CIRCULAR ADDRESS REGISTER

[0001] 1. Field of the Invention

[0002] This invention relates to the field of computer and programming systems, and in particular to the programming of systems that employ circular lists or iteratively process multiple-word operands.

[0003] 2. Description of Related Art

[0004] Circular lists are often used to implement First-In, First-Out (FIFO) buffers. The FIFO model represents a list of items wherein new items are added to the top of the list, and old items are removed from the bottom of the list. In effect, the FIFO is structured as a list of L data items; when a data item is added to the list, it is given the next available address of L1; when a data item is removed from the list, the address of each data item remaining in the list is decremented by one. The moving of each remaining data item down in the list each time a data item is removed from the list, however, is inefficient. Instead, the defined bottom of the list is moved up by one each time a data item is removed. The logical address of each data-item in the list is defined by its distance from this defined bottom of the list, rather than by its physical location in the continually growing list. Even though the removed data items may remain in the physical locations below the bottom of the list, they are no longer within the logical address space of the list, and are no longer considered to be a part of the list of L items. As would be evident to one of ordinary skill in the art, the continued advancement of the physical location used for each added data item, without a corresponding physical relocation of each data item to lower physical locations as data items are removed, will eventually exhaust the physical resources used to contain these data items.

[0005] A circular list is embodied as N contiguous physical memory locations, and a corresponding controlling mechanism that adds new data items to the list, starting at the beginning of the physical memory locations, and continuing to each next available physical location until the end of the contiguous locations is reached. When the next data item arrives after reaching the end of the contiguous locations, this next data item is placed at the beginning of the contiguous locations, replacing the prior data item that was stored at the beginning of the contiguous locations. Each next data item is placed into each next physical location, replacing the data items previously stored at each next physical location, until the end of the contiguous locations is again exceeded, at which point the next data item is placed at the beginning of the contiguous locations. This process continues ad infinitum without consuming more than the N contiguous physical memory locations at any time. Note, however, that the prior data items that are being replaced by the next data items must have been data items that were conceptually removed from the list. That is, at no time can the logical length of the list, L, exceed the physical length of the contiguous locations N. In a typical embodiment, the controlling mechanism maintains a pointer to the logical bottom of the list, and updates it accordingly as each data item is removed from the list, and also maintains a pointer to the logical top of the list, and updates it accordingly as each data item is added to the list. The controlling mechanism also inhibits the addition of a new data item when the top of the list becomes coincident with the bottom of the list, and waits for the bottom of the list to be advanced (by a removal of the oldest item in the list) before placing the new data item into the list.

[0006] Typically, multiple programming instructions are required to control the incrementing and decrementing of a pointer that is associated with a circular list. The instructions must test whether the pointer has gone beyond the upper or lower bounds of the physically contiguous memory locations, and adjust the pointer accordingly to provide the circular addressing effect.

[0007] It is an object of this invention to provide a device that facilitates the use of circular lists. It is a further object of this invention to ease the task of programming a device to facilitate the use of circular lists. It is a further object of this invention to ease the task of programming a device to facilitate the use of multiple-word data items by modeling the multiple-word data items as circular addressing process.

[0008] These objects and others are achieved by providing a device and corresponding programming instructions that facilitate a circular addressing process. The device is configured to provide an address output that is constrained to lie within specified bounds. When a circular increment or circular decrement instruction is executed that would cause the address to exceed a bound, the address is reset to the other bound. In a preferred embodiment, the programming instruction also sets condition flags that indicate when the address is at each bound. By providing these bounds flags in conjunction with the circular addressing operation, multiple-word data items can be processed efficiently. A base-address of N contiguous words in a memory is loaded into the circular register, and a circular addressing instruction is used to access each word of the N contiguous words in sequence; a bounds flag is set when the last word of the multi-word data item is accessed. This bounds flag may thus be used to signal the end of processing of N words of a multi-word data item, and can be used to trigger a load of a next multi-word data item, or to trigger a next operation on the current data-item, and so on. Other condition flags are also provided to facilitate the processing of multi-word data-items.

[0009] The invention is explained in further detail, and by way of example, with reference to the accompanying drawings wherein:

[0010] FIG. 1 illustrates an example block diagram of a processing system that provides circular addressing in accordance with this invention.

[0011] FIG. 2 illustrates an example block diagram of a circular-address-register in accordance with this invention.

[0012] Throughout the drawings, the same reference numerals indicate similar or corresponding features or functions.

[0013] FIG. 1 illustrates a processing system 100 that includes a circular-address-register 200 that is used to access a memory 110 comprising contiguous storage elements d₀-d_N1 111-119 via a circular addressing scheme. The circular-address-register 200 provides an address 101 to the contiguous storage elements 111-119, corresponding, for example, to the aforementioned circular list. The address 101 contained in the circular-address-register 200 can be incremented 102 or decremented 103 to advance the pointer up or down the elements 111-119, respectively. The circular-address-register 200 is configured to determine an associated upper and lower bound corresponding to the range of the addresses of the memory elements 111-119. In a preferred embodiment, a base address A_d 121 and a size N 122 are provided when the address register 200 is configured for processing the memory elements 111-119 as a circular list or buffer.

[0014] The processor 120 issues commands 123 to the circular-address-register 200 to effect desired operations, in response to program instructions 150. The following commands 123 are applicable to the address register 200 in accordance with this invention:

OPERATION	EFFECT
Circular	If Address > Ad	Then Address Ad
Increment (102)	(N 1)	Else Address Address 1.
Circular	If Address < Ad	Then Address Ad (N 1)
Decrement (103)		Else Address Address 1.

[0015] Note that the numeral 1 is used herein for ease of illustration. As is common in the art, the size of a word may differ from the scale that is used for addressing the elements d₀-d_N 111-119. That is, for example, the address 101 may be specified in terms of 8-bit bytes or 16-bit words, while the elements d₀-d_N1 111-119 may be configured to contain 32-bit words. As used herein, the term word is defined to be the size of the elements d₀-d_N1 111-119, and the increment and decrement of the address 101 contained in the circular-address-register 200 corresponds to a change of the address 101 by one word size.

[0016] A variety of embodiments of a circular-address-register that effects the above operations will be evident to one of ordinary skill in the art in view of this disclosure. In a preferred embodiment, both the circular increment operation and the circular decrement operation are provided, although a minimal embodiment may include one or the other. Also in a preferred embodiment, other operations are provided, as discussed below, to facilitate alternative and/or conventional uses of the circular-address-register 200. Note that a direct embodiment of the circular increment and decrement operations as defined above requires that the base address A_d 121 and N 122 be stored, for comparison with the current address 101 contained in the circular-address-register 200.

[0017] FIG. 2 illustrates a preferred embodiment of the circular-address-register 200 in accordance with this invention. The circular-address-register 200 includes a command processor 210, an index register 220, and an address register 230. The address register 230 contains an address Aout 231 that can be used to effect a circular addressing technique. As applied to FIG. 1, the output address Aout 231 will typically correspond to the address 101 of an addressed memory element 111-119; a different reference numeral is utilized because, as discussed below, the circular-address-register 200 in accordance with this invention can be utilized for other applications, such as counting functions, in addition to a circular addressing function.

[0018] The index register 220 contains an index 221 that ranges from 0 to (N1), and, in a circular addressing application, corresponds to the addressed memory element's position relative to a base address Ain 201. As applied to FIG. 1, the input base address Ain 201 will typically correspond to the base address A_d 121 of a circular list, or, as discussed below, the base address of a multi-word data-item, or a count number. The parameter N 122 may be a fixed value, for a circular-address-register 200 that is configured for a particular application, or it may be a variable that is set when the circular-address-register is configured for a specific use. If N is fixed, it is contained within the circular-address-register 200, and need not be provided by the processor 120. The size of the index register 220 determines the maximum value of N 123; for example, if the index register 220 comprises three bits, N must be less than or equal to 8. In like manner, the size of the address register 230 determines the maximum range of addresses 231.

[0019] The command processor 210 in a preferred embodiment accepts the command input 123, and supports the following operations, in dependence upon this input 123:

(1)	Load_L	Aout Ain;	Index 0.
(2)	Load_U	Aout Ain;	Index N 1.
(3)	Circular	If (Index (N 1))	Then Aout Aout (N 1);
	Increment		Index 0
			Else Aout Aout 1;
			Index Index 1.
(4)	Circular	If (Index 0)	Then Aout Aout (N 1);
	Decrement		Index (N 1)
			Else Aout Aout 1;
			Index Index 1.
(5)	Increment	Aout Aout 1;	If (Index (N 1))
			Then Index 0
			Else Index Index 1.
(6)	Decrement	Aout Aout 1;	If (Index 0)
			Then Index (N 1)
			Else Index 0.

[0020] Operations (1) Load_L loads the address register 230 with the input address Ain 201; for ease of reference, this input address Ain 201 is hereinafter termed the base address, or lower bound of the address space, when used in the context of a circular addressing application. Operation (1) Load_L sets the index to 0, corresponding to the word at the lower bound of the address space.

[0021] Operations (2) Load_U loads the address register 230 with the upper bound of the address space Ain 201, and sets the index to N1, corresponding to the word at the upper bound of the address space. These differing load operations (1) Load_L and (2) Load_U are provided to facilitate either low-to-high data access or high-to-low data access, by initializing the address and the index to each appropriate starting condition.

[0022] Operations (3) Circular Increment implements a circular increment addressing function. When the index is at its upper bound (indexN1), the next circular increment operation (3) resets the address Aout 231 to the lower address bound, by subtracting N1 from its current value, and resets the index to zero, corresponding to the word at the lower address bound. Otherwise, when the index is not at its upper bound, the next circular increment operation (3) increments the current contents of the address register and index register.

[0023] Operations (4) Circular Decrement implements a circular decrement addressing function. When the index is at its lower bound (index0), the next circular decrement operation (4) resets the address Aout 231 to the upper address bound, by adding N1 to its current value, and resets the index to N1, corresponding to the word at the upper address bound. Otherwise, when the index is not at its lower bound, the next circular decrement operation (4) decrements the current contents of the address register and index register.

[0024] Operations (5) Increment implements a modified form of a conventional increment operation. The increment operation (5) increments Aout 231, without regard to an upper address bound, similar to a conventional register-increment operation. The increment operation (5) modifies the index in the same manner as the circular increment operation (3), thereby performing a modulo N up-counting function via the index.

[0025] Operations (6) Decrement implements a modified form of a conventional decrement operation. The decrement operation (6) decrements Aout 231, without regard to a lower address bound, similar to a conventional register-decrement operation. The decrement operation (6) modifies the index in the same manner as the circular decrement operation (4), thereby performing a modulo N down-counting function via the index.

[0026] Note that by utilizing the index register 220, a number of advantages are gained. A test for an index equal to zero (Index 0) can be implemented with fewer logic devices than a test for corresponding addresses (Address> A_d). Also, because the parameter N, the size of the list, is often substantially smaller in magnitude than the available address space, a test for an index equal to the upper bound of the index (Index N1) can be implemented with fewer logic devices than a test for the upper bound of the address (Address> A_d(N1)). By providing the index register 220, and correlating its content with the content of the address register 230, the base address A_d need not be retained, and, assuming that N is smaller than the address space, the overall storage requirements are reduced. Similarly, the addition or subtraction of (N1) to the current contents of the address register Aout 231 to reset the address Aout 231 to the upper or lower bounds can typically be implemented with less circuitry and routing than an addition of (N1) to the base address A_d, and a load of this sum into the address register 230.

[0027] The command processor 210 also provides the state of the (Index 0) and (Index N1) comparisons as bounds condition flags 211, 212 that facilitate circular list and multi-word operations, as discussed below. Optionally, the command processor 210 could provide one or more flags 213 that identify other states of the index 221, including a separate flag for each value of the index 221. Also optionally, the index 221 could be provided as an output from the circular-address-register 200.

[0028] The bounds flags 211, 212 are provided to facilitate a processing of an N-word data item. When the last word of the N-word data item is accessed, at either the upper or lower address bound, depending upon whether the data item is being accessed in a low-to-high or high-to-low manner, the appropriate bounds flag 211, 212 is set. As would be evident to one of ordinary skill in the art, this set flag can be used to trigger the end of the N-word process, to commence another process on the N-word data item, or to load the contents of the memory elements 110-119 with a next N-word data item, and so on.

[0029] As discussed above, multiple address registers 200 are often utilized as registers that are configured to facilitate the processing of circular lists. For example, a pair of registers could be configured to provide the aforementioned top and bottom list pointers. Also, multiple address registers 200 can be configured for accessing multiple sets of memory elements 110-119, for example, to facilitate pipeline operations on multiple data items. As illustrated in FIG. 2, in a preferred embodiment, the next address 215 is also provided as an output from the circular-address-register, further facilitating pipeline operations; additional addresses, such as the next-next-address may also be provided to facilitate three-stage pipelines, and so on.

[0030] Although any register can be utilized as the index register 220 and the address register 230, special purpose registers are preferably utilized that are optimized to perform the circular addressing function. That is, for example, if the processing system is configured to process data items within a limited address space, the address register 230 need only be wide enough to accommodate this limited address space. In like manner, if the processing system is configured to process N-word data-items the index register 220 need only be wide enough to accommodate this size N, independent of the size of the address register.

[0031] The foregoing merely illustrates the principles of the invention. It will thus be appreciated that those skilled in the art will be able to devise various arrangements which, although not explicitly described or shown herein, embody the principles of the invention and are thus within its spirit and scope. For example, as illustrated in FIG. 2, an optional Aout 0 flag 214 may be provided to facilitate an efficient counter function in conjunction with the Load_L and Decrement or Increment operations. That is, the content of the address register 230 need not be an address pointer, and may be loaded with a count value M. Subsequent decrement operations will result in the Aout 0 being set, signaling a completion of M counts. These and other system configuration and optimization features will be evident to one of ordinary skill in the art in view of this disclosure, and are included within the scope of the following claims.