Skip to content

Program Files and Program Files (x86)

· engineering

Every Windows machine I used growing up had two folders sitting next to each other: Program Files and Program Files (x86). Some PCs said “32-bit” in their system settings and others said “64-bit”. I never asked why. This week I was poking at memory in a small Go project, and the question finally got an answer.

What the number means

The “32” or “64” is how much data the processor works with in one go. That chunk is called a word. On a 32-bit machine a word is 4 bytes; on a 64-bit machine it is 8.

The word size also sets the size of a memory address, and that is where the difference you can feel comes from. Every byte a program uses needs its own address.

  • With 32 bits you can count up to about 4.3 billion addresses. That is 4 GB, and it is the most memory a 32-bit program can ever see, however much RAM is in the machine.
  • With 64 bits the ceiling is so high it stops mattering. Today’s processors only use about 48 of those bits, which still comes to around 256 TB.

A 64-bit processor is also quicker with big numbers. It adds two 8-byte numbers in one step, while a 32-bit processor has to split the work into pieces.

Why some PCs said 32-bit

What Windows shows on its About page is the bit size of Windows itself, not of the processor. Those are separate layers:

  • The processor. Nearly every PC processor since around 2005 is 64-bit.
  • Windows. 64-bit Windows needs a 64-bit processor. 32-bit Windows runs on either kind.
  • Each app. 64-bit Windows runs both 32-bit and 64-bit apps. 32-bit Windows only runs 32-bit apps.

So most of the “32-bit” PCs I remember were probably 64-bit machines running 32-bit Windows. That was common up to the Windows 7 years, and it explains two things I saw back then without understanding them:

  • The RAM never added up. A PC with 8 GB installed showed about 3.5 GB usable. The 4 GB address limit was the cap, and part of those 4 GB of addresses was set aside for hardware such as the graphics card.
  • Big programs ran out of memory. On 32-bit Windows, each app got 2 GB by default, so heavy software crashed long before the machine was actually full.

Two folders for two kinds of app

64-bit Windows still has to run all the old 32-bit software people depend on. It does that through a built-in compatibility layer called WOW64, and it keeps the two kinds of app apart on disk:

  • Program Files is for 64-bit apps.
  • Program Files (x86) is for 32-bit apps.

“x86” is the old family name for 32-bit Intel and AMD processors, and “x64” means their 64-bit successors. Once you know that, the folder name just says “programs for the old 32-bit processors”. The same split shows up in Task Manager, where 32-bit apps get “(32 bit)” after their names.

The system folders follow the same logic with worse names. System32 holds the 64-bit system files, and SysWOW64 holds the 32-bit ones. They look backwards because Microsoft kept the old names so that old software, which expects System32 to exist, would not break.

Where it shows up for me

Windows 11 only comes in 64-bit, so 32-bit Windows is mostly gone. The split is still around in tooling, though. Go uses the same names for its build targets: GOARCH=386 builds a 32-bit x86 program and GOARCH=amd64 builds a 64-bit one.

It also changes how much memory a program uses. In Go, every pointer, every int and every string header is one word, so they are all twice as big on 64-bit. The same program takes more memory on a 64-bit machine, and in return it can reach far more of it.

The two folders I ignored for years turned out to be Windows quietly running two generations of software side by side.