Method and Apparatus for Writing to Memory Components

1. Field of the Invention

The present invention relates to the control of data to be written to memory components. More particularly, the present invention relates to a structure and method for implementing bit and byte enable signals and logic during write operations to memory components.

2. Art Background

FIG. 1 shows some of the common features of modern solid-state memory components. The memory components contain a two dimensional array of storage cells. These storage cells may be static (i.e. there is a bistable latch) or dynamic (i.e. there is a single capacitor holding a charge). A row of storage cells is read when a row address is applied to the row decoder and the appropriate read control signals are asserted. This entire row is held in a row of column amplifiers (sense amplifiers) which are typically static in nature. Any subset of this sensed row may be accessed via column addresses and control signals.

A write operation requires additional control. It is important to be able to write to a subset of a row. This is typically done as a two step process (although in a static memory component it may be done as a single step). The information to be written is placed on the write data signal lines. The write enable signal is asserted for only those bits of a row which have been selected by the column select circuitry and which are to be modified. Write enable signals are not asserted for the remaining bits. The Read/Write Column signal is then asserted, permitting those sense amplifiers of the row with the write enable signal asserted to be modified with the write data received across the write data signal lines. Subsequently, using the write row signals, the entire row may be rewritten from the column amplifiers back to the RAM array, with the subset of the storage cells modified according to the modified bits in the column amplifiers.

This technique works because the sense amplifier retains previous data written to the sense amplifier if the two write drivers driven by the write enable are allowed to float (write enable is not asserted). Although internally there is typically a write enable signal for each bit of write data, these signals are not available externally; therefore, most memory components are incapable of modifying any pattern of bits within a row.

FIG. 3 shows an example of a prior art memory component with byte enable controls for write masking. FIG. 3, and succeeding figures, use the notation set forth in FIG. 2.

In the prior art memory components, the value of m is typically one or two. The memory array 110 and the data receiver 115 typically transact the same quantum of information (m*b bits). There are typically "m" write enable signals for masking the individual bytes of write data within this transaction quantum, where m represents the number of bytes of data communicated by the memory component. The masking is controlled by the byte enable controls 120. The copy block duplicates the bit on the one input wire onto the "b" output wires.

FIGS. 4 and 5 illustrate another example of prior art in which the memory component includes a bit enable register (BitMask) 180 and write data register (WriteReg) 185. In addition to the byte enable control lines shown in the previous example of FIG. 3, two loadable internal registers, BitMask 180 and WriteReg 185 are connected to the receiver 190 for the data lines 195. Two multiplexers 200, 205 are provided for selecting the receiver 190 contents (the receivers typically have register or latch storage elements) or the contents of the two loadable registers 180, 185 for input to the write enable 210 and write data 215 input pins of the memory array 220.

As set forth in FIG. 4, this configuration permits several modes of operation. The first mode provides the functionality of the previous example set forth in FIG. 3. The second and sixth modes provide additional functionality.

FIG. 6 illustrates a simplified timing diagram for the second mode. It should be noted that the signals for the row and column portions of the address are multiplexed together. In a typical operation, a row address is received and latched by the memory component, and followed by multiple column addresses in subsequent clock cycles to access the sensed row. Referring to FIGS. 5 and 6, when the row address Row›a! is received on the address lines 225, the BitMask›a! value is also received on the data lines 195. The BitMask›a! value is latched by the bit mask register 180, and is used to perform bit masking on incoming data until a new value is received to overwrite it.

When each column address Col›a,i! is received on the address lines 225, the WData›a,i! value is also received on the data lines 195. This value is held in the receiver 190, and is used to drive the write data inputs 215 of the memory array 220. This is repeated for "n" different values of WData, using the same value of BitMask stored in the bit mask register 180. When the next row address Row›b! is received on the Address lines 225, a new BitMask›b! value is also received on the data lines 195. A second series of column write operations is then performed with the new bit mask value.

FIG. 7 shows a simplified timing diagram for Mode Six. Referring to FIGS. 5 and 7, when the row address Row›a! is received on the address lines 225, the WData›a! value is also received on the data lines 195. This value is latched by the WriteReg register 185, and is used until a new WData value is received and overwrites the value in the register 185. Thus, when each column address Col›a,i! is received on the address lines 225, the BitMask›a,i! value is also received on the data lines 195. This value is held in the receiver 190, and subsequently drives the Write enable inputs 210 of the Memory Array 220. This is repeated for "n" different values of the bit mask, using the same value stored in the WriteReg 185.

When the next row address Row›b! is received on the address lines 225, a new WData›b! value is also received on the data lines 195. A second series of column write operations is then performed using the new WData›b! value.

It is an object of the present invention to provide a sophisticated level of bit and byte masking in a memory system.

It is an object of the present invention to provide a memory system which can receive a new bit mask with each "m*b" bits of write data sent to the memory array.

In the memory system of the present invention, additional operating modes are provided to enhance the functionality and performance of the memory system. In one embodiment, a unique bit mask is supplied with the write data used in each column access. In an alternate embodiment, a bit mask register and byte mask register are provided to support bit level and byte level masking. Alternatively, the bit mask and write data Register are realized as a single register to provide the same functionality. These two registers are realized as a single register to provide the functionality while minimizing component space and cost.

In another embodiment, a separate bit mask and byte mask are provided. The byte mask is loaded with mask data in one cycle and is used during the next "b" column write accesses. This structure provides for operating modes with no bit masking, with bit masks supplied for every row access and with bit masks supplied with every column access.

In order to enhance the functionality of a system, such as a two-dimensional graphics system, in an alternate embodiment, the memory system is provided with two registers and a select control line to select data from one of two registers. In a computer graphics system, this is used to select between foreground and background colors. The embodiment can be utilized in conjunction with the other embodiments described to provide enhanced functionality and performance.

The objects, features and advantages of the present invention will be apparent from the following detailed description in which:

FIG. 1 is a block diagram illustration of the internal structure of a memory component.

FIG. 2 is a table which sets forth the notation for the block diagrams shown.

FIG. 3 is a block diagram illustration of a prior art memory system with byte enable control.

FIG. 4 is a table illustrating the possible operating modes in the prior art memory system of FIG. 5.

FIG. 5 is a block diagram illustration of a prior art memory system with a bit enable register.

FIGS. 6 and 7 are illustrations of the timing for two modes of the prior art memory system of FIG. 5.

FIG. 8 is a block diagram illustration of one embodiment of the memory system of the present invention.

FIGS. 9a and 9b illustrate the timing for two modes of the memory system of the present invention.

FIG. 10 is a block diagram illustration of one embodiment of the memory system of the present invention.

FIG. 11 is a block diagram illustration of an alternate embodiment of the memory system of the present invention.

FIG. 12 illustrates the timing for one mode of the memory system of the present invention illustrated in FIG. 11.

FIG. 13 is a block diagram illustration of an alternate embodiment of the memory system of the present invention.

FIG. 14 illustrates the timing for one mode of the embodiment shown in FIG. 13.

FIG. 15 is a table summarizing the operating modes available for the different embodiments of the memory system of the present invention.

In the following description, for purposes of explanation, numerous details are set forth in order to provide a thorough understanding of the present invention. However, it will be apparent to one skilled in the art that these specific details are not required in order to practice the present invention. In other instances, well known electrical circuits and structures are shown in block diagram form in order not to obscure the present invention unnecessarily.

FIG. 8 is a block diagram of one embodiment. The data paths are similar to those shown in FIG. 5. However, control logic 187, provides the control signals necessary for the data and bit mask transfer as shown by the timing diagrams of FIGS. 9a and 9b.

FIGS. 9a and 9b respectively illustrate the timing for two new modes hereinafter referred to as modes seven and eight, in accordance with the teachings of the present invention.

Referring to FIG. 9a, the row address Row›a! is received on the address lines. Prior to the issuance of each column address Col›a,i!, a BitMask›a,i! value is received and latched into the BitMask register. When each column address Col›a,i! is received on the address lines, the WData›a,i! value is also received on the Data wires. This WData›a,i! value is held in the receiver, and is subsequently used to drive the write data inputs to the memory array. This two cycle sequence is repeated for "n" different pairs of BitMask and WData values. Each column address can be received during the mask data cycle to provide more time to perform the access.

Mode seven, as illustrated in FIG. 9a, provides a significant performance advantage over, for example, mode two for the situation when a different bit mask must be applied to each piece of data to be written (WriteData) during a write operation. Mode seven of the system of the present invention is able to perform a write operation to the memory array in every two clock cycles of the interface, whereas mode two requires a new row access operation to apply a different bit mask to each piece of WriteData. As is well known in the art, a new row access typically takes four to eight times as long as a column access.

FIG. 9b illustrates the timing for Mode eight. The row address Row›a!, and column addresses Col›a,l!-Col›a,n! are received on the address lines. Prior to receipt of each column address Col›a,i!, a WData›a,i! value is received on the data lines and latched into the WriteReg register. When each column address Col›a,i! is received on the address lines, the BitMask›a,i! value is also received on the data lines. This value is held in the receiver, and drives the write enable inputs of the memory array. This two cycle sequence is repeated for "n" different pairs of WData and BitMask. The column address can be received when the write data is received, thereby providing more time to perform the access.

Mode eight is thus very similar to mode seven, with the only difference being the order in which BitMask and WData are supplied in each column write cycle. Thus, the same performance advantages are realized.

There are several trends in memory component design which will affect the techniques used for write control. These trends include the increase of interface speed (receivers and transmitters) relative to the speed of the memory array, and the multiplexing of control information onto data lines.

FIG. 10 illustrates a memory array and interface circuit. Referring to FIG. 10, the impact of increasing interface speed on a memory component can be shown. The number of data lines (d) 300 is less than the number of write data and write enable lines 305, 310 (m*b). The ratio of m*b/d is the same as the ratio of the cycle time of the memory array compared to the bit transmission across a time of a data line. With an aggressive interface technology this ratio might be eight or greater.

The greater the ratio, the larger the size of the BitMask and WriteReg registers. One optimization is to combine the two registers into a single BitMask register 320. At any time, this register holds either the bit mask or write data. Any drawbacks incurred for combining the registers is far outweighed by the space savings and cost savings enjoyed by eliminating a register.

As more information is written during each memory array cycle, it is also important to provide enough byte enable controls. This may be done, as before, by using "m" 325 control lines as dedicated byte enable control signals. This permits the most flexible control for applications requiring the manipulation of byte-sized quantities of information.

However, some memory components may not be able to dedicate "m" control lines for this control function. Instead, referring to FIG. 11, the byte mask control signals are multiplexed on the data lines 300 (like the bit mask controls). A byte mask register 380, is provided which enables a static set of byte enable signals to be loaded and used on multiple sets of write data during write operations by controlling the load byte mask (LoadBM) and select byte mask (SelBM) signals. Alternatively, a byte mask of m*q bits may be loaded into the byte mask register 380 prior to receiving each "q" (where q can be typically equal to b or b-1, depending upon implementation) blocks of m*b write data bits, since each byte mask bit controls "b" write data bits. It should be noted that, in the present embodiment, a multiplexer 390 with "q+1" inputs of "m" bits each is needed in order to select the proper byte mask for each block of data (and to also provide for selectively bypassing the byte mask and forcing all ones). It should be apparent that one can generate byte enable signals utilizing a compact encoding such as providing only the starting address and length information. Compact encoding technique can be applied to all embodiments. It should also be apparent that by storing values in both the bit mask and write data registers, and using those values multiple times with different addresses (under control of internal control logic), large bandwidth savings are permitted.

FIG. 12 illustrates the timing of a write operation in accordance with operating mode twenty one using the byte mask register as illustrated in the embodiment of FIG. 11. This operating mode can be viewed as an extension of operating mode one, wherein an additional clock cycle is used to transmit the byte mask. A row address is received in the first clock cycle. In the next cycle an "m*q" bit value ByteMask›a! is loaded into the byte mask register. In the next "q" column write accesses, a value WData›a,i! is received and driven to the write data inputs of the Memory Array. In each write access, one of the "q" values (each of which are "m" bits wide) is selected from the byte mask register and applied to the write enable inputs of the memory array.

Although a clock cycle is added to load the byte mask register, reducing the data bandwidth by 1/q, "m" byte enable control pins have been saved. In addition, the controller is relieved of the burden of merging the bit and byte masks. In a similar manner, operating modes two, six, seven, and eight can be extended to add a cycle to load the ByteMask register with a byte mask prior to every "q" column write accesses, thereby providing new operating modes 22 through 28.

In two-dimensional graphics systems it is advantageous to be able to write one of two color values into each pixel under control of a select mask. The two color values represent a foreground and a background color. The select mask chooses one of the two colors.

FIG. 13 illustrates an embodiment of the memory system of the present invention which provides this capability. The multiplexer 500 receives "m*b" bits from the WRegF 505 and WRegB 510 registers, and gets "m*b" select bits from the data receiver 515. FIG. 14 illustrates the timing of the memory system of the present invention operating in mode one using the two registers 505, 510. Referring to FIGS. 13 and 14, a row address is received by receiver 520 in the first cycle. In the next cycle an "m*b" bit value foreground color is loaded into the WRegF register 505. In the following cycle an "m*b" bit value background color is loaded into the WRegB register 510. In the next "n" column write accesses, a value WSel›a,i! is received and selects between the two color values on a bit-by-bit basis. Although no bitmask or bytemask is used in the above example, as shown in FIG. 13, the system is extendible to handle the additional functionality.

In a similar manner, operating modes two through eight can also be extended to include the two color registers and select signals, resulting in new modes 102 through 108.

FIG. 15 summarizes all the writing mode combinations that have been discussed, as well as all other combinations that have been implied. The first eight modes include the prior art writing modes and new extensions described earlier. Modes 11-18 are similar to modes 1-8, except that the byte mask is supplied by a single internal register rather than control inputs, as shown in FIG. 10. Likewise, in modes 21-28, the byte mask is supplied by the data input signal lines, interleaved with data input, as shown in FIG. 12, for other functions.

Modes 101, 102, 107, and 108 are similar to modes 1, 2, 7, and 8 except that the write data is supplied by two internal color registers and select information from the data inputs (combinations 3, 4, 5, and 6 are not applicable since the write data is not dependent upon the data input in these cases). Likewise, modes 111 through 118 are similar to modes 11 through 18 with the difference in the source of the write data, and modes 121 through 128 are similar to modes 21 through 28. Modes 201 through 628 permute the WDataF, WDataB, and WSelect sources through all relevant combinations.

The invention has been described in conjunction with a preferred embodiment. It is evident that numerous alternatives, modifications, variations and uses will be apparent to those skilled in the art in light of the foregoing description.