32-bit computing – the end of an era
We appear to be approaching the end of 32-bit computing. Much of the newer hardware uses 64-bit processors, with a 32-bit compatibility mode. However, support for 32-bit computing is on the wane.
I was reminded of this change this morning, when I updated my opensuse 12.3 systems. The update brought KDE to version 4.10.2. And on my one 32-bit system, I discovered that KDE activities have been crippled. The full activity capability is still there for 64-bit systems.
We see the same diminishing support elsewhere. The newer computers are coming with UEFI firmware, which is specifically oriented toward 64-bit systems. The newer hard drives are coming with GPT partitioning, due to limitations of the older FDISK partitioning scheme. And at least some of the software will refuse to use GPT partitioning unless the firmware is using UEFI. As I understand it, neither Windows 7 nor Windows 8 can be installed on the newer hardware, unless UEFI is disabled and the disk is repartitioned in classic FDISK mode.
There are still many smart phones and other mobile systems using 32-bit ARM hardware, but that is likely to move to 64-bit ARM within the next few years.
For myself, I have only one computer with 32-bit hardware. It is an older system, purchased in 2004. I am using it to test 32-bit compatibility, but I do not use it for anything important. All of my other systems are 64-bit. I do have 32-bit Windows XP and 32-bit Windows Vista installed on one box, but the processor for that box is 64-bit, and the linux that I mainly use on that box is 64 bit.
My first computing
The first computer that I ever used was 32-bit. It was an IBM S/370 mainframe. It had 128K of memory, which was huge for that time. I did not own that computer, of course. It was a campus computer, used for research and teaching. I only saw it as a user (a math professor teaching the occasional computer science class). It had 24-bit memory addressing, and 32-bit instructions.
The first computer that I owned
The first computer that I ever owned was an 8-bit Heathkit system. It used the Z80 processor (from Zilog), which was based on the Intel 8080 processor that helped start the PC revolution. It used an 8-bit data path, and 16-bit memory addressing. I had a few 16-bit arithmetic instructions, but it’s instruction set was mainly 8-bit.
I learned quite a bit from that Heathkit computer, both from building (from a kit) and from writing some of my own low-level device drivers. However, that was more than 30 years ago, and that computer has long since gone to the graveyard for old computers.
My experience with the IBM-PC series started mainly with the PC-AT, which we purchased relatively soon after they became available. The early computers in this series used the 8088 processor, with an 8-bit data bus and 16-bit instructions. The PC-AT actually used the 80286 processor, which had a 32-bit instruction set if put into protected mode. It was often said that the 80286 was brain-dead in 32-bit mode, so it was mainly used in 16-bit mode. The 32-bit mode was not backward compatible with the 8-bit 8088 and the 16-bit 8086 instruction sets.
For many people, 32-bit computing became a reality with the Intel 80386 processor. This was a 32-bit processor, with a 16-bit compatibility mode. It could be used with existing PC software in 16-bit mode, or it could be switched to protected mode to use the 32-bit instructions. The important feature of the 80386, was that in protected mode it could be set to emulate 16-bit processing. And, using that, the older 16-bit software could still run under the protected mode of the 80386.
The 80386 was a breakthrough processor. Although MS Windows versions existed for 16-bit systems, it was the 80386 that made them viable. A few years later, Microsoft Windows 95 came out for the pentium processors.
The last 32-bit system that I purchased had an 80686 processor, and is the one 32-bit system that I still have. It came with 192M of memory, but with a wide enough bus that I could later add an additional 1G.
My first 64-bit processor
When an even older 32-bit system failed in 2007, I purchased a Dell Dimension to replace it. And that system came with a 64-bit AMD two-core processor. It came with 32-bit Vista installed. I soon repartitioned the disk, and reinstalled Vista into a smaller partition. Back in those days, Dell provided re-install media.
With the repartitioning, I quickly installed WinXP to have something that I could use while getting the hang of Vista. And the next step was to install linux, which was what I would mostly use.
My first install of linux (SuSE 10.1) seemed to go well. However, it would not talk to the Internet. The system thought that the network card was working fine, but it failed to communicate.
My second install did better. I noticed an architecture option on the first install window, so I switched to a 32-bit install instead of a 64-bit install. And the network was just fine with the 32-bit install.
Since that time, 64-bit software has matured. I no longer have problems on that Dell box with a 64 bit opensuse install. So I see no need to go back to 32-bit.
So Rest in Peace, 32-bit computing. You have served us well but your time has passed.