpu--i8088 (i8086 simplified version) of the world's first PC, introduced by IBM in 1981, is also the X86 directive. In addition, the X87 chip series math coprocessor, which increases the ability of floating-point data processing in the computer, uses the X87 instruction, and the X86 instruction set and the X87 instruction set are collectively referred to as the X86 instruction set. Although with the development of CPU technology, Intel has developed a new i80386, i80486 until today's PENTIUM4 (the following P4) series, but in order to ensure that the computer can continue to run the various applications developed in the past to protect and inherit rich software resources, So all the CPUs produced by Intel still continue to use the X86 instruction set, so its CPU still belongs to the X86 series.

In addition to the Intel Company, AMD and Cyrix, and other manufacturers have been able to use the X86 instruction set of the CPU, because these CPUs can run all the software developed for the INTELCPU, so the computer industry will be listed as Intel's CPU compatible products. Because the INTELX86 series and its compatible CPUs all use the X86 instruction set, they form today's vast X86 series and compatible CPU lineup. Of course, in the current desktop (portable) computers are not all using the X86 series of CPUs, some servers and Apple (Macintosh) machine also use the United States Digital (digital) company Alpha61164 and PowerPC 604e series CPU.

Intel starts with 8086, 286, 386, 486, 586, P1, P2, P3, P4 all use the same kind of CPU architecture, collectively known as X86. Intel launched the X86 architecture for 30 years, compared with 486, Pentium a big step forward, and pⅱ the pace of progress is not so large, X86 CPU development seems to have come to an end. Intel is very aware that the X86 instruction set limits the further increase in CPU performance, so they are working with HP to develop next-generation instruction set architectures (Instructionset Architecture, ISA): EPIC (Explicitly Parallel instructioncomputing, explicit parallel instruction calculation). For Intel, IA-64 (Intel's 64-bit architecture) is the next 10-15-year architecture. The new Isa will allow Intel to get rid of the limitations of the X86 architecture, thus designing a new processor that transcends all existing risccpu and X86 CPUs. So where is Epic's advanced? Why would intel abandon the X86 architecture that made it a chip giant? IA-32 problems we know that engineers can improve performance by increasing the number of instruction executions per clock, and the primary purpose of Intel's new instruction set is to make instructions easier to decode and execute in parallel. This allows the development of new processors without restriction. However, for engineers, a 8086-compliant X86 instruction set is a must-have task. After all, compatibility with previous generations is a key factor in Intel's growth, but it also protects the user's original investments and uses millions of applications.

so why give up the whole X86 instruction set again? Where is the shortage of X86?
(1) Variable instruction length X86 instruction length is variable, and there are several different formats, resulting in X86CPU decoding work is very complex, in order to improve the CPU operating frequency, have to extend the CPU pipeline, and too long pipeline in the case of branch prediction error, It also brings the disadvantage that the CPU work has a long stagnation time.
(2) The poor X86 instruction set architecture for registers has only 8 universal registers, and only 6 are actually used. This situation with the modern superscalar CPU is very uncomfortable, although engineers use register renaming technology to compensate for this flaw, but caused the CPU too complex, pipelining too long situation.
(3) The memory Access X86 instruction accesses the memory address, while modern RISCCPU uses Load/store mode, only the load and store instructions can read data from memory to the register, and all other instructions only calculate the operand in the register. In the current CPU speed is 5 times times the memory speed or more than 5 times times the case, the latter mode of operation is the right way.
(4) floating-point stack X87FPU is currently the slowest FPU, one of the main reasons is that the X87 instruction uses an operand stack. If you don't have enough registers to compute, you'll have to use stacks to hold the data, which can waste a lot of time using the FXCH directive (that is, putting the right data on top of the stack).
(5) 4GB limit This does not seem to be the problem, but, 6 years ago, the mainstream PC only 4MB of memory, and most of the current PC equipped with more than 64MB of memory, is the previous 16 times times, so, in the next 10 years, the PC memory breakthrough 1GB will definitely not be surprising, And the current large server has used more than 1GB of memory, the breakthrough of 4GB memory will soon appear.
(6) The chip becomes larger all the methods used to improve the performance of X86CPU, such as register renaming, huge buffers, disorderly execution, branch prediction, X86 instruction conversion, etc., all make the chip area of the CPU become larger, also limit the work frequency further increase, and the additional integration of these transistors is only to solve the problem of X86 instructions.

PowerPC schema and X86 schema