-
2005-11-22
10/373,595
2003-02-24
US 6,968,452 B2
2005-11-22
-
-
Dennis M. Butler
2023-06-23
A method of synchronizing and reconfiguring configurable elements in a programmable unit is provided. A unit has a two- or multi-dimensional, programmable cell architecture (e.g., DFP, DPGA, etc.), and any configurable element can have access to a configuration register and a status register of the other configurable elements via an interconnection architecture and can thus have an active influence on their function and operation. By making synchronization the responsibility of each element, more synchronization tasks can be performed at the same time because independent elements no longer interfere with each other in accessing a central synchronization instance.
Get notified when new applications in this technology area are published.
This application is a continuation of U.S. patent application Ser. No. 09/537,932, filed Mar. 29, 2000, now U.S. Pat. No. 6,526,520 which is a division of U.S. patent application Ser. No. 08/946,812 filed on Oct. 8, 1997, now issued as U.S. Pat. No. 6,081,903, which is expressly incorporated herewith by reference in its entirety. This application also claims the benefit of the priority date of DE 19704728.9, filed on Feb. 8, 1997, under 35 U.S.C. Β§119.
1.1 State of the Art
1.2 Problems
With the units of today (FPGAS, DPGAs, etc.), synchronization of the configurable elements is usually based on the clock pulse of the unit. This type of time-controlled synchronization poses many problems, because it is often not known in advance how much time a task will require until a final result is available. Another problem with time-controlled synchronization is that the event upon which the synchronization occurs is not triggered by the element to be synchronized itself but instead by an independent element. In this case, two different elements are involved in the synchronization. This leads to much higher management overhead.
1.3 Improvement Through the Invention
This invention describes a method which makes it possible for the synchronization to be based on the elements to be synchronized themselves. Synchronization is no longer implemented or administered by a central instance. By making synchronization the responsibility of each element, many more synchronization tasks can be carried out at the same time, because independent elements no longer interfere with each other in accessing the central synchronization instance. The patent claims concern details and specific embodiments as well as features of the synchronization method according to this invention.
2. Description of the Invention
2.1 Overview of the Invention; Abstract
In a unit with a two- or multi-dimensional, programmable cell architecture (DFP, DPGA), any configurable element can have access to the configuration and status register of the other configurable elements via an interconnection architecture and can thus have an active influence on their function and operation. Thus, in addition to the usual method, the configuration may also take place from the processing array (PA, see PACT02) through a primary logic unit.
2.2 Detailed Description of the Invention
The invention starts with a run-time programmable, run-time reconfigurable unit. The configurable elements on the chip have one or more configuration registers for different tasks. Read and write access to these configuration registers is possible. In the method described here, it is assumed that for the following information, a configuration can be set in an element to be configured.
A cell is configured by a command which determines the function of the cell that is to be carried out. In addition, configuration data are entered to set the interconnection with other cells and the contents of the status register. The cell is ready for operation after this procedure.
To permit a flexible and dynamic interaction of multiple cells, each cell can have read or write access to all the configuration registers of another cell. The type of command with which the cell has been configured specifies to which of the many configuration registers will allow read or write access. Each command that can be executed by the cell exists in as many different modes of address as there are different independent configuration registers in an element to be configured.
Example: One cell has the configuration registers indicated above (interconnection, command and status) and is to execute the command ADD which performs addition. Through the various types of ADD command, it is now possible to select where the result of this function is to be transferred.
In addition to the result, each cell can generate a number of trigger signals. The trigger signals need not necessarily be transmitted to the same target cell as the result of processing the configured command. One trigger signal or a combination of multiple trigger signals triggers a certain action in the target cell or puts the cell in a certain status. A description of the states can be found in the following text. There are the following trigger signals:
Due to the possibility of indicating in the processing cell in which register of the target cell the result is to be entered and which type of trigger signal is to be generated, a quantity of management data can be generated from a data stream. This management data is not a result of the actual task to be processed by the chip, but instead serves only the function of management, synchronization, optimization, etc. of the internal status.
Each cell can assume the following states, which are represented by appropriate coding in the status register:
Due to these various states and the possibility of read and write access to the various registers of a cell, each cell can assume an active management role. In contrast with that, all existing units of this type have a central management instance which must always know and handle the overall status of the unit.
To achieve greater flexibility, there is another class of commands which change in type after the first execution. Using the example of the ADD command, the command has the following format:
This possibility can be expanded as much as desired, so that commands of the type ADD-C-V-A-C- . . . B are also conceivable. Each command can assume all permutated combinations of the different addressing and triggering types.
2.3 Hardware Expansion with Respect to PACT02
2.3.1 Additional Registers
A status register and a configuration register are added to the registers described in PACT02. Both registers are triggered by the PLU bus and they have a connection to the state machine of the SM UNIT (PACT02, FIG. 2: 0213).
2.3.2 Change in the PLU Bus
In PACT02, the configurable registers M/F-PLUREG are managed exclusively over the PLU bus (PACT02, FIG. 2: 0210). To ensure the function according to this invention, there must also be an additional possibility of access through the normal system bus (PACT02, FIG. 2: 0201). The same thing is true of the new status and configuration registers.
Only the part of the system bus that is interconnected with the PAE over the BM UNIT (PACT02, FIG. 2: 0210) is relevant for the registers. Therefore, the bus is relayed from the BM UNIT to the registers, where upstream multiplexers or upstream gates assume the switching between the PLU bus and the system bus relevant for the PAE.
The multiplexers or gates are wired so that they always switch through the system bus relevant for the PAE except after a reset of the unit (RESET) or when the ReConfig signal (PACT02, FIG. 3: 0306) is active.
2.3.3 Expansion of the System Bus
The system bus (PACT02, FIG. 2: 0201) is expanded to the extent that the information about the target register is transmitted together with the data. This means that an address which selects the desired register at the data receiver is also sent at the same time.
FIG. 1: This figure shows how a loop construct can be implemented by using triggers.
FIG. 2: This figure shows how a comparator construct can be implemented by using multiple triggers.
FIG. 3: This figure shows how a comparator construct with multiple outputs can be implemented by using multiple triggers and interleaving them.
FIG. 4: This shows the required extensions in comparison with PACT02.
FIG. 1. Macro 0103 will be executed 70 times in this example. Execution of the macro once requires 26 clock cycles. This means that counter 0101 may be decremented by one only once in every 26 clock cycles. One problem with freely programmable units is that it is not always possible to guarantee that the execution of macro 0103 is actually concluded after 26 cycles. A delay can occur, for example, due to the fact that a macro which is supposed to deliver input data for macro 0103 suddenly needs ten clock cycles longer. For this reason, the cell in macro 0103 sends a trigger signal to counter 0101 by means of which the result of the computation is sent to another macro. At the same time, processing of macro 0103 is stopped by the same cell. This cell βknowsβ exactly that the condition for termination of a computation has been reached.
The trigger signal sent in this case is a STEP trigger which causes counter 0101 to execute its configured function once. The counter decrements its count by one and compares whether it has reached the value zero. If this is not the case, a GO trigger is sent to macro 0103. This GO trigger signal causes macro 0103 to resume its function again. This process is repeated until counter 0101 has reached a value of zero. In this case a trigger signal is sent to macro 0102, where it triggers a function.
A very fine-grained synchronization can be achieved by this interaction of triggers.
FIG. 2 corresponds to the basic idea of FIG. 1. However, in this case the function in element 0202 is not a counter but a comparator. Macro 0201 sends a comparison value to comparator 0202 after each processing run. Depending on the output of the comparison, different triggers are again activated to cause an action in macros 0203, for example. The construct implemented in FIG. 2 corresponds to that of an IF query in a program language.
FIG. 3. As in FIG. 2, several comparators 0301, 0302 are used here to implement the construction of an IF-ELSE-ELSE construct (or a multiple choice). By using various types of triggers and connections of these triggers to macros 0303, 0304, very complex sequences can be implemented easily.
FIG. 4 shows the differences in comparison with PACT02 in FIG. 2. The differences are shown with shading. The configuration register (0401) and the status register (0402) are connected to the SM UNIT over the bus (0407). Registers 0401, 0402, F- and M-PLUREG are connected to a gate 0403 over an internal bus (0206) this connects the internal bus (0406) to the PLU bus 0405, depending on the setting, to permit configuration by the PLU, or to the BM UNIT over a bus 0408. Depending on the addressing on data bus 0404, the BM unit relays the data to the O-REG or to the addressed register 0401, 0402, F- or M-PLUREG.
6. Definition of Terms
BM UNIT: Unit for sending data on the bus systems outside the PAE. The data is sent over multiplexers for the data inputs and gates for the data outputs. oACK lines are implemented as open-collector drivers. The BM UNIT is controlled by the M-PLUREG.
Data receiver: The unit(s) that further process(es) the results of the PAE.
Data transmitter: The unit(s) that make(s) data available for the PAE as operands.
Data string: A data string consists of a bit sequence of any length. This bit sequence is a processing unit for a system. Commands for processors and similar units as well as strictly data can be coded in a data string.
DFP: Data flow processor according to (unexamined) German Patent DE 44 16 881.
DPGA: Dynamically programmable gate array. Known.
EALU: Expanded arithmetic and logic unit. An ALU that has been expanded by adding special functions that are needed or appropriate for operation of a data processing system according to German Patent DE 441 16 881 A1. These are counters in particular.
Elements: Collective term for all types of self-contained units which are used as one piece in an electronic unit. Elements are thus:
Event: An event can be analyzed by a hardware element in any appropriate manner for the application and it can trigger a conditional action as a response to this analysis. Events are thus, for example:
FPGA: Field-programmable gate array. Known.
F-PLUREG: Register in which the function of the PAEs is set. Likewise, the one-shot and sleep modes are set. The PLU writes into the register.
H level: Logic 1 level, depending on the technology used.
Configurable element: A configurable element is an element of a logic unit which can be set by a configuration string for a specific function. Configurable elements are thus all types of RAM cells, multiplexers, arithmetic and logic units, registers and all types of internal and external interconnection description, etc.
Configurable cell: See logic cells.
Configure: Setting the function and interconnection of a logic unit, a (FPGA) cell or a PAE (see reconfigure).
Configuration data: Any quantity of configuration strings.
Configuration memory: The configuration memory contains one or more configuration strings.
Configuration string: A configuration string consists of a bit sequence of any length. This bit sequence represents a valid setting for the element to be configured, so the result is a functional unit.
Primary logic unit: Unit for configuring and reconfiguring the PAE. Embodied by a microcontroller adapted specifically to its task.
Logic cells: Configurable cells used in DFPs, FPGAs, DPGAs which fulfill simple logical or arithmetic tasks according to their configuration.
L level: Logic 0 level, depending on the technology used.
M-PLUREG: Register in which the interconnection of the PAEs is set. The register is written by the PLU.
O-REG: Operand register for storing the operands of the EALU. Permits chronological and functional independence of the PAEs from the data transmitters. This simplifies the transfer of data because it can take place asynchronously or packet-oriented. At the same time, this creates the possibility of reconfiguring the data transmitters independently of the PAEs or reconfiguring the PAEs independently of the data transmitters.
PLU: Unit for configuring and reconfiguring the PAEs. Embodied by a microcontroller adapted specifically to its function.
SM UNIT: State machine UNIT. State machine that controls an EALU.
Switching table: A switching table is a ring memory which is addressed by a control. The entries in a switching table may contain any configuration strings. The control can execute commands. The switching table responds to trigger signals and reconfigures the configurable elements on the basis of an entry in a ring memory.
Reconfigure: Reconfiguring any quantity of PAEs while any remaining quantity of PAEs continue their own functions (see configure).
Processing cycle: A processing cycle describes the period of time needed by a unit to go from a defined and/or valid state to the next defined and/or valid state.
Cells: Synonym for configurable elements.
7. Conventions
7.1 Naming Conventions
Subassembly: UNIT
Operating mode:MODE
Multiplexer:MUX
Negated signal:not
Register for PLU visible: PLUREG
Register internal:REG
Shift registers:sft
7.2 Function Convention
| AND function & |
| A | B | Q |
| 0 | 0 | 0 |
| 0 | 1 | 0 |
| 1 | 0 | 0 |
| 1 | 1 | 1 |
| OR function # |
| A | B | Q |
| 0 | 0 | 0 |
| 0 | 1 | 1 |
| 1 | 0 | 1 |
| 1 | 1 | 1 |
| NOT function ! |
| A | Q | |
| 0 | 1 | |
| 1 | 0 | |
| GATE function G |
| EN | D | Q |
| 0 | 0 | β |
| 0 | 1 | β |
| 1 | 0 | 0 |
| 1 | 1 | 1 |
1. A method of synchronizing and reconfiguring configurable elements in units having a multi-dimensional programmable cell architecture, comprising:
generating by at least one of the configurable elements at least one synchronization signal within a data stream during processing by the configurable elements;
transmitting via a data bus the at least one synchronization signal to a first at least one other of the configurable elements for synchronization;
generating at least one configuration string from the data stream based on commands; and
transmitting via the data bus the at least one configuration string together with an address of a register to be addressed to at least one of (a) the first at least one other configurable element and (b) a second at least one other configurable element,
wherein the configurable elements are at least one of configurable and reconfigurable without influence of an external primary logic unit.
2. The method according to claim 1, wherein, during the synchronization, a configurable element carries out a single operation in response to a trigger.
3. The method according to claim 1, wherein, during the synchronization, a configurable element carries out a plurality of synchronization operations in response to a trigger.
4. The method according to claim 1, wherein, during the synchronization, execution of an operation by a configurable element is stopped in response to a trigger.
5. The method according to claim 1, wherein, during the synchronization, a configurable element is enabled for reconfiguration by a trigger.
6. The method according to claim 1, wherein a configurable element displays its instantaneous status in a status register.
7. The method according to claim 1, wherein the address of the resister is coded in commands.