International Business Machines Corporation

Harish Venkataraman

Saravanan Sethuraman

Gary A Tressler

Embodiments herein describe a memory system that queues program requests to a block of flash memory until a predefined threshold is reached. That is, instead of performing program requests to write data into the block as the requests are received, the memory system queues the requests until the threshold is satisfied. Once the buffer for the block includes the threshold amount of program requests, the memory system performs the stored requests. In one embodiment, the memory system erases all the pages in the block before writing the new data in the program requests into the destination pages. The data that was originally stored in the pages that are not destination pages is rewritten into the pages. In this example, the queued program requests can be written into the pages using one erase and write step rather than individual erase and write steps for each of the requests.

Delay programming requests in flash memory

Embodiments herein describe a 3D flash memory system that includes multiple blocks where each block contains multiple pages arranged in a vertical stack. Instead of having a single command line indicating whether a read or program is to be performed, separate command lines are coupled to each of the blocks. As a result, if the memory system identifies a read request and a program request to different blocks, the requests can be performed in parallel. In one embodiment, a program command line is used to perform a program request on a first block while a read command line is used to perform a read request on a second block in the 3D flash memory system in parallel. Furthermore, because a program request can take much longer to complete than a read request, the 3D flash memory system can perform multiple read requests in parallel with the program request.

Growing community of inventors

Bengaluru, India

Harish Venkataraman