Method and structure for performing pipeline burst accesses in a semiconductor memory

• US 5,784,705 A
• Filed: 07/15/1996
• Issued: 07/21/1998
• Est. Priority Date: 07/15/1996
• Status: Expired due to Term

• Current Assignee: Invensas Corporation
• Original Assignee: Mosys Incorporated (Peraso, Inc.)
• Inventors: Leung, Wingyu
• Primary Examiner(s): Swann, Tod R.
• Assistant Examiner(s): Chow, Christopher S.
• Application Number:
  US08/679,873
• Time in Patent Office: 736 Days
• Field of Search: 395/496, 365/189.05, 365/233, 711/169
• US Class Current: 711/169
• CPC Class Codes: G11C 7/1018 Serial bit line access mode, e.g. using bit line address shift registers, bit line address counters, bit line burst counters
• CPC Class Codes: G11C 7/1033 using data registers of which only one stage is addressed for sequentially outputting data from a predetermined number of stages, e.g. nibble read-write mode
• CPC Class Codes: G11C 7/1039 using pipelining techniques, i.e. using latches between functional memory parts, e.g. row/column decoders, I/O buffers, sense amplifiers

Abstract

A method and structure for implementing pipeline burst read and write operations in a semiconductor memory having a memory cycle time substantially longer than its I/O data cycle-time. The memory system includes a read buffer which stores all data required for a read burst transaction. All read burst data is loaded from the memory to the read buffer at the beginning of each burst read access. The memory is then isolated from the read buffer and prepared to perform the next burst access. During this time, the read data values are provided to the I/O device from the read buffer. A double-buffering technique provides gap-less output data for consecutive pipeline-burst read transactions. The memory system uses a two-entry write buffer in a first in, first out manner for pipeline-burst write operations. Each write buffer entry stores data for an entire burst transaction and a corresponding address. The first entry stores the data and address for a current write transaction and the second entry stores the data and address for a previous write transaction. The first and second entries are isolated to allow the second entry to provide its data and address to the memory while the first entry stores its corresponding data and address. A high-bandwidth transfer rate between the second entry and the memory with time to prepare the memory for the next transaction.

Claims

1. A memory system comprising:

a memory array having a memory cycle time and an input/output (I/O) cycle time, wherein the memory cycle time is longer than the I/O cycle time; and

a pipeline-burst access circuit coupled to the memory array, wherein the pipeline-burst access circuit enables pipeline-burst read operations or pipeline-burst write operations to be performed with the same access latency across the entire address range of the memory array;

wherein the pipeline-burst access circuit comprises a read buffer to the memory array, wherein the read buffer stores a set of data values for an entire burst read sequence; and

wherein the read buffer comprises;

a first data storage circuit coupled to the memory array; and

a second data storage circuit coupled to the first data storage circuit,

wherein the first and second data storage circuits are cascaded to form a double buffer circuit; and

wherein the pipeline-burst access circuit further comprises a read buffer control circuit coupled to the read buffer, the read buffer control circuit generating a first control signal which causes the set of data values to be stored in the first data storage circuit during a first clock cycle.

2. The memory system of claim 1, wherein the read buffer control circuit further generates a second control signal which causes the set of data values to be stored in the second data storage circuit during the first clock cycle.

3. The memory control system of claim 2, wherein the second control signal causes second data storage circuit to be isolated from the first data storage circuit during a second clock cycle.

4. The memory system of claim 1, wherein the first data storage circuit comprises a plurality of data amplifiers.

5. The memory system of claim 1, wherein the second data storage circuit comprises a plurality of transparent latches.

6. The memory system of claim 1, wherein the read buffer further comprises a multiplexer coupled to the second data storage circuit.

7. The memory system of claim 6, further comprising a sequencer circuit coupled to the multiplexer, wherein the sequencer circuit generates control signals which cause the multiplexer to sequentially route the set of data values from the second data storage circuit.

8. The memory system of claim 1, wherein the memory cycle time is three times the I/O cycle time.

9. The memory system of claim 1, wherein the pipeline-burst access circuit comprises a write buffer coupled to the memory array, wherein the write buffer stores a set of data values for an entire burst write sequence.

10. The memory system of claim 9, wherein the write buffer comprises:

an input/output (I/O) device coupled to receive write data and write addresses;

a first data storage circuit coupled to receive write data from the I/O device;

a first address storage circuit coupled to receive write addresses from the I/O device;

a second data storage circuit coupled to the first data storage circuit, wherein the first and second data storage circuits are cascaded to form a first double buffer circuit;

a second address storage circuit coupled to the first address storage circuit, wherein the first and second address storage units are cascaded to form a second double buffer circuit; and

a tri-state buffer circuit coupled between the second data storage circuit and the memory array, the tri-state buffer circuit also being coupled between the second address storage circuit and the memory array.

11. The memory system of claim 10, wherein the pipeline-burst access circuit further comprises a write buffer control circuit coupled to the write buffer, the write buffer control circuit generating a first control signal which places the tri-state buffer circuit in a conductive state, thereby providing a first set of data values stored in the second data storage circuit and a first address value stored in the second address storage circuit to the memory array during a first clock cycle, the first control signal further causing a second address value to be stored in the first address storage circuit during the first clock cycle.

12. The memory system of claim 11, further comprising a sequencer circuit which generates a second control signal which is applied to the first data storage circuit to cause a second set of data values to be sequentially written to the first data storage circuit over a plurality of clock cycles.

13. The memory system of claim 12, wherein the write buffer control circuit further generates a third control signal which is applied to the second data storage circuit and the second address storage circuit, wherein the third control signal causes the second set of data values to be written to the second data storage circuit and causes the second address value to be written to the second address storage circuit during a second clock cycle.

14. The memory system of claim 13, wherein the first control signal causes second data storage circuit and the second address storage circuit to be isolated from the memory array during the second clock cycle.

15. The memory system of claim 9, wherein the write buffer comprises a two entry buffer configured to route information in a first in, first out (FIFO) manner.

16. The memory system of claim 15, wherein each entry of the two entry buffer comprises a data buffer for storing data values for a full burst write operation and an address buffer for storing the memory address of the data values in the data buffer.

17. The memory system of claim 15, wherein the write buffer further comprises a plurality of shift registers.

18. A memory system comprises:

a memory array having a memory cycle time and an input/output (I/O) cycle time, wherein the memory cycle time is longer than the I/O cycle time; and

a pipeline-burst access circuit coupled to the memory array, wherein the pipeline-burst access circuit enables pipeline-burst read operations or pipeline-burst write operations to be performed with the same access latency across the entire address range of the memory array;

wherein the pipeline-burst access circuit comprises a write buffer coupled to the memory array, wherein the write buffer stores a set of data values for an entire burst write sequence;

wherein the write buffer is controlled to transfer a plurality of data values to the memory array simultaneously.

19. A memory system comprising:

a dynamic random access memory (DRAM); and

a memory control sequencer for accessing the DRAM, wherein the memory control sequencer generates a row access signal and a column access signal for controlling the memory operations of the DRAM in response to a single external access signal and a clock signal;

wherein the memory control sequencer comprises circuitry for detecting the reception of the external access signal during a first cycle of the clock signal, asserting the row access signal during the first cycle and a consecutive second cycle of the clock signal, asserting the column access signal during the second cycle of the clock signal, de-asserting the row and column access signals during a consecutive third cycle of the clock signal, and precharging the DRAM during the third cycle of the clock signal.

20. A memory system comprises:

a dynamic random access memory (DRAM); and

a memory control sequencer for accessing the DRAM wherein, the memory control sequencer generates, a row access signal and a column access signal for controlling the memory operations of the DRAM in response to an external access signal and a clock signal;

wherein the memory control sequencer comprises circuitry for blocking the external access signal for three clock cycles upon detecting the external access signal.

21. A method of operating a memory array having a memory cycle time and an input/output (I/O) cycle time, wherein the memory cycle time is longer than the I/O cycle time, the method comprising the steps of:

initiating a read access to the memory array during a first clock cycle;

reading a first set of data values from the memory array during a second clock cycle;

storing the first set of data values read from the memory array in a first data storage circuit during the second clock cycle;

storing the first set of data values in a second data storage circuit during the second clock cycle, wherein the second data storage circuit is cascaded with the first data storage circuit; and

then

isolating the second data storage circuit from the first data storage circuit during a third clock cycle.

22. The method of claim 21, further comprising the step of precharging the memory array during the third clock cycle.

23. The method of claim 22, further comprising the steps of:

initiating a second read access to the memory array during a fourth clock cycle;

reading a second set of data values from the memory array during a fifth clock cycle;

storing the second set of data values read from the memory array in the first data storage circuit during the fifth clock cycle; and

reading data values of the first set of data values out of the second data storage circuit during the second, third, and fourth clock cycles.

24. The method of claim 23, further comprising the steps of:

reading a data value of the first set of data values out of the second data storage circuit during the fifth clock cycle;

storing the second set of data values in the second data storage circuit during the sixth clock cycle; and

then

isolating the second data storage circuit from the first data storage circuit during a seventh clock cycle.

25. The method of claim 21, further comprising the steps of:

performing a row access operation on the memory array during the first clock cycle;

performing a column access operation on the memory array during the second clock cycle, thereby reading the first set of data values from the memory array.

26. The method of claim 21, further comprising the steps of:

sequentially reading the first set of data values from the second data storage circuit; and

initiating a write access of the memory array one clock cycle after the first set of data values have been read from the second data storage circuit.

27. A method of operating a memory array having a memory cycle time and an input/output (I/O) cycle time, wherein the memory cycle time is longer than the I/O cycle time, the method comprising the steps of:

storing a first set of data values in a first data storage circuit during a first write transaction;

storing a first address corresponding to the first set of data values in a first address storage circuit during the first write transaction;

shifting the first set of data values from the first data storage circuit to a second data storage circuit during the first write transaction;

shifting the first address from the first address storage circuit to a second address storage circuit during the first write transaction;

providing the first set of data values and the first address to the memory array during a second write transaction, whereby the first set of data values are written to the memory array during a single clock cycle;

storing a second set of data values in the first data storage circuit during the second write transaction;

storing a second address corresponding to the second set of data values in the first address storage circuit during the second write transaction;

shifting the second set of data values from the first data storage circuit to a second data storage circuit after the first set of data values has been provided to the memory array; and

shifting the second address from the first address storage circuit to a second address storage circuit after the first address has been provided to the memory array.

28. The method of claim 27, further comprising the steps of:

initiating a read access of the memory array during the second write transaction; and

reading data values from the memory array one clock cycle after storing the second set of data values in the first data storage circuit.