A+ Core 1 3.5: CPU Features — Class Notes

Class notes · Professor Messer A+ Core 1 (220-1201) · Section 3, lesson 5

Quick reference: the short version of this lesson lives in the A+ Core 1 Section 3 resource sheets. This is an eighth lesson under objective 3.5. Where the compatibility lesson matched a CPU to a socket, this one is about the CPU’s own features: how wide it is, what architecture it uses, and how many cores it has.

Learning objectives

By the end of these notes you should be able to:

1. Explain the difference between 32-bit and 64-bit processors, including their address space and the x86/x64 labels.
2. Describe how bit-width affects drivers and applications, and where Windows installs each.
3. Describe the ARM architecture and its advantages.
4. Explain what CPU cores are and how they improve performance.

1. 32-bit and 64-bit

Open your system information and you will see lines like “64-bit operating system” and, perhaps, “ARM-based processor”. The 32-bit / 64-bit figure describes the CPU’s capability: how much information it can process at once, and the total address space a single CPU can reference.

The address space is the key idea. A 32-bit processor addresses up to 2^32 locations; a 64-bit processor addresses up to 2^64. In more usable terms:

• 32-bit can reference about 4 GB of memory.
• 64-bit can reference about 17 billion GB of memory (2^64 bytes is 16 exabytes, i.e. roughly 17,000,000,000 GB).

No real machine has anything like 17 billion GB fitted, and yours could not hold it — but a 64-bit processor lets operating systems scale up to very large amounts of memory, which a 32-bit address space simply cannot reach.

Address reach (memory a single CPU can reference):

  32-bit  |####|                              about 4 GB
  64-bit  |###########################...###|  about 17 billion GB
                                              (2^64 bytes = 16 EB)

Note (beyond this lesson): in practice a 32-bit version of Windows makes only about 3.25 GB usable, because part of that 4 GB address space is reserved for hardware. The 4 GB figure is the theoretical ceiling.

1.1 The x86 and x64 labels

In the Intel world you will see 32-bit processors called x86 — a nod back to the old 8086 CPU — and 64-bit processors called x64.

Note (beyond this lesson): x64 is also written x86-64 or AMD64; AMD designed the 64-bit extension of the x86 instruction set, and Intel adopted it (branded Intel 64). All are the same 64-bit x86 architecture.

1.2 Drivers and applications

Bit-width has to match in two places:

• Drivers. A 64-bit OS needs 64-bit hardware drivers; a 32-bit OS needs 32-bit drivers. They are not interchangeable.
• Applications. A 32-bit OS cannot run 64-bit applications. The reverse usually works: a 64-bit OS can generally run 32-bit applications.

Windows shows this split in its folders. 64-bit applications install to Program Files, while 32-bit applications install to Program Files (x86) — the “(x86)” folder is where the 32-bit software lives.

Exam tip: remember the one-way rule — 64-bit can run 32-bit, but 32-bit cannot run 64-bit. Pair it with the folders: plain Program Files = 64-bit, Program Files (x86) = 32-bit.

1.3 Worked example — reading your own system

You can check bit-width and architecture from a shell. The command differs by platform, which is itself a good illustration of the “architecture” idea:

$env:PROCESSOR_ARCHITECTURE     # Windows: AMD64, x86 or ARM64

uname -m                        # Linux/macOS: x86_64, aarch64 or armv7l

echo %PROCESSOR_ARCHITECTURE%   :: Windows CMD: AMD64, x86 or ARM64

If Windows reports AMD64, you have a 64-bit x86 OS; ARM64 means a 64-bit ARM OS; x86 means 32-bit. Knowing this tells you which drivers to fetch and whether a given application will install.

2. ARM

A second architecture has become hugely popular: ARM, created by ARM Limited. ARM stands for Advanced RISC Machine. ARM Limited writes the chip specification and licenses it to third parties, who build the actual processors.

Correction: the transcript expands ARM as “Advanced Risk Machine”. It is Advanced RISC Machine — RISC being Reduced Instruction Set Computing (and the name was originally “Acorn RISC Machine”). It is “RISC”, not “risk”.

ARM processors are prized for efficiency: they use less power, generate less heat, and are fast at processing. That efficiency is why roughly 99% of mobile phones use ARM processors — and why ARM is increasingly turning up in desktops and laptops too.

Note (beyond this lesson): ARM’s efficiency comes partly from being a RISC design (a smaller, simpler instruction set) versus the CISC style of traditional x86. Modern ARM PCs include Apple Silicon (the M-series) and Windows on ARM devices (e.g. Snapdragon X), which run many x86/x64 apps through an emulation layer.

3. CPU cores

We talk about a CPU as a single device, but inside that one CPU package are several individual processors called cores. A processor is often described by its core count — an 8-core or 16-core processor, for instance.

Each core contains its own processing logic (effectively its own CPU) and its own cache memory dedicated to that core. Because each core can work independently, multiple cores execute multiple instructions simultaneously, raising the overall efficiency of the machine. Look at a magnified CPU die and you can almost pick out the individual cores and the cache belonging to each — a rendering of a 16-core chip shows sixteen such blocks.

One CPU package, multiple cores:

+-------------------------------------------+
|  Core 0     Core 1     Core 2     Core 3   |
|  [CPU|L1]   [CPU|L1]   [CPU|L1]   [CPU|L1]  |
|                                           |
|  each core has its own logic and cache;   |
|  all cores run instructions at once       |
+-------------------------------------------+

Note (beyond this lesson): two extensions of this idea. First, a cache hierarchy — the per-core cache is usually L1/L2, with a larger L3 cache often shared across cores. Second, threads: with SMT (Intel calls it Hyper-Threading) one core presents as two logical processors, so an 8-core chip may show 16 threads. Cores are physical; threads are how the OS schedules work onto them.

4. Security perspective

CPU features are mostly a performance and compatibility topic, but three points matter to a defender:

• 64-bit is not just bigger, it is better defended. Moving off 32-bit is a security win as much as a capacity one. 64-bit Windows enforces driver signing, runs kernel patch protection (PatchGuard), and makes memory-safety mitigations (DEP/NX, ASLR) far more effective than on 32-bit. A machine still on a 32-bit OS is usually older, closer to end-of-support, and missing these — a larger and less-defended attack surface.
• Architecture decides what code runs — including malware. A binary built for x86, x64 or ARM only runs on that architecture, which is why analysts care which one a sample targets, and why moving to ARM breaks straight x86 malware. The flip side is the emulation layers (Windows on ARM’s x86/x64 emulation, Apple’s Rosetta) that let old binaries run on new chips — convenient, but an extra layer to account for when reasoning about what can execute.
• Shared core resources enable side channels. Because cores share caches and microarchitectural state, timing-based side-channel attacks (the Spectre and Meltdown family) can leak data across boundaries. It is an advanced topic, but the takeaway is practical: keep CPU microcode/firmware and the OS patched, since that is how these are mitigated.

Summary

• 32-bit vs 64-bit describes how much a CPU processes at once and its address space: 32-bit reaches ~4 GB (2^32), 64-bit reaches ~17 billion GB (2^64).
• Intel labels them x86 (32-bit, from the 8086) and x64 (64-bit).
• Drivers must match the OS bit-width. A 32-bit OS can’t run 64-bit apps; a 64-bit OS usually runs 32-bit apps. Windows: Program Files (64-bit) vs Program Files (x86) (32-bit).
• ARM (Advanced RISC Machine, licensed by ARM Limited) is efficient — less power and heat, fast — so ~99% of phones use it, and more PCs now do.
• A single CPU package holds multiple cores, each with its own logic and cache, running instructions in parallel; processors are rated by core count (8-core, 16-core).

Glossary

Review questions

1. What two things does the “32-bit/64-bit” figure describe about a CPU?
2. How much memory can a 32-bit processor address, and why?
3. Roughly how much can a 64-bit processor address?
4. What do the labels x86 and x64 refer to?
5. If you run a 64-bit OS, what must be true of your hardware drivers?
6. State the one-way rule about 32-bit and 64-bit applications.
7. In Windows, where do 32-bit versus 64-bit applications install?
8. What does ARM stand for, and what is ARM Limited’s role?
9. Give three advantages of ARM processors and one place they dominate.
10. What is a CPU core, and what does each core contain?
11. Why do more cores improve performance?
12. Scenario: an old line-of-business app is 64-bit only, but a PC runs a 32-bit OS. Will it run? What’s the fix?
13. Scenario: you’re told to move a fleet off a 32-bit OS “for security, not just RAM.” Name two protections 64-bit Windows adds.

Answer key

1. How much information it processes at once, and the total address space it can reference. Capability and reach.
2. About 4 GB, because it addresses up to 2^32 locations. The address width sets the ceiling.
3. About 17 billion GB (2^64 bytes = 16 exabytes). Vastly larger, so OSes can scale.
4. x86 = 32-bit (from the 8086); x64 = 64-bit. Intel-world shorthand.
5. They must be 64-bit drivers — 32-bit drivers won’t work. Bit-width must match the OS.
6. A 32-bit OS can’t run 64-bit apps; a 64-bit OS usually can run 32-bit apps. One-way compatibility.
7. 64-bit apps in Program Files; 32-bit apps in Program Files (x86). The “(x86)” folder holds 32-bit software.
8. Advanced RISC Machine; ARM Limited designs the specification and licenses it to third parties. It doesn’t (mainly) make the chips itself.
9. Less power, less heat, fast processing — dominant in mobile phones (~99%). Efficiency is the theme.
10. An independent processor inside the CPU package, each with its own processing logic and dedicated cache. A CPU within the CPU.
11. Multiple cores execute multiple instructions simultaneously. Parallel work raises throughput.
12. No — a 32-bit OS can’t run a 64-bit app; move the PC to a 64-bit OS (with 64-bit drivers). The OS bit-width is the blocker.
13. Enforced driver signing, kernel patch protection (PatchGuard), and stronger DEP/ASLR — any two. 64-bit adds real mitigations.