Configurable IC with error detection and correction circuitry

Description

CROSS REFERENCE TO RELATED APPLICATION

This Application is related to the following applications with the same filing date: U.S. patent application Ser. No. 11/390,973, filed Mar. 27, 2006 now U.S. Pat. No. 7,529,992.

FIELD OF THE INVENTION

The present invention is directed towards a configurable IC with error detection circuitry.

BACKGROUND OF THE INVENTION

The use of configurable integrated circuits (“IC's”) has dramatically increased in recent years. One example of a configurable IC is a field programmable gate array (“FPGA”). An FPGA is a field programmable IC that often has logic circuits, interconnect circuits, and input/output (I/O) circuits. The logic circuits (also called logic blocks) are typically arranged as an internal array of circuits. The logic functions of these circuits are typically configured by the contents of an internal Static Random Access Memory (SRAM) which is loaded with data from an external source prior to operation.

FIG. 1 illustrates an example of a configurable logic circuit 100. This logic circuit can be configured to perform a number of different functions. As shown in FIG. 1, the logic circuit 100 receives a set of input data 105 and a set of configuration data 110. The configuration data set is stored in a set of SRAM cells 115. From the set of functions that the logic circuit 100 can perform, the configuration data set specifies a particular function that this circuit has to perform on the input data set. Once the logic circuit performs its function on the input data set, it provides the output of this function on a set of output lines 120. The logic circuit 100 is said to be configurable, as the configuration data set “configures” the logic circuit to perform a particular function, and this configuration data set can be modified by writing new data in the SRAM cells. Look-up tables are examples of configurable logic circuits.

FIG. 2 illustrates an example of a configurable interconnect circuit 200. This interconnect circuit 200 connects a set of input data 205 to a set of output data 210. This circuit receives configuration data bits 215 that are stored in a set of SRAM cells 220. The configuration bits specify how the interconnect circuit should connect the input data set to the output data set. The interconnect circuit 200 is said to be configurable, as the configuration data set “configures” the interconnect circuit to use a particular connection scheme that connects the input data set to the output data set in a desired manner. Moreover, this configuration data set can be modified by writing new data in the SRAM cells. Multiplexers are one example of interconnect circuits.

During operation, the contents of SRAM 220 are susceptible to random bit errors (soft errors) due to high-energy atomic particles (such as alpha particles or neutrons) traveling through the chip releasing an avalanche of electrons. The electrons find their way either to the positive supply or to an SRAM storage cell representing one bit of memory. If the total electrical charge reaching the storage cell is larger than the critical charge to upset logic, a soft error takes place. As semiconductor geometries and voltage levels are reduced in order to provide more functionality per chip, the critical charge level is reduced, making soft errors more likely. Undetected, such errors can seriously effect the logical operation of the FPGA and thus affect the system in which it is used, often in unpredictable ways.

One prior art solution to this problem involves a process called scrubbing, which periodically reads back the contents of the configuration memory to an external system, and compares the read-back data to the original data. If a discrepancy is detected, the subsystem using the FPGA can be stopped, the FPGA configuration memory downloaded again and then operation continued. One problem with this approach is that it takes time away from the external process which could have been used to execute other tasks. A second problem with this approach is that it results in a high probability that a system fault will occur between the times the external system checks the configuration memory.

Another prior art solution to the problem involves the application of triple modular redundancy (TMR). With TMR, the IC requires three copies of the user circuit, all of which are initially loaded with the same data by a configuration network. During operation, the outputs of the three circuits are compared to each other. If one of these differs from the other two, then only data from one of the two matching circuits is used. The primary problem with this method is that it requires more than three times as much IC real estate to implement.

Accordingly, there is a need for a configurable IC that has the ability to check and correct its configuration memory continuously, during normal operation of the system in which it is embedded. Moreover, there is also a need for a configurable IC that has the ability to check and correct its configuration memory independently of the external system.

SUMMARY OF THE INVENTION

Some embodiments of the invention provide a configurable IC that performs error detection and correction of its configuration data. This configurable IC includes a configuration memory for storing configuration data. This IC further includes (1) error detection circuitry for detecting an error, and (2) a circuit that outputs from the IC an uncorrectable error signal indicating the detection of an error. In some embodiments, the configurable IC has a circuit that causes the IC to reset when the error circuitry detects an uncorrectable error. In some embodiments, a circuit outside of the IC causes the IC to reset when the error circuitry detects an uncorrectable error.

BRIEF DESCRIPTION OF THE DRAWINGS

The novel features of the invention are set forth in the appended claims. However, for purpose of explanation, several embodiments of the invention are set forth in the following figures.

FIG. 1 illustrates an example of a configurable logic circuit.

FIG. 2 illustrates an example of a configurable interconnect circuit.

FIG. 3 illustrates an example of a circuit for ECC error detection and correction of a configuration memory where the ECC logic precedes a multiplexer and corrected data writeback is directly to configuration memory.

FIG. 4 illustrates the timing between the Odd/Even clock signals, Odd context signals, and state signals ST0, and ST1.

FIG. 5 illustrates the interface between ECC logic output of error-corrected context data and writeback input to configuration memory.

FIG. 6 illustrates an example of a circuit for ECC error detection and correction of a configuration memory where the ECC logic precedes a multiplexer and corrected data writeback is through a configuration network.

FIG. 7 illustrates an example of a circuit for ECC error detection and correction of a configuration memory where the ECC logic is located after a multiplexer and corrected data writeback is directly to configuration memory.

FIG. 8 illustrates the timing between an Odd/Even clock, odd and even context signals, and corrected and retimed context signals.

FIG. 9 illustrates a group of components which can reload configuration memory upon detection of an uncorrectable error.

FIG. 10 illustrates a portion of a configurable logic circuit that includes ECC detection of uncorrectable context data errors.

FIG. 11 illustrates a process to signal an external system when the ECC logic detects an uncorrectable context data error.

FIG. 12 illustrates a two-tiered interconnect and memory structure.

FIG. 13 illustrates the timing between the CLK signals and the two sets of four multiplexer control signals.

FIG. 14 illustrates an application electronic system which includes a configurable IC.

DETAILED DESCRIPTION OF THE INVENTION

In the following description, numerous details are set forth for purpose of explanation. However, one of ordinary skill in the art will realize that the invention may be practiced without the use of these specific details. In other instances, well-known structures and devices are shown in block diagram form in order not to obscure the description of the invention with unnecessary detail.

Some embodiments of the invention provide a configurable IC that performs error detection and correction of its configuration data. This configurable IC includes a configuration memory for storing configuration data and error correction data. This IC further includes error correction circuitry for (1) receiving the configuration data, (2) correcting a particular error in the received configuration data when the particular error exists in the configuration data, and (3) outputting the configuration data without the particular error. The configurable IC has a configurable circuit (e.g., a configurable logic circuit or a configurable interconnect circuit) that receives the potentially error-corrected configuration data from the error correction circuitry. In some embodiments, the configurable IC includes circuitry to write the corrected configuration data and error data back into the configuration memory.

I. Memory Error Detection and Correction Structures

FIG. 3 illustrates a portion of a configurable IC of some embodiments of the invention. This configurable IC has an error correcting circuitry 300 that stores, outputs, and processes configuration and error correction bits for a particular core reconfigurable circuit 355. The circuitry 300 provides a configuration data word to the core reconfigurable circuit 355 at a particular rate. In some embodiments, this particular rate is a sub-cycle rate that is faster than a design cycle rate, where the design cycle rate is the rate of a design that the configurable IC is implementing. Examples of configurable IC's that implement IC designs by operating reconfigurable circuits at rates faster than the design rate are provided in U.S. patent application Ser. No. 11/081,859 “Configurable IC with Interconnect Circuits that also Perform Storage Operations”.

As shown FIG. 3, the circuitry 300 includes a configuration memory 305, ECC logic 335 and a multiplexer 350. The configuration memory 305 includes a read/write port 310, two banks of storage elements 321 and 322, and two read only ports 315 and 316. This port 310 stores the received configuration and error-correction data in one of two banks of storage elements. In some embodiments, the storage elements are SRAM cells. The read/write port receives configuration and error correction data from the configuration network. The configuration network is the source of configuration information and a shared interface to an application electronic system, which is further described below.

Each word of configuration data accessed from one of the two banks of storage elements is termed a context as it defines the operation (i.e., the logic or interconnect operation) that the core reconfigurable circuit 355 (i.e., logic or interconnect circuit) performs for a particular sub-cycle out of the set of operations that the circuit 355 can perform. Each configuration data word is read out with an associated error-correction word (i.e., with an associated error-correction code, or ECC) that contains the data necessary for detection and correction of an error in its associated configuration data word. Each configuration data word and its associated ECC form a configuration/ECC data word. In some embodiments, each configuration data word is stored in the same storage-element bank as its associated ECC.

The configuration memory 305 uses two banks 321 and 322 of storage elements because the circuitry 300 utilizes a two-tiered multiplexer structure to provide configuration data words at an effective data rate of 4× to the core reconfigurable circuit 355. This two-tiered multiplexer structure (1) uses a set of two or more out-of-phase signals ST operating at a rate of X to read the two banks 321 and 322 in an interleaved manner, and (2) uses a clock signal CLK that operates at a rate of 2× to operate the multiplexer 350.

More specifically, FIG. 3 conceptually illustrates the circuitry for reading out the two storage-element banks 321 and 322 as the read-only ports 315 and 316. These read-only ports (i.e., the circuitry for reading out the two storage-element banks 321 and 322) receive the set of two or more out-of-phase signals ST, as shown in FIG. 3. In response to these two clock signals, the read-only ports 315 and 316 output configuration/ECC words to the ECC logic 335 at a data rate of 2×. The operation of the read-only ports 315 and 316 will be further described below by reference to FIG. 9 and FIG. 10.

As shown in FIG. 3, the ECC logic 335 has two components: (1) an even correction circuit block 340 for processing configuration/ECC words that are read out of the even storage-element bank 321, and (2) an odd correction circuit block 345 for processing configuration/ECC words that are read out of the odd storage-element bank 322. As further described below, each of these error correction blocks (1) detects particular errors (e.g., one bit errors) in any configuration/ECC word that it receives, (2) if there is an error, corrects it, and (3) provides the potentially corrected configuration data word of the configuration/ECC word to the multiplexer 350 at a rate of 2×. If an error-correction block detects and corrects an error in a configuration/ECC word, the error-correction block writes back the corrected configuration/ECC word back to the appropriate storage-element bank. This write-back is further described below by reference to FIGS. 5-7.

As mentioned above, the multiplexer 350 receives configuration data words at a 2× rate from the ECC logic 335. The select line of the multiplexer is controlled by a clock signal that operates at a rate of 2×. This allows the multiplexer to output error-corrected configuration data words at a rate 4× to the core reconfigurable circuit 355.

FIG. 4 provides the timing diagram for the circuitry 300 in some embodiments. In this example, the circuitry stores four configuration data words that are provided to the reconfigurable circuit in four looping cycles. Each configuration data word can have any arbitrary number of bits.

To enable this “four-loopered” design, the read-only ports 315 and 316 receive two 90° out-of-phase clock signals ST0 and ST1 that operate at a rate X. In response to these two clock signals, the read-only port 315 outputs configuration/ECC words to the even ECC logic 340 at a data rate of 2× during the even phases. Similarly, the read-only port 316 outputs configuration/ECC words to the odd ECC logic 345 at a data rate of 2× during the odd clock phases.

The multiplexer 350 receives on its select line a clock signal CLK that operates at a rate 2×. The relationship between the clock signals ST0, ST1, and CLK are illustrated in FIG. 4. As shown in this figure, an odd configuration/ECC data word is read on the falling edge of the clock CLK. Similarly, an even configuration/ECC data word is read on the rising edge of the clock CLK. The ECC blocks correct the read out configuration/ECC data words in half the cycle of the clock CLK. Accordingly, to provide error corrected configuration data to the core reconfigurable circuit 355, the read-only port reads out each storage element bank at a data rate of 2×, in an interleaved manner. Hence it can provide corrected configuration data words to the multiplexer at the rate of 2×. The multiplexer 350 then provides the reconfigurable circuit 355 the corrected configuration data at a rate 4×, as its select is controlled by the clock signal CLK.

II. Write Back

As mentioned above, the ECC blocks 340 and 345 write back any configuration/ECC word that they correct to the storage-element banks 321 and 322. FIGS. 5 and 6 illustrate two different approaches that some embodiments use to achieve this write back.

FIG. 5 illustrates the circuitry needed to route the corrected configuration/ECC data from the ECC logic directly back to the configuration memory. As shown in this figure, this circuitry includes two multiplexers 505 and 510. The even multiplexer 505 selects between the configuration network and the even ECC block of the ECC logic 335, based on the write signal from the ECC logic 335. Similarly, the odd multiplexer 510 selects between the configuration network and the even ECC block of the ECC logic 335, based on the write signal from the ECC logic 335.

The function of these multiplexers is to select whether the configuration network or the ECC logic 335 provides write data to the configuration memory 305. If the ECC logic detects and corrects an error in the context data it outputs this write signal. The write signal also causes the selected data from the multiplexer to be written into the storage elements 320 via the read/write port 310. Asserting the write signal causes the corrected context data to be written into the configuration memory 305 on a priority basis over requests to write data from the configuration network.

Instead of directly writing back an error corrected, configuration/ECC data word, some embodiments write back such a data word through the configuration network. FIG. 6 illustrates one such approach. To write back the corrected configuration/ECC data words through the configuration network, the circuitry 600 of FIG. 6 includes a configuration network interface 620 for each of the ECC blocks 610 and 615 in the ECC logic 605.

When the ECC logic detects an error in the context data, it signals the configuration network interface with an error signal. This interface then sends the corrected configuration/ECC data to the configuration network and requests it to write the data back to the configuration memory through the read/write port 310. The odd and even ECC logic also outputs the corrected data to the multiplexer 350 which feeds the core reconfigurable circuit 355.

III. Alternate Position for the ECC Block

In some embodiments, design tradeoffs in circuit speed, die area and routing may make it advantageous to structure the IC with a single ECC logic circuit rather than the even and odd sections shown in FIG. 3. FIG. 7 illustrates a portion of such a configurable IC. Specifically, this figure illustrates an error correcting circuitry 700 that stores, outputs, and processes configuration and error correction bits for a particular reconfigurable circuit 355.

Like the circuitry 300 of FIG. 3, the circuitry 700 of FIG. 7 provides a configuration data word to the reconfigurable circuit 355 at a particular rate. Also, like the circuitry 300, the circuitry 700 includes a configuration memory 305 with two storage-element banks 321 and 322. It also has a multiplexer 705 that is similar to the multiplexer 350. However, the multiplexer 750 is before an ECC logic 710, instead of after the ECC logic as in circuitry 300. The circuitry 700 also includes a retiming circuit 720.

As in the circuitry 300, the configuration memory 305 of the circuitry 700 outputs even and odd configuration/ECC data words in an interleaved manner at a rate 2×. The multiplexer 705 receives on its select line a clock signal CLK that operates at a rate 2×. Accordingly, the multiplexer 705 can output the received configuration/ECC data words at a rate 4× to the single ECC logic block 710.

The ECC logic block 710 then performs error detection/correction operation on the configuration/ECC data word that it receives. Specifically, for each configuration/ECC data word that it receives, the ECC logic block (1) detects particular errors (e.g., one bit errors) in the configuration/ECC word, (2) if there is an error, corrects it, and (3) provides the potentially corrected configuration data word of the configuration/ECC word to the retiming circuit at a rate of 4×. If the error logic block 710 detects and corrects an error in a configuration/ECC word, this block writes back the corrected configuration/ECC word back to the appropriate storage-element bank. To perform the write back, the ECC logic block 710 can use the approaches described above by reference to FIGS. 5 and 6.

The retiming circuit 720 receives the potentially error-corrected configuration data words from the ECC logic block 710. In some embodiments, the retiming circuit is implemented as a set of double edge-triggered flip-flops which can hold or pass-thru the data from the ECC logic to the reconfigurable circuit. Alternately, the retiming circuit could be implemented as a set of edge-triggered flip-flops operating with a 4× clock. The purpose of this retiming register is to allow for context data-dependent variations in the time it takes for data to flow thru the ECC logic and settle within one sub-cycle. Register 720 latches these bits and passes them on to the core reconfigurable circuit 355 at the start of the next sub-cycle. Without register 720, the reconfigurable logic might not receive the configuration data word that is for a particular cycle at the start of that cycle.

The timing relationships governing the operation of the circuit shown in FIG. 7 is illustrated in FIG. 8. In particular, it shows the relationship of the odd and even clock signals to the context data as it enters the ECC logic, is corrected and finally retimed and sent to the reconfigurable logic.

IV. Uncorrectable Error Detection and Signaling

In some embodiments described above, the error correction logic can detect and correct a single erroneous bit of a configuration/ECC word. If more than one bit of a configuration/ECC word were in error, then the ECC logic of these embodiments will not be able to correct either or both errors. This situation would then result in use of incorrect configuration data by the reconfigurable circuit 355.

Accordingly, some embodiments of the invention provide the capability to detect, but not correct, a configuration/ECC word with more than one erroneous bit. Upon detection of this condition, the configurable IC of some embodiments generates a signal that causes the configurable IC or a device external to the IC to reload configuration data in the configuration IC. In some embodiments, the configurable IC resets and restarts its operation after such reloading of its configuration data.

FIG. 9 illustrates a group of components of some embodiments which accomplishes this function. Specifically, this figure illustrates a group of components 900 that can detect uncorrectable errors in configuration memory data and initiate configuration memory reloading.

The group 900 includes a configurable circuit arrangement (e.g., a circuit array) 905, a nonvolatile memory 915, and a configuration controller 920. The configurable circuit arrangement 905 contains configuration memory and associated error detection circuitry. The nonvolatile memory 915 may be loaded with configuration data. When the error detection circuit detects an uncorrectable error, it sends an uncorrectable error signal to the configuration controller 920. In response, the controller 920 retrieves configuration and ECC data from the nonvolatile memory 915 and reloads the configuration and ECC data into the configurable circuit arrangement 905. The controller reloads the data through the configuration network of the configurable circuit arrangement 905 of some embodiments. U.S. patent application Ser. No. 11/375,363, entitled “Configuration Network for a Configurable IC,” and filed on Mar. 13, 2006. The reloading of the configuration and error correction data eliminates the error in this data.

In some embodiments, the nonvolatile memory 915 is Flash memory. In other embodiments, this memory may be Electrically Programmable Read Only Memory (EPROM) or ROM programmed during manufacturing of the memory or after manufacturing of the group of components 900. In some embodiments, the group of components 900 could be fabricated by dividing the functions of 905, 915 and 920 among several ICs (i.e., several chips) that are placed on one or more printed wiring boards. In other embodiments, all three component functions, 905, 915 and 920 could be implemented on a single IC die (on a single IC chip). One of ordinary skill in the art would appreciate that any other system configuration may also be used in conjunction with the invention, and these system configurations might use different nonvolatile memory technologies and have fewer or additional components.

FIG. 10 illustrates a portion of a configurable IC which can detect and signal uncorrectable errors in its configuration data. Specifically, this figure illustrates an error correcting circuitry 1000 that stores, outputs and processes configuration and error correction bits for a particular reconfigurable circuit 355. The circuitry 1000 for FIG. 10 is similar to the circuitry of 700 of FIG. 7, except that the ECC logic circuit 1010 in the circuit 1000 not only can detect and correct n-bit errors in the configuration data words, but also it can detect m-bit errors in the configuration data words, where m>n. In some embodiments, above and below, n equals 1 and m equals 2. However, in other embodiments, n may be greater then 1 and m greater then 2. One of ordinary skill in the art will recognize that other embodiments may be realized of the configurable IC that could detect and correct errors in configuration/ECC words of any lengths m and n where m>n.

When the ECC circuit 1000 in FIG. 10 detects an m-bit error in a configuration data word, it outputs an uncorrectable-error signal to signify that it has detected an uncorrectable error. FIG. 11 illustrates one approach that some embodiments take to address such an uncorrectable-error signal. As shown in this figure, the process 1100 of FIG. 11 starts (at 1105) when an ECC logic circuit 1010 of the error-correcting circuitry 1000 of the configurable IC detects a multi-bit error in a configuration data word. Section V describes how some embodiments detect an m-bit error in a configuration data word.

Next, at 1110, the ECC logic circuit that detected the m-bit error in the configuration data word, generates an uncorrectable-error signal. The configurable IC is then reset (at 1115). In some embodiments, the resetting of the configurable IC entails setting the value of several state elements (e.g., latch, register, storage element, memory, etc.) within the IC to a pre-specified value. In some embodiments, circuitry inside of the configurable IC receives the uncorrectable-error signal and in response directs the configurable IC to be reset. In other embodiments, circuitry outside of the configurable IC receives this error signal and in response directs the IC to be reset. Also, when the configurable IC is reset, the system that contains the configurable IC is notified. Such a system can be fault-tolerant and might not need to be reset. In other words, such a system might contain fault-tolerant mechanisms to deal with the resetting of the configurable IC. Alternatively, the system might not be fault tolerant and therefore might have to reset when the configurable IC resets.

After resetting the configurable IC, configuration and ECC data is reloaded (at 1120) into the configurable IC through a configuration network as described above. In some embodiments, the configuration and ECC data is stored in (and hence loaded from) a non-volatile memory outside of the configurable IC. In other embodiments, the configuration and ECC data is stored within (and hence loaded from) a non-volatile memory of the configurable IC. Once the configuration and ECC data is reloaded in the configurable IC, the configurable IC restarts (at 1125) its operation.

V. Error Correction Coding

The purpose of the ECC logic is to ensure the reliability of the context bits in the face of random occurrences of errors introduced into the configuration SRAM. One embodiment of ECC utilizes the block-based algorithms otherwise known as Hamming codes. The simplest such code is the Single-bit Error Correction (SEC) ECC algorithm. That is, a system utilizing the SEC ECC algorithm can detect and correct a data word as long as it contains no more than one erroneous bit. The basic idea of SEC ECC is to map a given data vector into a longer codeword. All valid SEC ECC codewords differ from each other by at least 3 bits in the bit pattern. This difference is known as a Hamming distance of 3. The distance of 3 bits ensures that any given valid SEC ECC codeword can sustain a one bit error, and can unambiguously recover the original valid SEC ECC codeword. The advantage of SEC ECC is that it is relatively inexpensive to implement. SEC ECC requires the use of (log 2 d)+1 check bits (c) for d data bits. The external system initially computes the check bits and combines them with the data bits to form the codeword. These codewords are loaded into the configuration memory by the configuration network prior to operating the reconfigurable IC. Subsequently, the ECC circuits illustrated above maintain the integrity of the data by continuously checking and correcting the check and data bits and writing the result back to the configuration memory.

As described above, the process of encoding a bit vector into a space where the bit vectors have a Hamming distance of 3 means that a single bit error and a double bit error are indistinguishable from each other. Therefore, errors longer than one bit cannot be detected. However, if the data vectors encoded into an ECC codeword are in a space where the codewords have a Hamming distance of 4 from each other, then a one-bit error and a two-bit error can be distinguished from each other. This process is termed Single bit Error Correction with Double bit Error Detection (SECDED) ECC. However, in SECDED ECC, two-bit errors cannot be corrected because multiple valid SECDED codewords can result in the same error vector with different two-bit errors. The cost of SECDED ECC over SEC ECC alone is the addition of one more ECC bit in the data vector. ECC circuit 1000 in FIG. 10 of some embodiments illustrates an example of a portion of a configurable IC in which. SECDED is implemented to detect two-bit errors as well as correcting one-bit errors.

In some embodiments, the number of check bits for SEC ECC is c=5 for d=24 data bits. For SECDED ECC, c=6 for 24 data bits. One of ordinary skill in the art will recognize that c and d could be other values so long as c is greater than or equal to (log 2 d) for SEC ECC and (log 2 d)+1 for SECDED ECC. One of ordinary skill in the art will also recognize that other algorithms for error detection and correction could be utilized and may result in the ability to detect and correct more than one bit and/or result in different values of c and d.

VI. Two Tiered Structure for Retrieving Data

As mentioned above, the read-only port 315 of the various circuits mentioned above conceptually illustrates the circuitry for reading out the two storage-element banks 321 and 322 in an interleaved manner. As further discussed above, this read out circuitry is part of a two-tiered multiplexer structure that can provide configuration data words at a rate of 4× to the reconfigurable circuit 355. This two-tiered multiplexer structure uses (1) a set of two or more out-of-phase signals ST that operate at a rate of X to read the two banks 321 and 322 in an interleaved manner, and (2) uses a clock signal CLK that operates at a rate of 2× to operate the multiplexer 350 or 705.

FIG. 12 illustrates an example of the circuitry for reading out the two storage-element banks 321 and 322. In this example, the circuitry stores eight configuration data words, which are provided to the reconfigurable circuit in eight looping cycles. To enable this “eight-loopered” design, the read-only ports 315 and 316 receive four out-of-phase signals ST0(0)-ST0(3) that operate at a rate X.

FIG. 12 illustrates eight storage cells 1250a-1250h that each store one value for one bit of configuration or error-correction data for each of the eight looping cycles. Four of the cells are in the even storage-element bank, while the other four cells are in the odd storage-element bank.

FIG. 12 also illustrates a multiplexer 1260, which represents either the multiplexer 350 or 705 of FIG. 3 or 7. This multiplexer 1260 represents the second tier of multiplexers in the two-tiered multiplexer structure. This multiplexer receives on its select line a clock signal CLK that operates at a rate 2×.

FIG. 12 further illustrates a first tier of multiplexers that are built into the sensing circuitry of the storage elements 1250. Specifically, this figure illustrates four storage elements 1250a-1250h that are arranged in two columns 325 and 330. Each storage element stores one logical bit of data in a complementary format.

Each of the two complementary outputs of each storage element 1250 connects to a pair of stacked NMOS transistors 1220 and 1225. One transistor 1225 in each stacked pair of NMOS transistors is part of a first tier multiplexer structure. Specifically, in the two-tiered circuit structure 325 and 330 illustrated in FIG. 12, the first tier multiplexer structure is formed by the 16 transistors 1225, which receive the sub-cycle signals ST0[0]-ST0[3] and ST1 [0]-ST1 [3].

Through the sub-cycle signals ST0[0]-ST0[3] and ST1[0]-ST1[3], the multiplexer transistors 1225 selectively connect the NMOS transistors 1220 to the cross-coupled PMOS transistors 1205 and 1210. One pair of PMOS transistors 1205 and 1210 exists in each column and form part of the sensing amplifier for the storage elements in that column.

When the NMOS transistors 1220 associated with one storage element 1250 connect to the PMOS transistors 1205 and 1210, they form a level-converting sense amplifier. This amplifier then translates the signals stored in the storage element to the bit lines 1235 or 1240. The circuits 325 and 330 provide the content of the storage elements through level-converting sense amplifiers, because, in some embodiments, the storage elements are storage cells that use a reduced voltage to store their data in order to conserve power.

The bit lines 1235 and 1240 connect to the two-to-one multiplexer 1260. As described above, this multiplexer is controlled through the clock signal CLK and its complement. Accordingly, when the clock signals CLK and CLK, and the sub-cycle signals, ST0[0]-ST0[3] and ST1[0]-ST1 [3] have the timing relationship illustrated in FIG. 13, the first tier multiplexer (formed by the transistors 1225) and the second tier multiplexer 1260 operate to output data from the storage elements 1225 at a rate that is twice the rate of the clock signal CLK.

As mentioned above, the circuit 325 and 330 of FIG. 12 can be used to provide data on a sub-cycle basis, or any other reconfiguration cycle basis. By building the first multiplexer stage into the sense amplifier section of the storage elements, this circuit reduces signal path delay from the storage elements. Also, it operates with storage elements that have less power consumption. Furthermore, it reduces power consumption by using NMOS transistors 1220 that are not driven by full voltage levels, and sharing the PMOS transistors 1205 and 1210 that are necessary for level conversion between two storage elements.

One of ordinary skill will realize that other embodiments might implement the two tiered circuit 325 and 330 differently. For instance, some embodiments might have one or more circuits 1245 between the multiplexer 1260 and the storage element section (i.e., the section with the storage elements 325 and their sense amplifiers).

VII. External Electronic System

FIG. 14 conceptually illustrates an example of an electronic system 1400 that has an IC 1405, which includes one of the invention's configurable circuit arrangements that were described above. The system 1400 can be a stand-alone computing or communication device, or it can be part of another electronic device. As shown in FIG. 14, the system 1400 not only includes the IC 1405, but also includes a bus 1410, a system memory 1415, a nonvolatile memory 1420, a storage device 1425, input devices 1430, output devices 1435, and communication interface 1440.

The bus 1410 collectively represents all system, peripheral, and chipset interconnects (including bus and non-bus interconnect structures) that communicatively connect the numerous internal devices of the system 1400. For instance, the bus 1410 communicatively connects the IC 1405 with the nonvolatile memory 1420, the system memory 1415, and the permanent storage device 1425.

From these various memory units, the IC 1405 receives data for processing and configuration data for configuring the IC's configurable logic and/or interconnect circuits. When the IC 1405 has a processor, the IC also retrieves from the various memory units instructions to execute. The nonvolatile memory 1420 stores static data and instructions that are needed by the IC 1405 and other modules of the system 1400. The storage device 1425, on the other hand, is a read-and-write memory device. This device is a non-volatile memory unit that stores instructions and/or data even when the system 1400 is off. Like the storage device 1425, the system memory 1415 is a read-and-write memory device. However, unlike storage device 1425, the system memory is a volatile read-and-write memory, such as a random access memory. The system memory stores some of the instructions and/or data that the IC needs at runtime.

The bus 1410 also connects to the input and output devices 1430 and 1435. The input devices enable the user to enter information into the system 1400. The input devices 1430 can include touch-sensitive screens, keys, buttons, keyboards, cursor-controllers, microphone, etc. The output devices 1435 display the output of the system 1400.

Finally, as shown in FIG. 14, bus 1410 also couples system 1400 to other devices through a communication interface 1440. Examples of the communication interface include network adapters that connect to a network of computers, or wired or wireless transceivers for communicating with other devices. One of ordinary skill in the art would appreciate that any other system configuration may also be used in conjunction with the invention, and these system configurations might have fewer or additional components.

While the invention has been described with reference to numerous specific details, one of ordinary skill in the art will recognize that the invention can be embodied in other specific forms without departing from the spirit of the invention. For example, even though several embodiments are described above by reference to reconfigurable IC, one of ordinary skill will realize that the invention can be practiced in any type of configurable IC. Thus, one of ordinary skill in the art would understand that the invention is not to be limited by the foregoing illustrative details, but rather is to be defined by the appended claims.