Abstract: One dynamic task scheduling method includes: receiving a task, wherein the task comprises a kernel and a plurality of data items to be processed by the kernel; dynamically partitioning the task into a plurality of sub-tasks, each having the kernel and a variable-sized portion of the data items; and dispatching the sub-tasks to a plurality of computing devices of a heterogeneous computing system. Another dynamic task scheduling method includes: receiving a task, wherein the task comprises a kernel and a plurality of data items to be processed by the kernel; partitioning the task into a plurality of sub-tasks, each having the kernel and a same fixed-sized portion of the data items; and dynamically dispatching the sub-tasks to a plurality of computing devices of a heterogeneous computing system.

Abstract: One dynamic task scheduling method includes: receiving a task, wherein the task comprises a kernel and a plurality of data items to be processed by the kernel; dynamically partitioning the task into a plurality of sub-tasks, each having the kernel and a variable-sized portion of the data items; and dispatching the sub-tasks to a plurality of computing devices of a heterogeneous computing system. Another dynamic task scheduling method includes: receiving a task, wherein the task comprises a kernel and a plurality of data items to be processed by the kernel; partitioning the task into a plurality of sub-tasks, each having the kernel and a same fixed-sized portion of the data items; and dynamically dispatching the sub-tasks to a plurality of computing devices of a heterogeneous computing system.

Abstract: An operation method of a memory includes the steps of calculating an offset of sequential write commands and the beginning of pages of a block of a non-volatile memory; shifting the block by the offset; and directly writing data from a host to the pages except the first and last pages of the block by the sequential write commands. In an embodiment, the pages are logical pages providing optimal writing efficiency and are determined before calculating the offset. The step of shifting the block by the offset is to increase corresponding logical block addresses (LBA) in the pages by the offset.

Abstract: A wear leveling limit and/or an overall erase count threshold used for activating wear leveling in a non-volatile memory may be adjusted by determining a stage according to a highest erase count, and determining the wear leveling limit and/or the overall erase count threshold corresponding to the stage. Wear leveling may then be performed according to the wear leveling limit and/or the overall erase count threshold.

Abstract: A flash memory system includes a path selector to determine to write to a non-volatile memory, a volatile memory or both the non-volatile memory and the volatile memory when the flash memory system is to write data. A record is stored in the non-volatile memory which is updated the status of the non-volatile memory after each one or more writing operations. When the flash memory system is powered on after a power loss, it could be recovered to a command executed prior to the power loss or to any checkpoint prior to the power loss by using the record.

Abstract: The block groups of a multiple data channel flash memory storage device are detected for defective blocks. The block group containing any defective blocks is divided into subgroups, each of which contains only defective blocks or only good blocks. The subgroups containing only good blocks are selected to establish a new block group having the same amount of blocks as that of the original block groups.

Abstract: The block groups of a multiple data channel flash memory storage device are detected for defective blocks. The block group containing any defective blocks is divided into subgroups, each of which contains only defective blocks or only good blocks. The subgroups containing only good blocks are selected to establish a new block group having the same amount of blocks as that of the original block groups.

Abstract: A wear leveling limit and/or an overall erase count threshold used for activating wear leveling in a non-volatile memory may be adjusted by determining a stage according to a highest erase count, and determining the wear leveling limit and/or the overall erase count threshold corresponding to the stage. Wear leveling may then be performed according to the wear leveling limit and/or the overall erase count threshold.

Abstract: An operation method of a memory includes the steps of calculating an offset of sequential write commands and the beginning of pages of a block of a non-volatile memory; shifting the block by the offset; and directly writing data from a host to the pages except the first and last pages of the block by the sequential write commands. In an embodiment, the pages are logical pages providing optimal writing efficiency and are determined before calculating the offset. The step of shifting the block by the offset is to increase corresponding logical block addresses (LBA) in the pages by the offset.

Abstract: A flash memory system includes a path selector to determine to write to a non-volatile memory, a volatile memory or both the non-volatile memory and the volatile memory when the flash memory system is to write data. A record is stored in the non-volatile memory which is updated the status of the non-volatile memory after each one or more writing operations. When the flash memory system is powered on after a power loss, it could be recovered to a command executed prior to the power loss or to any checkpoint prior to the power loss by using the record.