TL;DR: An original concept of thermally assisted MRAM in which memory and logic functions are combined in the same stack, which is particularly suited for security applications and Content Addressable Memories.
Abstract: This paper describes an original concept of thermally assisted MRAM in which memory and logic functions are combined in the same stack. The memory cell is represented by a magnetic tunnel junction having an exchange biased storage layer and a soft reference layer (called sense layer), replacing the conventional pinned reference layer. The write of the storage layer is ensured by a combination of heating pulses and magnetic stray fields created by the soft reference layer. The read is performed in a self-referenced manner by measuring the resistance variation associated to a field variation. This makes these memories much more tolerant to cell to cell variability. In addition, this stack intrinsically performs both a storage function and a comparison (XOR) function. This device called a Magnetic Logic Unit (MLU™) is particularly suited for security applications and Content Addressable Memories.
TL;DR: In this article, a near out-of-memory condition in a memory space is detected by creating softly reachable objects which are garbage collected when the memory space becomes full, which can be cleared at the discretion of the garbage collector when heap memory is running low.
Abstract: A near out-of-memory condition in a memory space is detected by creating softly reachable objects which are garbage collected when the memory space is becoming full. The softly reachable objects are objects that can be cleared at the discretion of the garbage collector when heap memory is running low. An agent process of an application can create soft reference objects which reference the softly reachable objects, and periodically poll the soft reference objects to determine if the softly reachable objects have been cleared. If they have been cleared, the agent reports to the application so that a graceful shutdown of the application can be initiated. A report can also be sent to a user interface or other output device. Additional information regarding the memory space can be gained by using softly reachable objects of different sizes and/or ages. Further, a wait period for the polling can be set adaptively.
TL;DR: An exemplary magnetic memory cell comprises a data layer, a soft reference layer having a lower magnetic energy than the data layer and a spacer layer between the data and the soft reference layers as discussed by the authors.
Abstract: An exemplary magnetic memory cell comprises a data layer, a soft reference layer having a lower magnetic energy than the data layer, and spacer layer between the data layer and the soft reference layer.
TL;DR: In this paper, a garbage collector determines a target amount of heap space to deallocate, estimates an amount of space reachable by a plurality of soft references by determining a cumulative size of no more than an exploration bound N number of objects reachable from each soft reference, and deallocates heap space based on the target amount and the estimate of the heap space reachedable from the soft references.
Abstract: A garbage collector determines a target amount of heap space to deallocate, estimates an amount of heap space reachable by a plurality of soft references by determining a cumulative size of no more than an exploration bound N number of objects reachable from each soft reference, and deallocates heap space based on the target amount and the estimate of the heap space reachable from the soft references. Deallocating heap space may include clearing at least one soft reference. If the estimate is inaccurate, it may be utilized regardless or modified to account for inaccuracy. The least-recently-used or the largest soft reference may be cleared until the total cleared space reachable exceeds the target amount. By performing a bounded analysis, the garbage collector may be able to make a more informed decision about whether to clear a soft reference without consuming the full amount of resources consumed by an exhaustive analysis.
TL;DR: In this article, an image cache is realized through memory cache and SD(Secure Digital) card cache, where the memory cache comprises two levels of cache, and the SD card cache comprises a layer of cache.
Abstract: The invention discloses an image data caching method. The image cache is realized through memory cache and SD(Secure Digital) card cache. The memory cache comprises two levels of cache, and the SD card cache comprises a layer of cache. The specific realization process is that firstly a caching space is opened up in the memory, the cache stores the data through a strong reference type, the caching space is an eighth of the memory allocated for each application; a second-level caching space is opened up in the memory, so as to store the cache removed from the first-layer memory through the soft reference type; then a catalogue is established in the SD card to store the image data obtained through network request; and finally the data are read. Compared with the prior art, the image data caching method has the advantages of realizing a caching method including the memory cache and the SD card cache through LRU(Least Recently Used) algorithm, being capable of conveniently and rapidly showing the cached image in UI(User Interface) interface, improving user experience, saving flow and strong practicability.