A discussion of the Central Processing Unit (CPU) opens this chapter. CPUs have been around long enough that they are more commonly just called processors. Features of CPUs that are used to rate and compare them are discussed. This is this year's list:
Intel processors used to have simple names (286, 386), then they got more creative names (Pentium) , and now they are grouped into generation sets, such as 7th and 8th generation, but they also categorize them as Core i8 and Core i9, which does not mean the same thing.
AMD processors are both less and more confusing. Let's listen to a kind gentleman who wants to speak about both brands.
Okay, so he tells us a lot. Some chips have graphics ability included, but gamers should still install graphics cards. The Intel Core iNumber notation does not tell us how many cores we get, only that a bigger number means it's better than one with a smaller number, but only within the same generation. Sort of. AMD notation is similar, but not clearer. Consider the Applying Concepts exercise on page 124. It shows that there are often many choices, each with different prices and performance features. Making choices in computer parts often comes down to what it is expected to do and how much can be spent to do it.
The text gives us a few pages on installing a new Intel processor, an AMD processor, and installing a cooling assembly on each. Note the use of thermal paste on top of the processor to conduct heat to the cooler. Failing to do so can lead to unpleasant results.
We do not recommend cooking on a processor: it ruins the food and the processor.
RAM (Random Access Memory) has come in several sorts of packages. Some RAM chips were soldered onto motherboards in the early days, or on expansion cards. RAM that is meant to be replaced by the user is more often contained on a SIMM (Single Inline Memory Module) or a DIMM (Dual Inline Memory Module). SIMMs will only be found in older computers now. They had chips on one side of a small card, while DIMMs typically have chips on both sides of the card. DIMMs also have little notches in their connectors, to identify whether they use 3.3 volts or 5 volts, and to identify them as registered (RFU), buffered, or unbuffered. Different systems will require different DIMMs. Laptops and small form factor devices may require SO-DIMMs, which are Small Outline DIMMs. They fit smaller slots, and take up less space.
DIMMs can be either DRAM (dynamic) or SRAM (static). The word static is misleading here. DRAM chips need to have their data refreshed dynamically every few milliseconds (often, every 4 milliseconds) with a jolt of power. SRAM chips do not need to have this periodic jolt, and will retain the data placed on them without the jolt, but only until the computer's power is turned off. In this respect, DRAM and SRAM chips will both lose their data when the computer is turned off. SRAM chips are more expensive, but give faster access, so they are used where this speed advantage is necessary.
The text lists several generations of DIMMs, starting with DDR (Double Data Rate) DIMMs, which could be read twice in a motherboard's clock cycle instead of once. This technology was followed by DDR2, DDR3, and DDR4, each of which is faster than its predecessor. Note the illustrations on page 138:
DIMMs can support dual, triple, and quad channeling, which allows the system memory controller to read two, three, or four DIMMs at the same time.
Memory comes in various speeds, measured in several ways. Some older memory was measured in nanoseconds (ns). A nanosecond is one billionth of a second. To picture it, electricity travels 11.78 inches along a copper wire in a nanosecond. In a microsecond (one millionth of a second), it travels 984 feet. (For readers outside the United States, these measurements are in US terms. The words millionth and billionth can have different meanings elsewhere. See the discussion of nanosecond on WhatIs.com.) If memory is rated at 70 nanoseconds (ns), it is slower than memory rated at 60 ns. Smaller numbers are faster.
DDR memory is measured differently. It is rated in MHz, or in a PC rating. For example, a MHz rating may state that the RAM is capable of running at 333 or 400 MHz. A PC rating will tell us the throughput of the module, in round numbers. A module that can process 64 bits at a time is processing 8 bytes at a time. We multiply the 8 bytes times the MHz rating of the module to get a PC rating: a module rated at 400 MHz times 8 bytes gives us a PC rating of PC3200.
DIMMs can be parity, ECC, or non-parity/non-ECC.
DIMMS can also be registered, buffered, or unbuffered. A buffer and a register are the same kind of thing, a temporary holding area. They just use different technology. Each holds an incoming signal and amplifies it before writing to the DIMM, making it more reliable.
Memory can also be measured and rated by its latency. In general, latency means "how long do we have to wait?" How long does it take to actually read or write to the module? Not a long time, to be sure, but the lower the latency rating, the faster the memory is. It may be expressed as column access strobe (CAS) latency or row access strobe (RAS) latency. They refer to the number of clock cycles it takes to read or write a column or row of data in the memory module.Obviously, low latency is better than high latency.
Adding and upgrading RAM are discussed. The author presents a list of questions that should be answered before buying RAM: