Memory Hierarchy

 Figure 1.4 shows the memory hierarchy in real terms, where we see can see how the various memory elements grow exponentially in size and decrease exponentially in access time. Closest memory to the CPU is the smallest, typically in the range of 1 Kbyte of data (1,000 8-bit charac-ters) to 1 Mbyte of data (1,000,000 characters). These on-chip cache memories can have access times as short as ½ to 1 nanosecond or one-billionth of a second. As a benchmark, light can travel about one foot through the air in one nanosecond. We call this cache the Level 1, or L1 cache. Below the L1 cache is the Level 2, or L2 cache. In today’s Pentium-class processors, the L2 cache is usually on the processor chip itself. In fact, if you could lift the lid of a Pentium or Athlon processor and look at the silicon die itself under a microscope you might be surprised to find that the biggest percentage of chip area was taken up by the cache memories. The secondary cache sits your computer’s main memory. These are the “memory sticks” that you can purchase in a computer shop to increase your computer’s perfor-mance. This memory is considerably slower than on-chip cache memory, but you usually have a much larger main memory then the on-chip memory. You might wonder why adding more memory will give you better performance. To see this, consider the next level down from main memory, the hard disk drive. Your hard drive has lots of capacity, but it comes at a price, speed. The hard disk is an electro-mechanical device. The data is stored on a rotating platter and, even at a rotational speed of 7200 revolutions per minute, it takes precious time for the correct data to come under the disk read heads. Also, the data is organized as individual tracks on each side of multiple platters. The heads have to move from track-to-track in order to access the correct data. This takes time as well. Now, your operating system, whether its MAC O/S, Linux or one of the flavors of Windows, uses the hard