MAGRAM™ is a ferromagnetic based digital memory technology which combines the best attributes of all existing memory technologies and offers a total memory solution.

Potentially replacing all conventional memories (varieties of RAM, ROM and magnetic media) with a unified solid-state system that does not require refresh, retains data indefinitely after power off (i.e., is non-volatile) and delivers nanosecond random read/write capabilities, MAGRAM promises to radically re-define computer architectural and operational paradigms.

Magram memory combines all the positive characteristics of other memory technologies without their drawbacks. In particular MAGRAM offers:

There is increasing pressure in the electronics industry in general, specifically the data processing industry, to develop a "true" NVRAM: a random access memory that retains digital data without the use of primary or backup power. Matters concerning security, economy and convenience will not be fully addressed until a viable NVRAM exists in the commercial market. This holy grail of memory technology needs to address a variety of growing concerns including protection against inadvertent power failure, rising processor speeds, power consumption, transportability, shrinking product size and the move away from slow, failure-prone mechanical data storage devices.

Today there exists no single true and viable NVRAM technology on the market (so-called NVRAMs are primarily standard RAMs with battery back up). Non-volatility, speed and random accessibility of memory data are the key criteria in computer memory, but these factors have been more or less mutually exclusive. Therefore, most computers use three main types of memory devices which collectively possess the required attributes: Read-Only Memory (ROM), mechanical magnetic media such as hard and floppy disks and Dynamic Random Access Memory (DRAM).

Today’s focal point for computer memory is DRAM. The "D" stands for dynamic and represents the fact that DRAM requires refresh. This means that DRAM can only remember stored data for a short period of time, typically between 4 and 30 milliseconds (thousandths of a second). Thus, before the data in DRAM fades completely, it must be restored, a process which places a heavy demand of time and power resources on a computer.

The upside of DRAM is that it is dense, fast and relatively inexpensive. Its density is stated in terms of the number of bits and bytes in a given area on a chip, and is typically very high. DRAM is fast: write and read times are considerably below 100 nanoseconds, a factor vital for the CPU (Central Processor Unit) which runs at very high clock speeds (desktop PCs are at 450 million cycles per second and climbing) and relies on DRAM for instructions and immediate data storage.

One of the major downsides to DRAM, however, is that it loses data instantly upon power off (volatility). Because DRAM cannot remember past power off, there are two other major forms of data storage used in computers to make up for this deficiency. These are ROMs and disk drives -- both diskettes and hard or fixed disks.

ROM

ROMs, which are integrated circuit (IC) devices, are non-volatile but fast, random accessibility is restricted to read operations only. ROM data, depending upon type and sophistication, is usually written into the device prior to installation, and cannot be changed thereafter without special considerations.

Since writing new data onto ROM is slow and relatively difficult, once information is stored in ROM, it is seldom changed. For this reason, ROM is used principally to permanently store data required during the first moments after system power-on (BIOS). This is program data required by the CPU to initiate computer activities generally referred to as boot routines. These routines usually test the computer, then point the CPU toward program data stored on DRAM or other memory within the computer.

Disk Drives (diskettes and hard disks)

Disks, including diskettes and hard drives, take over the data source task from ROM subsequent to the boot-up or initialization phase of computer power-up. These disks are non-volatile and are capable of storing large amounts of data. The CPU refers to them to transfer programs and data into DRAM for upcoming work sessions where faster read/write functionality is necessary.

The principal drawbacks of disks are that they are slow (millisecond vs. nanosecond speeds of the solid-state devices), bulky and have certain environmental restrictions to their use. Disks are slow for two reasons: one, they are mechanical (that is, the medium and the write/read head or heads move relative to each other); two, they are mainly written and read in a serial fashion (i.e., they are non-random). Data is written and read bit-by-bit rather than byte-by-byte, although some state of the art drives are configured as multi-track, allowing for byte-by-byte read/write. Disks, especially hard drives, also must be kept relatively stable during use and become unstable at high altitudes. Hard drives have a "mean time before failure" (MTBF) figure associated with them: they are expected to fail at some point in the foreseeable future. In some computer systems, identical data is stored in more than one drive to protect against this inevitable occurrence.

Static Random Access Memories (SRAM) are high-speed, low density write/read IC memory devices. Although volatile, they do not require refresh (thus static). They are very useful in so-called cache memory, CPU-close RAM containing repetitive operational data. Cache memory has become de rigeur in modern, high-speed systems because CPU speeds are so high that DRAM is beginning to look comparatively slow.

The latest generational step with ROMs is flash memory. Flash memory is dense and features DRAM-fast read speeds. Although it can be written to after installation (i.e., without removal from the system), it is a slow, cumbersome process, and therefore lacks a principal requirement to make it a "true" NVRAM: random write. Flash memory is also expensive, and exhibits a problem termed "write fatigue;" that is, after so many operations, it’s usable life declines measurably.

In the past few years, the most encouraging prospects for NVRAMs have been those based on the ferroelectric principle. NVRAMs using this technology, however, suffer from write fatigue problems. Over time and with use, the ferroelectric crystals tend to break down, losing their ability to store data. Manufacturers of these devices use such terms as durability and remanence (retentivity) factors. For the most part, devices made using this technology have been relegated to DRAM, SRAM, register or cache (interim RAM) and data back-up at power-off to reduce the number of aging cycles while capturing essential data held in RAM prior to power off.

There are several versions of ferromagnetic memories, exclusive of hard drives and diskettes. They are solid state to the extent that there are no moving parts. Core memories, plated wire and bubble are three examples. Core is bulky, power-hungry, tends to run hot and reads data using destructive readout, which requires refreshing the data after each read access. This process requires the use of extra computer clock cycles, since the data destroyed at read time must be restored, much like refresh in DRAMs, thus limiting access by the CPU. Bubble and plated wire technologies are slow and use a serial write/read scheme.

Note: There is no relationship between the ferromagnetic MAGRAM and ferroelectric memories, whose physics rely on the movement of an atom within a crystal, and whose input and output is essentially electrostatic, as in DRAM. It is these crystals that tend to break down over time, and thus suffer from long-range reliability problems. MAGRAM, on the other hand, is a purely solid state technology wherein individual data bits, or groups of bits (bytes) can be accessed randomly, at high speed (nanosecond access times), without destructive readout or the need for refresh. The disadvantages make these other memory technologies inappropriate for many applications due to their slow speed, volatility and vulnerability to power outages and mechanical failures.

There are other NVRAM technologies, outside of MAGRAM, which, as of this writing, have not been marketed, as ferroelectric devices have. These others depend on three basic physical principles: spin-valve, GMR (Giant Magneto-Resistance) and phase-change. In each case, production difficulties, cost, fatigue problems, speed and power consumption limit their viability, especially in the face of the simplicity and other favorable factors contained in the MAGRAM technology.

MAGRAM combines all the desirable features these other memories possess, including non-volatility, speed, random read and write, all while retaining the capability to match DRAM-like density.

In a computer system designed around such a comprehensive technology, only one type of memory would be needed. ROM, DRAM and hard drives would inevitably be eliminated. A system centered around MAGRAM would be considerably faster, safer for data retention and rely on less software overhead (for disk I/O operations).

Power-off sequences, whether planned or emergent, could be graceful, in that current register contents would be literally saved to RAM. The operating system could be resident in RAM, rather than having to be re-loaded from the hard drive at power-on time. After power is withdrawn and then restored, it would be possible for the exact program position to be restored, and operations to continue where ceased.

MAGRAM is a true NVRAM technology and can potentially replace virtually all memory devices outside of those used for archival storage (diskette, mass storage hard drive, tape, etc.).

SRAM can be replaced by MAGRAM because of the clearly defined advantages of non-volatility and density.

Since ROM is a read-only memory, MAGRAM devices are a logical replacement choice. Unlike ROM, MAGRAM can be randomly written to after installation and at DRAM speed, meaning that boot and fixed routines could be upgraded easily and at will.

DRAM is an obvious target for replacement by MAGRAM. Since MAGRAM is non-volatile and competitive in the areas of density and speed, CPU-close MAGRAM would have the advantage of increased speed and accessibility without refresh, as well as being able to retain processed data and programs after power loss. In a disastrous power-off scenario, CPU register information would be stored securely in the MAGRAM-based RAM, providing for a smooth power-on and easy transition back to work in progress.

In theory, diskettes (used typically as portable back-up media) could also be replaced by a diskette-sized MAGRAM device, although this is not expected to be a near-term target.

Hard drives, the mass-storage devices now most widely used to maintain program and process data, can also be replaced by MAGRAM devices. Since MAGRAM is expected to resemble DRAM in speed and density, time to access mass storage data, whether in write or read mode, would be reduced by a factor in the range of 100,000.

With mass storage operating the same as RAM, RAM itself could then become a part of mass storage, or vice-versa. Further, the need to allocate memory for operating systems and other such programs is eliminated, since they would be accessed and run directly from a single memory system.

Thus, MAGRAM takes the place of all three standard digital memory devices in a typical computer.

MAGRAM is unique in its applicability to multiple markets and has virtually unlimited applications. It can be implemented as a sub-component, a component or as a stand-alone system:

Today, there are in the range of over 200 million digital cell phones. Virtually every cell phone contains flash memory ICs, which consume considerable power. Cell phone manufacturers strive for the longest life per battery charge, and since MAGRAM consumes less power and does not have write-cycle limitations (fatigue), it is therefore a better solution.

Small, handheld devices such as 2-way pagers, palm PCs and organizers do not have disk memory to store information. They typically use flash memory to retain data. MAGRAM is a preferred alternative due to the high cost and inability to perform random read/write functions of flash memory.

Application Specific Integrated Circuits (ASICs):

Many devices use ASICs (custom-designed ICs) to replace a handful of components and thus reduce size, weight, cost and power consumption. The auto industry is a major ASIC user. These ASICs have many advantages; however, one disadvantage is that they typically require long development cycles. Additionally, if an ASIC has a design flaw – even a small one – they must be modified, and that process may add 2-6 months to the design cycle. Because time-to-market is critically important, designers would greatly benefit from the ability to incorporate MAGRAM into their ASIC designs, allowing for changes without redesign.

Numerous household appliances contain memory devices to store simple instructions. MAGRAM provides a low-cost, non-volatile memory to store functions and user preferences that do not need to be reprogrammed after power is turned off or lost, then restored. Examples are digital clocks, VCRs, microwaves, answering machines and calculators. In fact, anything electrical that has a digital memory can benefit from MAGRAM.