How they improved the performance of CPU
Improving the CPU quality is closely related to electronic technology improvement. According to industry research, there will be new type of CPU or PC launch to market every 3 to 4 years time. Here are a few methods they have used to improve the CPU performance
This is very obvious and everybody know that faster CPU internal Clock speed (or) CPU speed make better CPU performance. The Clock Speed is increased from 8Mz ( Intel 8086 CPU) to 500Mhz (Intel Pentium-III CPU) gradually whenever electronic technology is available to do so.
Data Bus width (internal & external data bus width) is one of the important factor to improve the CPU performance. Basically, smaller Data Bus width takes more time to read/transfer data from one place to another. The external Data Bus width of very first CPU (Intel 8086 CPU) was only 8-bits. Then next model (Intel 80386 CPU) used 16-bits external Data Bus width. After some time later newer Intel 80386 CPU and Intel 80486 CPU used 32-bits Data Bus size. Latest Pentium/Pentium-II series CPUs are using 64-bits external Data Bus size which performance is obviously better than old CPUs.
The purpose of cache memory is to enable the processor to run in a constant state of flux. Whenever the processor is made to wait on data or instructions will slowdown the overall computation. Memory caching is effective because most programs access the same data or instructions over and over. By keeping as much of this information as possible in cache, the computer avoids accessing the slower system memory. When data is read from, or written to, main memory a copy is also saved in the cache, along with the associated main memory address. The cache monitors addresses of subsequent reads to see if the required data is already in the cache. If it is (a cache hit) then it is returned immediately and the main memory read is aborted (or not started). If the data is not cached (a cache miss) then it is fetched from main memory (main memory is the RAM memory on motherboard) and also saved in the cache. The cache is built from faster memory chips than main memory, so a cache hit takes much less time to complete than a normal memory access.
Level 1 Cache is typically 8K to 64K in size. It is built directly into the CPU and is extremely effective because it operates at the same speed as the processor. Due to this speed and the highly-integrated design, large L1 caches produce very large performance gains. Level l cache uses very fast type SRAM memory, which are built into the architecture of microprocessors. Such internal caches are expensive and often called Primary Cache memory (or) L1 caches.
The Intel 80486 microprocessor, for example, contains an 8K memory cache, and the Pentium and Pentium-Pro has a 16K L1 Cache. Pentium-MMX, Pentium-II, Celeron and Pentium-III contains 32K L1 Cache memory.
AMD K5 CPU has 24K and AMD K6 CPU has 64K L1 cache.
Cyrix 6x86/ 6x86MX CPU contain 16K and CyrixM-II CPU has 64K L1 cache.
Level 2 Cache is located externally to the CPU. It is generally 256K to 1024K in size. "Conventional" L2 caches generally run at speeds which are substantially slower than the CPUs they service, yet they are significantly faster than the computer's main System Memory. L2 cache improves system-level performance by improving the CPU’s memory read and write performance, as well as decreasing the system bus utilization. A larger, more efficient L2 cache will result in less processor read requirements to main memory which is implemented in slower DRAM thus reducing the number of times the processor needs to access the bus.
Start from Pentium-Pro CPU, Intel coupled 256K/512K L2 Cache memory together with CPU and let them run at full speed of processor. All Intel Pentium-II CPU has 512K L2 cache memory except Celeron with 128K only.
AMD and Cyrix CPUs do not contain L2 cache within CPU module. These CPUs look for L2 on motherboard and size is usually 256K or 512K depend on motherboard manufacturer.
MMX is a technology from Intel that is designed to run faster when playing multimedia applications. According to Intel, a PC with an MMX microprocessor runs a multimedia application up to 60% faster than one with a microprocessor having the same clock speed but without MMX. In addition, an MMX microprocessor runs other applications about 10% faster.
The MMX technology consists of three improvements over the non-MMX Pentium microprocessor:
6. 3Dnow! Technology
3DNow! technology is a technology from AMD that is a new Superscalar SIMD (single instruction multiple data) instruction set for improving 3D graphics, sound and or imaging within your PC. 3DNow! technology enables new levels of realism on your PC with lifelike visuals, big-screen sound, and more realistic 3D imaging. 3DNow! technology is the first innovation to the x86 architecture that significantly enhances 3D graphics, multimedia, and other floating-point-intensive PC applications to enable a superior visual computing experience.
3DNow! technology is a set of 21 instructions that use SIMD (Single Instruction Multiple Data) and other performance enhancements to open the performance bottleneck in the 3D graphics pipeline between the host CPU and the 3D graphics accelerator card. 3DNow! works hand-in-hand with leading 3D graphics accelerators to achieve faster frame rates on high-resolution scenes, improved physical modeling of real-world environments, realistic 3D graphics and images, and theater-quality audio and video.
7. Others
This section is under construction.