Drum memory was a magnetic data storage system invented by Gustav Tauschek in 1932 in Austria. Drums were widely used within the 1950s and into the 1960s as computer memory. Many early computer systems, known as drum computer systems or drum machines, used drum memory as the main working memory of the computer. Some drums have been additionally used as secondary storage as for example various IBM drum storage drives and the UNIVAC FASTRAND sequence of drums. Drums were displaced as major laptop memory by magnetic core memory, which offered a greater stability of measurement, speed, cost, reliability and potential for additional enhancements. Drums have been then replaced by exhausting disk drives for secondary storage, which had been both less expensive and supplied denser storage. A drum Memory Wave or drum storage unit contained a large steel cylinder, coated on the surface surface with a ferromagnetic recording materials. It might be thought of the precursor to the hard disk drive (HDD), however in the form of a drum (cylinder) rather than a flat disk.

In most designs, one or more rows of mounted learn-write heads ran alongside the long axis of the drum, one for each observe. The drum's controller simply chosen the correct head and waited for the info to look underneath it because the drum turned (rotational latency). Not all drum units have been designed with every track having its own head. Some, such because the English Electric DEUCE drum and the UNIVAC FASTRAND had a number of heads transferring a brief distance on the drum in contrast to trendy HDDs, which have one head per platter floor. In November 1953 Glenn E. Hagen of Logistics Research, Inc. published a paper disclosing "air floating" of magnetic heads in an experimental sheet metallic drum. Flying heads turned normal in drums and laborious disk drives. Magnetic drum models used as primary memory have been addressed by word. Drum items used as secondary storage were addressed by block. Several modes of block addressing were doable, relying on the device.

Blocks took up an entire observe and had been addressed by monitor. Tracks had been divided into fastened length sectors and addressing was by track and sectors. Blocks have been variable length, and blocks were addressed by monitor and document number. Blocks have been variable length with a key, and could be searched by key content. Some units have been divided into logical cylinders, and addressing by observe was truly logical cylinder and observe. The performance of a drum with one head per observe is comparable to that of a disk with one head per track and is set nearly fully by the rotational latency, Memory Wave whereas in an HDD with shifting heads its efficiency includes a rotational latency delay plus the time to place the pinnacle over the desired monitor (search time). Within the era when drums have been used as major working memory, programmers typically did optimum programming-the programmer-or the assembler, for example IBM’s Symbolic Optimal Meeting Program (Soap)-positioned instructions on the drum in such a way as to reduce the period of time wanted for the following instruction to rotate into place below the pinnacle.

They did this by timing how lengthy it might take after loading an instruction for the pc to be able to learn the next one, then inserting that instruction on the drum so that it would arrive below a head just in time. This methodology of timing-compensation, called the "skip factor" or "interleaving", was used for many years in storage Memory Wave Audio controllers. Tauschek's original drum memory (1932) had a capacity of about 500,000 bits (62.5 kilobytes). One of the earliest functioning computer systems to employ drum memory was the Atanasoff-Berry pc (1942). It stored 3,000 bits