Static information storage and retrieval – Systems using particular element – Magnetic thin film
Utility Patent
1998-01-12
2001-01-02
Dinh, Son T. (Department: 2824)
Static information storage and retrieval
Systems using particular element
Magnetic thin film
C365S009000, C365S158000
Utility Patent
active
06169687
ABSTRACT:
FIELD OF THE INVENTION
This invention relates generally to high speed memories constructed from magneto-electronic devices. Such devices can include magnetic spin transistors, hybrid hall effect devices, etc., and can be constructed in a variety of different ways, including as elements integrated in silicon with conventional semiconductor elements. The new magneto-electronic devices can be implemented as replacements for other conventional memories, including semiconductor random access SRAM, DRAM, Flash, as well as intermediate storage magnetic or optical memory types such as magnetic disk drives, tape drives, optical drives, etc.
BACKGROUND OF THE INVENTION
The breakdown of conventional memory in contemporary computing machines is depicted in FIG.
1
. As can be seen there, memory coupled to a processing unit
100
is generally broken up into four classifications: (1) cache memory
105
; (2) system memory (primary storage)
110
; (3) intermediate (secondary) storage
115
; and (4) archival (long term) storage
120
.
Each of these types has unique characteristics, requirements and associated costs. The arrow on the left hand side of the figure is intended to illustrate a rough comparison of the speed and cost associated with such systems, and the various blocks shown for cache, system, intermediate and long term are also roughly scaled in size in the diagram to denote the concept of relative capacity differences in such systems as well.
For example, cache memory
105
usually resides onboard a microprocessor (or similar computing device), is extremely fast (less than 10 ns access time), expensive, and volatile. Typically, SRAM technology has been used to date for such implementations. This type of memory is usually used for storing data and instructions that are most frequently used by the system to accelerate execution speeds of operating system/applications programs. An example of this is a software loop comprised of multiple instructions which needs to be executed frequently. To avoid having to re-load this entire routine every time it is needed, it is common to cache such routines directly onboard the microprocessor. To a large extent caches make use of the spatial temporalism inherent in programs; i.e., that if an instruction is recently executed, that same instruction, or another instruction close to it (from a logical address space perspective) is likely to be needed soon as well. For this reason, it is common to pre-fetch and cache a number of instructions and related data in anticipation of their later use by the processor.
Main or system memory
110
(sometimes called primary storage as well) typically refers to memory implemented outside the microprocessor chip, and which is shared with other devices, such as other processors. The general functional constraints on this memory are related to the fact that it need only be fast enough to feed program instructions to a computing device (such as a microprocessor, RISC processor) or data to peripheral devices at a rate specified by such devices and which is otherwise permissible on a system bus. In other words, instructions and data are usually loaded from the main memory system
110
directly through a system bus to registers, latches, etc. on the processing unit or peripheral device for later processing. System memory
110
is usually DRAM based technology, and while somewhat slower (30-40 ns access time) than cache, is much less expensive, resides off-chip, but is also unfortunately volatile. Generally speaking, the bulk of operating system and applications programs are run from system memory
110
, and logic onboard the computing device is generally responsible for moving data and instructions into the onboard cache
105
as appropriate. In rough terms, system memory
110
capacity is typically on the order of 10 to 100 times larger than the cache memory
105
used within a computing system.
Secondary storage
115
includes such items as floppy disks, hard drives, and assorted magnetic media board, which have access times in the millisecond range, but which also have the advantage of being non-volatile and relatively inexpensive compared to the above semiconductor memories from a dollars/megabyte perspective. The bulk of operating system and user application programs is stored with this type of media, and is only accessed when needed. Such storage is also accessible through the system bus, but the data and instructions here must first be loaded into main memory where they can be accessed at the necessary speed by the processing unit. Again, in rough terms, secondary storage
115
capacity can be tens, hundreds or thousands of times larger than the primary memory
110
used within a computing system. Generally speaking, high capacity and low cost are the most important aspect of secondary storage
115
, while speed is the most important characteristic of primary memory
110
and cache memory
105
. At present, magnetic hard disk is the leading technology used for secondary storage, but there are a number of electro-mechanical, tribological and electrical factors which are limiting the potential of such devices to achieve higher speeds and densities. For example, significantly challenging engineering problems exist in improving the density of recording as well as size, sensitivity and bandwidth of read heads in such devices.
Some inroads into intermediate non-volatile storage have been made with semiconductor Flash memories, but for the most part their speed, density and cost considerations have constrained them to limited environments.
Finally, archival memory
120
is also non-volatile, substantially slower and cheaper than secondary storage. This type of media includes CD ROMs, optical disks, magnetic tape, etc., and usually data and programs that are only rarely needed, or which need to be stored long term, are loaded here. Capacity and cost are the primary considerations in such systems, and they can be many times larger than a secondary storage system.
All of these types of memory can co-exist within a single computing system, with each performing its role in accordance roughly with the discussion above. However, each type of memory system is made with completely different materials and manufacturing techniques. A hard drive, for example, includes a complex combination of interacting electrical and moving mechanical components, while a DRAM has no moving parts and is manufactured entirely using conventional semiconductor processing techniques. Furthermore, each technology has associated controller requirements, and must be connected to “other” memory systems through a bus and often using complicated interface technologies. Thus, within any conventional system, there are not only four different types of memory technology, but also four different types of controllers and four different types of data transfer interfaces all of which attempt to cooperate on a single bus. The manufacturers of separate memory technologies generally do not coordinate with each other, because, to date, there is no available technology that bridges their respective operating domains.
It is apparent, nevertheless, that it would be extremely advantageous to impart non-volatility to DRAM and similarly to impart random access capability and higher operations speed to magnetic disk technology. For example, when beginning operation of a machine, users of conventional computing systems typically must endure several minutes of “boot-up” sequences during which time necessary drivers, operating system procedures, etc. are transferred or loaded from magnetic disk to system memory where they can be executed. Furthermore, anytime an application program or data not already in system memory is required, another seek and load must be performed on the magnetic disk.
Along the same lines, it would be extremely beneficial if the cost and complexity associated with four different types of controllers and interfaces could be eliminated.
There is a tremendous unfulfilled need, therefore, for a simpler, faster, unified system memory technology that combines th
Dinh Son T.
Law +
Nguyen Tuan T.
LandOfFree
High density and speed magneto-electronic memory for use in... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with High density and speed magneto-electronic memory for use in..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and High density and speed magneto-electronic memory for use in... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2557440