IEEE Solid-State Circuits Magazine - Spring 2016 - 17

Drumming up Business

of memory. Memory, as an enabler
of computing solutions, has taken
computing along a surprising path
over the past decades.

A History of Memory
If we exclude papyrus, paper, scribes,
monks, Gutenberg, and their cellulosecentered media from consideration,
modern information storage may be
traced to the invention by Danish engineer Valdemar Poulsen of a magnetic
wire recording system in 1899. (His
patent for the device is reproduced in
Figure 1.) The tenet pioneered by this
invention-movement of a magnetic
medium past read and write (R/W)
heads-has proven to be long-lived,
with hard drives using this basic principle to this day.
In an improvement over the
wire-based system, Austrian Fritz
Pfleumer invented a magnetic tape
system for recording audio in 1927.
The paper tape was coated with
iron-oxide powder, using lacquer as
a bonding agent.
Magnetic tape was first used
to store computer data in 1951 on
the UNIVAC 1 computer (Figure 2).
This system stored 128 characters/
in on an actual metal tape 0.5-in
wide. Multiple R/W heads laid down
eight tracks across the width of the
tape, thus giving an areal density
of 25,600 characters/in2. The data
rate was 12,800 characters/s, with
a character consisting of six data
bits, one parity bit, and a timing
bit. After overhead was removed,
the data rate was actually closer to
7,200 characters/s.

Figure 1: Valdemar Poulsen's magnetic
wire recording patent. (U.S. Patent and
Trademark Office.)

Of course, one of the negative
characteristics of any tape system
is long latency. Balancing this out
is the high density offered by tape,
which has allowed tape to survive
even today. Just two years ago,
Sony introduced a tape cartridge
with an areal density of 148 Gb/in2
for a cartridge capacity of 185 TB.
Another benefit of magnetic tape is
endurance-with advertised endurance spans of 15-30 years, as long
as there are functional tape drives
remaining [2].

Magnetic drum memory was invented
by Gustav Tauschek in 1932 in Austria. However, in 1941 the AtanasoffBerry Computer used capacitive
drum memory as the machine's main
working memory (Figure 3). Called a
"regenerative capacitor memory," the
system used a pair of drums, each
containing 1,600 capacitors, with
connections to the capacitors covering the surface of the drum and
rotating at one r/s. The system gave
3,000 b of total usable memory. The
core of the computational engine was
synchronized directly to the drum.
Similar to another memory type that
would make its appearance some 25
years later, regenerative capacitor
memory required periodic refresh.
Magnetic drum memory was much
more popular than capacitive drum
memory. In many ways, magnetic
drum memories were a reconfiguration
of tape systems, albeit with much lower
total capacity. What these systems gave
up in capacity, however, they gained
in random-access performance. Similar to the Atanasoff-Berry Computer,
many systems using magnetic drum
memory used this as the main memory of the computer. With most drum
memories, rotational latency was a
performance-limiting factor, but careful data placement was used to minimize the impact. Most drum memories
had multiple heads that were electronically switched and, with careful

IEEE SOLID-STATE CIRCUITS MAGAZINE

S P R I N G 2 0 16

Table of Contents for the Digital Edition of IEEE Solid-State Circuits Magazine - Spring 2016