What you’ll learn:
This iteration of Electronic Design: Now and Then was inspired when I was reading "Using Magnetic Cores in Computers" from our archives. This article was originally published in Electronic Design April 1955, Vol. 3 No. 4. which was a while ago. In fact, I was born that year.
We will get to the usual use of magnetic cores for storage, but first a mention about the article. As it turns out, iron cores and magnetics have lots of uses. They're still used as chokes in power and analog circuits. However, in this case, the idea was to use magnetic cores to implement logic (Fig. 1). Check out the article for details. Though the technology never made it into the commercial realm, it worked.
So, next, we take a look at what was happening then with magnetic cores and then we get into what's happening now in storage, where magnetic cores were originally used. Lots of memory technologies have come and gone with new ones continuing to emerge.
What are Magnetic Cores
Magnetic cores are magnetic materials much like iron. The orientation of the magnetic field can be used to store a single bit of information. Magnetic cores employed for storage purposes were mounted in a grid with a set of wires crisscrossing the array.
Magnetic cores used for analog and power purposes come in a wide variety of forms depending on their use. They're normally a one-off within a design, unlike magnetic cores used for storage that are replicated in the millions.
Then: Using Magnetic Cores for Storage
Magnetic cores were used as main memory for mainframes and minicomputers (Fig. 2). The cores were very tiny and the grids were stacked to provide more memory in a module. In most systems, multiple modules were applied to provide a memory capacity that's dwarfed by today’s solid-state solutions.
One big advantage of magnetic core memory, or simply core memory, is its non-volatile nature. This was a big advantage given how long it took to load things from card decks, tape drives, and, later, disk drives. A boot loader would typically checksum the portion of memory containing the operating system, and possibly applications, and use that if the checksum was correct, just in case something went wrong beforehand.
I encountered core memory with the Burroughs large system (aka mainframe), 48-bit B5000 series with 32 kwords of memory. It supported multiple processors and virtual memory, and was programmed in Algol and support languages like COBOL and FORTRAN. The operating system was the Burroughs Master Control Program (MCP).
Using software like large, in-memory databases makes more sense when main memory isn't volatile. These days, main memory tends to be volatile, dynamic RAM (DRAM).
Bubble memory, like Intel’s 7110, also took advantage of the magnetic properties of the underlying material. The system could move magnetic domains known as bubbles using an external magnetic field. It worked more like a shift register or delay-line memory of the past requiring a refresh process that differs from how you refresh DRAM although the basic purpose is the same since data is lost if the process is incomplete or is stopped.
Magnetic storage lives on in hard disk drives and tape drives, although the approach is much different than magnetic cores. Drives and tapes require movement of the magnetic material past read/write heads that change the polarity of magnetic regions. Like core memory, data is maintained once it's written.
Now: Flash Memory Dominates Non-Volatile Storage
Capacity and speeds in data centers are massive compared to core storage decades ago. Cloud data centers measure storage in petabytes and transfer rates are in GB/s or higher.
On the embedded side, the JEDEC UFS 4.1 standard supports 4.2 GB/s. KIOXIA’s UFS 4.1 flash-memory device is a single-chip example that targets automotive applications and supports up to 1 TB of BiCS NAND flash (Fig. 4).
Magnetics remain a primary electronics technology, but no longer for storage purposes. Still, it's interesting to look back to see the origins of our technology.