Static information storage and retrieval – Addressing – Plural blocks or banks
Reexamination Certificate
2000-08-24
2002-04-23
Zarabian, A. (Department: 2824)
Static information storage and retrieval
Addressing
Plural blocks or banks
C365S233100, C365S063000
Reexamination Certificate
active
06377503
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to a synchronous Dynamic Random Access Memory (DRAM) for burst read/write operations.
2. Description of the Prior Art
FIG. 1
shows a conventional dynamic RAM (DRAM) with a conventional basic architectural configuration.
FIG. 2
shows a detailed drawing of the conventional DRAM shown in FIG.
1
.
In the conventional basic architectural configuration of the dynamic RAM (DRAM), as shown in
FIG. 1
, data read out of a memory cell selected by a word line is transferred to a sense amplifier (S/A) via a bit line.
A pair of data items amplified by the S/A are read out to an output buffer
104
through a pair of FETs
101
(shown in
FIG. 2
) through which the pair of data items are controlled by a signal on a column select line CSL.
In the conventional basic architectural configuration of the DRAM shown in
FIGS. 1 and 2
, we will describe one of architectural configurations of a conventional synchronous DRAM (SDRAM) below.
FIG. 3
shows a path of the synchronous data read/write operations for the input and output of one unit of data. These operations will now be briefly explained.
During the output of one string of serial data, when the head address of the data in the string is provided, two adjacent CSLs corresponding to column select lines CSL
1
and CSL
2
are selected, and four items of data from memory cells are read out through four pairs of DB lines. This is a 2-bit prefetch system whereby data read out of two columns within two clock cycles simultaneously is transferred serially, and two pairs of DB lines are selected to coincide with serial access addressing from the four pairs of DB lines. This selection is performed by a DB selector. The data on the two pairs of selected DB lines is transferred to two pairs of RWD lines RWD
1
and RWD
2
. Data in the first two cycles on the two pairs of RWD lines are stored into registers R
1
and R
2
, and data in the next two cycles are stored into registers R
3
and R
4
.
In this write operation to the registers R
1
to R
4
, the sequence for storing the data from the RWD lines RWD
1
and RWD
2
in the registers R
1
to R
4
is determined by RWD switches RWDS
1
and RWDS
2
.
The data passing through these switches RWDS
1
and RWDS
2
is stored in access sequence into the registers R
1
to R
4
by register transfer gates RTG
1
and RTG
2
which open alternately every two cycles to provide high speed data output.
The RWD switches
1
,
2
and the register transfer gates RTG
1
and RTG
2
, as shown in
FIG. 3
, are made up of gates of FETs. The data stored in the registers R
1
, R
4
, for example, as shown in
FIG. 4
, is read out to the output buffer
104
.
FIG. 5
shows a timing chart of the data transfer state in this data read operation described above. In
FIG. 5
, the data transfer state is illustrated under the condition that the burst length is 8 and the number of latency is 3 counted after address is determined or latched.
In
FIG. 5
, the operational state of each of the configurational elements shown in
FIG. 3
is illustrated. These will now be explained in order.
First, in a clock cycle (CLK), a Column Address Strobe (/CAS) is switched from high to the low, the head address of one string of burst data is set, and access is commenced. After the head address is determined, according to the addressing sequence of the burst data access, an internal address is produced for every two cycles and an access operation is carried out at the rise of levels of every two column select lines CSL.
When the column select line CSL rises, the DB line pair immediately enters to a busy state. When the data has been kept satisfactorily on the DB line pair, using the DB selector, data from two pairs in four-pair DB lines is transferred to the RWD line pair, and the RWD lines enter to the busy state every two cycles.
When data is kept sufficiently on the RWD lines, the data is stored into the register by the operation of one of the register transfer gates RTG
1
, RTG
2
and one of the RWD switches RWD
1
and RWD
2
.
In this data store operation, the RWD switches
1
or
2
are suitably selected by addressing for the burst data and turned ON, normally the register transfer gates
1
and
2
are alternately ON, and the data is stored in the register.
When the respective register transfer gates RTG
1
and RTG
2
are turned ON, the contents of the register are immediately rewritten and data is transferred serially from an OUTPUT which enters the busy state.
While these burst data transfer are controlled, after the access for the burst data transfer is completed, the clock cycle for commencing a new burst transfer access is restricted because the internal operation is operated in two clock cycles. In other words, a time restriction is produced so that a new access is not commenced from an optional cycle after the burst data transfer is completed. When a new burst data transfer access is commenced from an optional cycle after the previous burst data transfer is completed, it is necessary to temporarily reset the control of the clock period and commence the new burst data transfer after two clock cycles.
For this reason, a data burst completion signal is generated internally at a time when the burst data transfer access is completed and when it becomes unnecessary to control the burst data transfer access. The control system is reset from the clock cycle in which the data burst completion signal is generated. This clock cycle is designated by the reference number CLK
9
shown in FIG.
5
.
Because if the reset is not completed it is not possible to commence a new burst data transfer cycle and a time period of several tens of ns is required for the reset, the setting of a new starting address for a new burst data transfer occurs from a clock cycle
11
. For this reason, it is not possible to set a new burst access in clock cycles CLK
9
and CLK
10
. Accordingly, the output of a new burst data transfer is not possible after the thick dotted line in
FIG. 5
, so that data output of the new burst data transfer is only possible after the thin dotted line, which is disadvantageous in high speed burst data transfer.
As can be seen from the foregoing description, the reset operation described above is required in a conventional synchronous DRAM during the transfer for a burst data string. Because this reset operation takes a comparatively long time, it is very troublesome to transfer burst data continuously at high speed.
In addition, in a conventional synchronous DRAM, the data transfer system for cell arrays of multibank architectural configuration is not arranged in an optimum manner, necessitating an increase in the area of the chip.
SUMMARY OF THE INVENTION
Accordingly, an object of the present invention is, with due consideration to the drawbacks of such conventional synchronous DRAM, to provide a synchronous DRAM wherein the arrangement of cell arrays with a multibank architectural configuration and a data transfer system is optimized and/or high speed burst data transfer is achieved.
In accordance with one aspect of the present invention, there is provided a synchronous Dynamic Random Access Memory (DRAM) comprising:
a plurality of cell arrays, each of which contains cells, arranged in rows and columns, and being divided into a plurality of banks which being accessed asynchronously; and
a plurality of n bit I/O buses for the simultaneous input and output of n bits of data among said plurality of cell arrays,
wherein said each bank is divided into m blocks (m is a positive integer) with a plurality of said cell arrays,
the n-bit I/O buses (n is a positive integer), located between adjacent banks, is used for time sharing between adjacent banks in common,
said n bit I/O buses, used for time sharing between adjacent banks in common, are grouped into n/m bit I/O buses, every n/m bits for each block of m blocks of bank, and
in each block in each bank, data input/output are carried out between said n/m-bit I/O buses and data bus lines in each bank.
In the synchronous DRAM descr
Hogan & Hartson L.L.P.
Zarabian A.
LandOfFree
Synchronous dynamic random access memory does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Synchronous dynamic random access memory, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Synchronous dynamic random access memory will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2875484