Dynamic Random Access Memory Having Small Cycle Time Period

1. Field of the Invention

The present invention relates to a random access memory (RAM) which has a small cycle time period compatible with static RAM's which can carry out a so-called read-modified write operation.

2. Description of the Related Art

Recently, dynamic RAM's have come into increased use for storing video data. Such devices must allow a large amount of data to be read out and written therein per unit time, i.e., must offer a reduced cycle time period. Generally, in a dynamic RAM, each cycle time period includes one access time period for an actual read/write operation and one reset time period for precharging each portion of the device prior to each access operation. The reset time period is of the same order as an access time period. Therefore, each cycle time period of a dynamic RAM is relatively long.

In the prior art, a small cycle time period is available with a static column type dynamic RAM (F. Baba et al, "A 35 ns 64K Static Column DRAM", 1983 IEEE ISSCC Digest of Technical Papers, Article WPM6.5, pp.65-64, Febuary 1983). This dynamic RAM offers a cycle time period during a page mode about the same as its access time period. This dynamic RAM, however, includes static circuits. Therefore, there is a disadvantage of a rather large power consumption.

In another prior art dynamic RAM offering a small cycle time period (U.S. Pat. No. 4,376,989), each circuit thereof is automatically reset, i.e., precharged by the subsequent circuit thereof after the completion of its own operation. Therefore, its own reset time period is small, thereby reducing the cycle time period. In such a dynamic RAM, however, since all the circuits thereof are automatically reset, it is impossible to carry out a read-modified write operation. As a result, a dynamic RAM of this type is not compatible with conventional dynamic RAM's.

It is an object of the present invention to provide a dynamic RAM which has a small cycle time period and is compatible with conventional dynamic RAM's which can carry out a read-modified write operation.

It is another object of the present invention to provide a dynamic RAM having low power consumption.

According to the present invention, in a dynamic RAM wherein a memory access operation is started by receiving an external signal, such as a row address strobe signal RAS or a column address strobe signal CAS, address signals are applied from address buffers to address decoders, which include NOR gates for decoding the address signals and drivers for driving an internal circuit including word lines, memory cells, bit lines, and the like. The address buffers and the NOR gates are automatically reset earlier during an access operation defined by the strobe signal, while the drivers and their subsequent circuits are reset after the completion of the active period defined by the strobe signal. As a result, since the reset time period decreases, the cycle time period also decreases.

Also, since the internal circuit remains in a selected state even after the address buffers and the NOR gates are reset, it is possible to perform a write operation upon the internal circuit. Therefore, the dynamic RAM according to the present invention is compatible with conventional dynamic RAM's.

Further, the dynamic RAM according to the present invention has no static circuits, and therefore has a low power consumption.

The present invention will be more clearly understood from the description as set forth below with reference to the accompanying drawings, wherein:

FIG. 1 is a block diagram of a prior art dynamic RAM;

FIGS. 2A and 2B are circuit diagrams of the address decoders of FIG. 1;

FIG. 3 is a waveform diagram of the operation of the device of FIG. 1;

FIG. 4 is a waveform diagram of the operation of another prior art dynamic RAM;

FIG. 5 is a block diagram of an embodiment of the dynamic RAM according to the present invention;

FIG. 6 is a waveform diagram of the operation of the device of FIG. 5;

FIGS. 7A and 7B are circuit diagrams of the address decoders of FIG. 5; and

FIG. 8 is a waveform diagram of the operation of the circuit of FIG. 7A.

In FIG. 1, which illustrates a prior art dynamic RAM, reference numeral 1 designates a memory cell array which includes 1 Mbit memory cells such as MC.sub.ij (i, j=0, 1, . . . , 1023) linked between word lines such as WL.sub.i (i=0, 1, . . . , 1023) and bit lines such as BL.sub.j (j=0, 1, . . . , 1023).

One of the word lines is selected by row address buffers 2 and row address decoders 3, which are controlled by a row control circuit 4. Upon receipt of an external control signal, i.e., a row address strobe signal RAS, the row control circuit 4 generates sequentially the following clock signals: a row enable signal RE, a word line driving signal WD, a latch enable signal LE for sense amplifiers 5, and a word line reset signal WDR.

One of the bit lines, i.e., one column selection line such as CL.sub.j corresponding thereto, is selected by the column address buffers 6 and the column address decoders 7, which are controlled by a column control circuit 8. Upon receipt of an external control signal, i.e., a column address strobe signal CAS, the column control circuit 8 generates sequentially a column enable signal CE, a column selection line driving signal CD, a clock signal .phi..sub.R for a data output buffer 9, and a column selection line resetting signal CLR. Reference numeral 10 designates input/output gates, each of which is controlled by one of the column selection lines, such as CL.sub.j.

Reference numeral 11 designates a data input buffer which is controlled by a read/write control circuit 12. The read/write control circuit 12 receives an external control signal, i.e., a write enable signal WE, and generates a clock signal .phi..sub.W for the data input buffer 11.

The row address decoders 3 and the column address decoders 7 have the same configuration, as shown in FIGS. 2A and 2B. Note that FIG. 2A illustrates a one-bit decoder for the word line WL.sub.i, and FIG. 2B illustrates a one-bit decoder for the column selection line CL.sub.j. Each of the decoders of FIGS. 2A and 2B includes a NOR gate portion 31 (71) formed by transistors Q.sub.1, Q.sub.20, . . . , and Q.sub.28 and a driver portion 32 (72) formed by transistors Q.sub.3, Q.sub.4, Q.sub.5, and Q.sub.6. The transistor Q.sub.1 is controlled by the row enable signal RE or the column enable signal CE, while the transistors Q.sub.20, . . . , Q.sub.28 are controlled by the row address signals RA.sub.0 (or RA.sub.0) to RA.sub.9 (or RA.sub.9 or the column address signals CA.sub.0 (or CA.sub.0) to CA.sub.9 (or CA.sub.9). When the potential at node N1 is high and the potential of the word line driving signal WD (or of the column selection line driving signal CD) is high, the potential at the word line WL.sub.i (or the column selection line CL.sub.j) is high. In addition, when the potential of the word line resetting signal WLR (or of the column selection line resetting signal CLR) is high, the transistor Q.sub.6 is turned on, so that the potential at node N2 is high, thereby resetting the word line WL.sub.i (or the column selection line CL.sub.j).

The operation of the device of FIG. 1 will be explained with reference to FIG. 3. At time t.sub.R0, the potential of the row address strobe signal RAS falls, thereby starting an active operation. Note that such an active operation is defined by the row address strobe signal RAS. In this case, the row control circuit 4 makes the row enable signal RE low, and, as a result, at time t.sub.R1, the row address buffers 2 generate row address signals RA.sub.0, . . . , and RA.sub.9 of a metal-oxide semiconductor (MOS) level and their inverted signals RA.sub.0, . . . , RA.sub.9, by receiving external address signals ADD of a transistor-transfer logic (TTL) level, and these address signals are transmitted to the row address decoders 3. Then, at time t.sub.R2, the row control circuit 4 makes the word line driving signal WD high, so that the row address decoders 3 apply a high-level voltage to the word line WL.sub.i, thereby selecting it. After that, at time t.sub.R3, the row control circuit 4 makes the latch enable signal LE high so as to operate the sense amplifiers 5. Thus, the data stored in the memory cell MC.sub.ij is obtained at the bit line BL.sub.j.

On the other hand, at time t.sub.C0 (later than time t.sub.R0), the potential of the column address strobe signal CAS falls. Accordingly, at time t.sub.C1, the column control circuit 8 makes the column enable signal CE low. As a result, the column address buffers 6 generate column address signals CA.sub.0, . . . , and CA.sub.9 of a MOS level and their inverted signal CA.sub.0, . . . , CA.sub.9, by receiving external address signals ADD of a TTL level. These signals are transmitted to the column address decoders 7. Note that, in this case, the external address signals ADD are multiplexed for the two kinds of signals. Then, at time t.sub.C2, the column control circuit 8 makes the column selection line driving signal CD high, so that the column address decoders 7 apply a high-level voltage to the column selection line CL.sub.j. As a result, at time t.sub.C3, the data at the bit line BL.sub.j is transmitted via the input/output gates 10 to data bus DB, which is connected to the data output buffer 9 and the data input buffer 11. Then, at time t.sub.C4, the column control circuit 8 makes the clock signal .phi..sub.R high, so as to activate the data output buffer 9, thereby obtaining read data D.sub.out at the output thereof. Thus, one read operation for the memory cell MC.sub.ij is completed.

Subsequently, at time t.sub.W1, the potential of the write enable signal WE falls, thereby starting a write operation for the same memory cell MC.sub.ij. In this case, the read/write control circuit 12 controls the column control circuit 8, thereby resetting the data output buffer 9. Then, at time t.sub.W1, the read/write control circuit 12 makes the clock signal .phi..sub.W high, so that write data D.sub.in is transmitted from the data input buffer 11 to the data bus DB. Further, at time t.sub.W2, the potential at the bit line BL.sub.j changes in response to the potential at the data bus DB. Thus, one write operation is completed.

Then, at time T.sub.RS1, the potential of the row address strobe signal RAS and the potential of the column address strobe signal CAS both rise, thereby starting a reset operation for precharging each circuit of the device. That is, the row control circuit 4 makes the signals WD, and LE low and makes the signals RE and WDR high, while the column control circuit 8 makes the signal CD low and makes the signals CE and CLR high. As a result, the row address signals RA.sub.0 to RA.sub.9 (RA.sub.0 to RA.sub.9), the word line WL.sub.i, the bit lines, the column address signals CA.sub.0 to CA.sub.9 (CA.sub.0 to CA.sub.9), the column selection line CL.sub.j, and the data bus DB are reset.

After a predetermined reset time period passes, the potential of the row address strobe signal RAS again falls, thereby starting another cycle time period.

Thus, in the device of FIG. 1, during the same active time period, a data read operation and a data write operation are both performed upon the same memory cell, thus carrying out a read-modified write operation.

In the device of FIG. 1, however, a reset operation for each portion thereof is carried out only after the potential of the row address strobe signal RAS rises. As a result, the reset time period is large, thus increasing the cycle time period.

The operation of another prior art dynamic RAM will be explained with reference to FIG. 4 (U.S. Pat. No. 4,376,989). Note that this dynamic RAM has a similar configuration to that of the device of FIG. 1. Therefore, the details of the construction of this device are omitted. In FIG. 4, at time t.sub.R0, the potential of the row address strobe signal RAS falls, thereby starting an active operation, i.e., a read mode. Even in this case, the row control circuit 4 makes the row enable signal RE low, and, as a result, at time t.sub.R1, the row address buffers 2 generate row address signals RA.sub.0, . . . , and RA.sub.9 of a MOS level and their inverted signals RA.sub.0, . . . , RA.sub.9, by receiving external address signals ADD of a TTL level. These address signals are transmitted to the row address decoders 3. Then, at time t.sub.R2, the row control circuit 4 makes the word line driving signal WD high, so that the row address decoders 3 apply a high-level voltage to the word line WL.sub.i, thereby selecting it. After that, at time t.sub.R3, the row control circuit 4 makes the latch enable signal LE high so as to operate the sense amplifiers 5. Thus, the data stored in the memory cells is obtained at the bit lines.

In this case, the row address signals RA.sub.0 to RA.sub.9 and RA.sub.0 to RA.sub.9 are automatically reset after the completion of a selecting operation for the word line WL.sub.i. That is, the row control circuit 4 makes the row enable signal RE high after a predetermined time period from the rising of the word line driving signal WD. Also, the word line WL.sub.i is automatically reset after the completion of operation of the sense amplifiers 5. That is, the row control circuit 4 makes the word line driving signal WD low and makes the word line resetting signal WDR high after a predetermined time period from the rising of the latch enable signal LE.

On the other hand, at time t.sub.C0 (later than time t.sub.R0), the potential of the column address strobe signal CAS falls. Accordingly, at time t.sub.Cl, the column control circuit 8 makes the column enable signal CE low. As a result, the column address buffers 6 generate column address signals CA.sub.0, . . . , and CA.sub.9 of a MOS level and their inverted signal CA.sub.0, . . . , CA.sub.9, by receiving external address signals ADD of a TTL level. These signals are transmitted to the column address decoders 7. Then, at time t.sub.C2, the column control circuit 8 makes the column selection line driving signal CD high, so that the column address decoders 7 apply a high-level voltage to the column selection line CL.sub.j. As a result, at time t.sub.C3, the data at the bit line BL.sub.j is transmitted via the input/output gates 10 to the data bus DB. Then, at time t.sub.C4, the column control circuit 8 makes the clock signal .phi..sub.R high, so as to activate the data output buffer 9, thereby obtaining read data D.sub.out at the output thereof.

In this case, the column address signals CA.sub.0 to CA.sub.9 and CA.sub.0 to CA.sub.9 are automatically reset after the completion of a selecting operation for the column selection line CL.sub.j. That is, the column control circuit 8 makes the column enable signal CE high after a predetermined time period from the rising of the column selection line driving signal CD. Also, the column selection line CL.sub.j is automatically reset after the completion of a selecting operation for the data output buffer 9. That is, the column control circuit 4 makes the column selection line driving signal CD low and makes the column selection line resetting signal CLR high after a predetermined time period from the rising of the clock signal .phi..sub.R. Also, the bit lines and the data bus DB are automatically reset within the active time period.

At time t.sub.RS1, the potential of the row address strobe signal RAS and the potential of the column address strobe signal CAS both rise, thereby starting a reset operation for precharging each circuit of the device which has not been reset at this stage. In this case, the data output buffer 9 is reset.

In a dynamic RAM which carries out the operation as shown in FIG. 4, since most of the circuits and the signals are reset before the rising of the row address strobe signal RAS, it is possible to reduce the reset time period from time t.sub.RS1 to time t.sub.RS2, thus reducing the cycle time period. In such a dynamic RAM, however, each circuit or signal is automatically reset or precharged within an active time period defined by the strobe signal, so it is impossible to carry out a read-modified write operation. If it is desired to carry out a write operation upon the same memory cell MC.sub.ij accessed by the above-mentioned read mode, a similar operation has to be again carried out after the potential of the row address strobe signal RAS falls. Thus, this operation has two cycle time periods as compared with the read-modified write mode which can be completed within a single access cycle in the conventional RAM's.

In FIG. 5, which illustrates an embodiment of the dynamic RAM according to the present invention, elements which are the same as those of FIG. 1 are denoted by the same reference numerals. In FIG. 5, each of the row address decoders 3' has a NOR gate portion and a driver portion electrically separable from each other. Such separation of the portions is controlled by the row control circuit 4', which generates clock signals .phi..sub.0 and .phi..sub.1 therefor. Similarly, each of the column address decoders 7' has a NOR gate portion and a driver portion electrically separable from each other. Such separation of the portions is controlled by the column control circuit 8' which generates clock signals .phi..sub.0 ' and .phi..sub.1 ' therefor. The row address decoder 3' and the column address decoder 7' will be explained in more detail later.

The operation of the device of FIG. 5 will be explained with reference to FIG. 6. At time t.sub.R0, the potential of the row address strobe signal RAS falls, thereby starting an active operation. In this case, the row control circuit 4' makes the row enable signal RE low, and, as a result, at time t.sub.R1, the row address buffers 2 generate row address signals RA.sub.0, . . . , and RA.sub.9 of a MOS level and their inverted signals RA.sub.0, . . . , RA.sub.9, by receiving external address signals ADD of a TTL level, and these address signals are transmitted to the row address decoders 3'. Then, at time t.sub.R2, the row control circuit 4' makes the word line driving signal WD high, so that the row address decoders 3' apply a high-level voltage to the word line WL.sub.i, thereby selecting it. After that, at time t.sub.R3, the row control circuit 4' makes the latch enable signal LE high so as to operate the sense amplifiers 5. Thus, the data stored in the memory cell MC.sub.ij is obtained at the bit line BL.sub.j.

In this case, the row address signals RA.sub.0 to RA.sub.9 and RA.sub.0 to RA.sub.9 are automatically reset after the completion of a selecting operation for the word line WL.sub.i. That is, the row control circuit 4' makes the row enable signal RE high after a predetermined time period from the rising of the word line driving signal WD. Also, the NOR gate portions of the decoders 3' are automatically reset after the completion of a selecting operation for the word line WL.sub.i. That is. the row circuit 4' makes the clock signal .phi..sub.0 low and makes the clock signal .phi..sub.1 high after a predetermined time period from the rising of the latch enable signal LE. However, the driver portions of the row address decoder 3', the word line WL.sub.i, and the bit lines are not reset until the row address strobe signal RAS rises.

On the other hand, at time t.sub.C0 (later than time t.sub.R0), the potential of the column address strobe signal CAS falls. Accordingly, at time t.sub.C1, the column control circuit 8' makes the column enable signal CE low. As a result, the column address buffers 6 generate column address signals CA.sub.0, . . . , and CA.sub.9 of a MOS level and their inverted signal CA.sub.0, . . . , CA.sub.9, by receiving external address signals ADD of a TTL level. These signals are transmitted to the column address decoders 7'. Then, at time t.sub.C2, the column control circuit 8' makes the column selection line driving signal CD high, so that the column address decoders 7' apply a high-level voltage to the column selection line CL.sub.j. As a result, at time t.sub.C3, the data at the bit line BL.sub.j is transmitted via the input/output gates 10 to the data bus DB. Then, at time t.sub.C4, the column control circuit 8' makes the clock signal .phi..sub.R high, so as to activate the data output buffer 9, thereby obtaining read data D.sub.out at the output thereof.

In this case, the row address signals CA.sub.0 to CA.sub.9 and CA.sub.0 to CA.sub.9 are automatically reset after the completion of a selecting operation for the column selection line CL.sub.j. That is, the column control circuit 8' makes the column enable signal CE high after a predetermined time period from the rising of the column selection line driving signal CD. Also, the NOR gate portions of the column address decoders 7' are automatically reset after the completion of a selecting operation for the data output buffer 9. That is, the column control circuit 8' makes the clock signal .phi..sub.0 ' low and makes the clock signal .phi..sub.1 ' high, after a predetermined time period from the rising of the clock signal .phi..sub.R. However, the driver portions of the column address decoders 7' and their subsequent circuits are not reset until the column address strobe signal CAS rises. Thus, one read operation for the memory cell MC.sub.ij is completed.

As explained above, even after the read operation is completed, the driver portions of the row address decoders 3' and their subsequent circuits and the driver portions of the column address decoders 7' and their subsequent circuits are not reset, so that the memory cell MC.sub.ij remains in a selected state. In this state, at time t.sub.W1, the potential of the write enable signal WE falls, thereby starting a write operation for the same memory cell MC.sub.ij. In this case, the read/write control circuit 12 controls the column control circuit 8', thereby resetting the data output buffer 9. Then, at time t.sub.W1, the read/write control circuit 12 makes the clock signal .phi..sub.W high, so that write data D.sub.in is transmitted from the data input buffer 11 to the data bus DB. Further, at time t.sub.W2, the potential at the bit line BL.sub.j changes in response to the potential at the data bus DB. Thus, one write operation is completed.

At time t.sub.RS1, the potential of the row address strobe signal RAS and the potential of the column address strobe signal CAS both rise, thereby starting a reset operation. In this case, the driver portions of the row address decoders 3' and their subsequent circuits, and the driver portions of the column address decoders 7' and their subsequent circuits, are reset.

Thus, in the device of FIG. 5, it is possible to carry out a read-modified write operation. Also, in the device of FIG. 5, since the row address buffers 2, the NOR gate portions of the row address decoders 3', the column address buffers 6, and the NOR gate portions of the column address buffers 7' are automatically reset after their own operations, it is possible to reduce the reset time period (i.e., the period from time t.sub.RS1 when the potential of the row address strobe signal RAS rises to time t.sub.RS2 when it falls), thus reducing the cycle time period. That is, in this case, the driver portions of the row address decoders 3' and their subsequent circuits, and the driver portions of the column address decoders 7' and their subsequent circuits, are reset after time t.sub.RS1. Such a reset operation may be completed before the NOR gate portions of the row address decoders 3' operate at time t.sub.R1 ' of the next cycle period. Therefore, even if the reset time period from time t.sub.RS1 to time t.sub.RS2 is small, no problem occurs.

The row address decoders 3' and the column address decoders 7' of FIG. 5 will now be explained with reference to FIGS. 7A and 7B. In FIG. 7A, each of the row address decoders 3' includes a separation circuit 33 formed by transistors Q.sub.7 and Q.sub.8 linked between the NOR gate portion 31 and the driver portion 32. The transistors Q.sub.7 and Q.sub.8 of the circuit 33 are controlled by the clock signals .phi..sub.0 and .phi..sub.1 of the row control circuit 4'. Similarly, in FIG. 7B, each of the column address decoders 7' includes a separation circuit 73 formed by transistors Q.sub.7 and Q.sub.8 linked between the NOR gate portion 71 and the driver portion 72. The transistors Q.sub.7 and Q.sub.8 of the circuit 72 are controlled by the clock signals .phi..sub.0 ' and .phi..sub.1 ' of the column control circuit 7'.

The operation of each of the row address decoders 4', which is similar to that of each of the column address decoders 7', will be explained with reference to FIG. 8. The case wherein the word line WL.sub.i is selected is illustrated in the upper portion of FIG. 8. In this case, at time t.sub.o, when the row address strobe signal RAS falls, the row enable signal RE falls at time t.sub.1. As a result, the transistor Q.sub.1 is cut off, and, at time t.sub.2, the row address signals RA.sub.0 (or RA.sub.0) to RA.sub.8 (or RA.sub.8) are applied to the gates of the transistors Q.sub.20, . . . , Q.sub.28. In the case where the word line WL.sub.i is selected, all the gates of the transistors Q.sub.20, . . . , Q.sub.28 are at a low level. Accordingly, these transistors are all cut off. As a result, the node N1 is at a high level. At this time, since the clock signal .phi..sub. 0 is at a high level, the transistor Q.sub.3 is in an on state. Since the potential at node N1 is high, the potential at node N3 is also high. In this state, when the word drive signal WD is caused to be high, the gate of the transistor Q.sub.3, i.e., the potential at node N3 is further raised by the capacitance between the drain and gate of the transistor Q.sub.3 and, simultaneously, the word drive signal WD is transmitted via the transistor Q.sub.4 to the word line WL.sub.i, so that the potential at the word line WL.sub.i is pulled up to a high level. Next, at time t.sub.5, when the clock signal .phi..sub.0 becomes low and the signal .phi..sub.1 becomes high, the transistor Q.sub.3 is cut off so that the NOR gate portion 31 and the driver portion 32 are separated electrically from each other. In this case, since the potential at node N3 is higher than the power supply voltage V.sub.cc, and the potential at the word line WL.sub.i and the clock signal .phi..sub.1, are both high, the transistor Q.sub.5 is cut off and the word line WL.sub.i remains high. Next, at time t.sub.6 the row address signals RA.sub.0 (or RA.sub.0) to RA.sub.9 (or RA.sub.9) are reset so that they become low, and, in addition, the potential of the row enable signal RE is changed from low to high. As a result, the transistor Q.sub.1 of the NOR gate portion 31 is turned on so as to precharge node N1, thereby resetting the NOR gate portion 31. However, since the transistor Q.sub.8 is cut off so that the NOR gate portion 31 and the driver portion 32 are separated electrically from each other, even when the NOR gate portion 31 is reset, no problem occurs.

The output potential of the driver portion 32, i.e., the potential at the word line WL.sub.i, remains high, and the NOR gate portion 31 and its pre-stages are reset. In this state, if necessary, the write enable signal WE is caused to be low, thereby carrying out a write operation. At time t.sub.7, when the row address strobe signal RAS is again caused to be high, the word line driving signal WD is caused to be low, and the word line resetting signal WDR is caused to be high. As a result, the transistor Q.sub.6 is turned on so as to pull up the potential at node N2. Thus, the transistor Q.sub.4 is turned on and the transistor Q.sub.5 is turned off, so that the potential at the word line WL.sub.i is caused to be low, that is, the word line WL.sub.i is reset.

After the word line WL.sub.i is reset, the clock signal .phi..sub.1 is pulled down to a low level and the signal .phi..sub.0 is pulled up to a high level. As a result, the transistor Q.sub.8 is cut off and the transistor Q.sub.7 is turned on.

The case wherein the word line WL.sub.i is not selected is illustrated at the lower portion of FIG. 8. In this case, when the row address strobe signal RAS falls and the row enable signal RE falls, the address signals applied to the gates of the transistors Q.sub.20, . . . , Q.sub.28 turn on at least one of the transistors Q.sub.20, . . . , Q.sub.28. As a result, at time t.sub.3, the potential at node N1 is changed from high to low. At this time, since the clock signal .phi..sub.0 is at a high level, the transistor Q.sub.3 is in an on state. Therefore, the potential at node N3 is changed from high to low. In this state, at time t.sub.4, even when the word line driving signal WD is caused to be high, the transistor Q.sub.4 is cut off. Therefore, the word line WL.sub.i remains low. Next, at time t.sub.5, when the clock signal .phi..sub.0 is caused to be low and the clock signal .phi..sub. 1 is caused to be high, the transistor Q.sub.7 is cut off so that the NOR gate portion 31 and the driver portion 32 are separated electrically from each other. In this state, the transistor Q.sub.8 is turned on, so that the potential at node N3 is clamped at a low level, thereby preventing the potential at node N3 from being in a high-level floating state by the application of the word line driving signal WD, which is helpful in stabilization of the memory access operation. Next, at time t.sub.6, the row address signals RA.sub.0 (or RA.sub.0) to RA.sub.9 (or RA.sub.9) become low. In addition, the row enable signal RE is changed from low to high. As a result, the transistor Q.sub.1 of the NOR gate portion 31 is turned on so as to precharge node N1, since all the transistors Q.sub.20, . . . , and Q.sub.28 are turned off. At time t.sub.7, when the row address strobe signal RAS is again caused to be high, the word line driving signal WD is caused to be low, and the word resetting signal WDR is caused to be high, thereby resetting the word line WL.sub.i. However, in the case where the word line WL.sub.i is not selected, the word line WL.sub.i was already low, so that the potential of the word line WL.sub.i is unchanged.

After the word line WL.sub.i is reset, the clock signal .phi..sub.1 is pulled down to a low level and the clock signal .phi..sub.0 is pulled up to a high level. As a result, the transistor Q.sub.8 is cut off and the transistor Q.sub.7 is turned on. The conduction of the transistor Q.sub.7 charges up node N3 by node N1. At the time when the charging-up operation is completed, a stand-by state for the next cycle time period is established.

Here, since it takes a time period from t.sub.0 to t.sub.2 for the row address buffer 2 to generate the row address signals RA.sub.0 to RA.sub.9 and RA.sub.0 to RA.sub.9, it is possible to start the next cycle time period at time t.sub.9 when the clock signals .phi..sub.0 and .phi..sub.1 are reversed. Thus, it is possible to reduce the reset time period, i.e., the precharging time period, from t.sub.7 to t.sub.8.

As explained above, according to the present invention, it is possible to reduce the reset time period, thereby reducing the cycle time period. In addition, since the dynamic RAM according to the present invention can carry out a read-modified write operation, the device according to the present invention is compatible with conventional dynamic RAM's. Further, the dynamic RAM according to the present invention has no static circuits therein, so that this device has low power consumption.