公开(公告)号：GB2507049A

公开(公告)日：2014-04-23

申请号：GB201218550

申请日：2012-10-16

Applicant: IBM

Inventor： WONG KELVIN ,  PALMER MICHAEL JOHN

IPC: G01R31/3185

Abstract: A logic circuit comprises a plurality of functional logic units (20) each having an independent clock signal (CLK1, CLK2) and each having a trace bus (41,42) for carrying trace data. A trace gathering logic unit (30) collects trace data from the functional logic units. The trace gathering logic unit has a clock signal (CLK_TRC) independent of the clock signals of the functional logic units and a trace bus {71,72) for receiving trace data from the functional logic units. Multiplexors {45,46) can select portions of trace data from different functional logic units. Synchronization logic is coupled to the multiplexing logic, and includes latches 84, 85, 97, 98 and detection logic (103) to communicate data based on CLK_TRC. When valid synchronized data is available at registers 110 and 111 for placing on the bus of the trace gathering logic unit, logic (112, 113) signals this to the trace gathering unit.

公开(公告)号：GB2507961A

公开(公告)日：2014-05-21

申请号：GB201220442

申请日：2012-11-14

Applicant: IBM

Inventor： WONG KELVIN ,  PALMER MIKE ,  SMITH PETE

IPC: G06F11/14 ,  G06F11/07

Abstract: Â A method and controller system are provided for back-up of data to flash memory. The method comprises organising backed-up data into stripes (where a stripe 830 is a set of pages across all available flash memory devices which have the same block and page address), and when an error 821 in a block of flash memory is encountered during back-up of a stripe of data, re-writing the stripe 840 of data starting at the next available page address and excluding the page of flash memory for the block having the error, and writing subsequent stripes of data in the block to pages excluding the page 850 of flash memory for the block having the error. The method may also include embedding metadata for a stripe of backed-up data, including an identification number of the stripe of data, a stripe length field of the amount of data saved in a stripe, an indication of the location of a grown bad block, and an indication of known bad blocks. The method is particularly suited to the back-up of DRAM when a main power failure occurs.

公开(公告)号：GB2517435A

公开(公告)日：2015-02-25

申请号：GB201314810

申请日：2013-08-19

Applicant: IBM

Inventor： PALMER MICHAEL JOHN ,  WONG KELVIN

IPC: G06F11/10 ,  G06F12/02

Abstract: Memory controller 1000 backs-up stripes of data to a flash memory system. The system comprises channels, e.g. flash devices, divided into dies, each having two planes, where each plane includes blocks, and where each block comprises a number of pages for data payload. Stripes are sets of pages across all available channels, dies, and planes having the same block and page addresses. Each stripe further comprises first metadata indicating known bad blocks. Grown bad block component 1034 determines when an error occurs in a block of flash memory during back-up. Component 1034 rewrites the stripes to exclude block with errors and embeds second metadata with locations of newly discovered bad blocks. After checking first and second metadata, component 1034 writes to more planes simultaneously if no bad bock indication is given. By flagging new bad-block locations, time is not wasted erasing blocks when new bad-blocks are encountered at back-up.