A method and device for forming an application specific integrated circuit (ASIC), in particular, in response to engineering changes.
An application specific integrated circuit (ASIC) is an integrated circuit designed or customized for a specific use or task. Generally, a customer or client requests a manufacturer or ASIC design center to fabricate an ASIC to perform specific logic. The customer may provide a “netlist,” which represents the desired logic operation for the application, or some other representation of the logic for performing the desired task to a designer or engineer. The designer or engineer takes the customer's logic and synthesizes it into gate logic. This gate logic is then placed and routed to form a physical design of the chip, which can take, e.g., many weeks to months to complete.
During the placing and routing procedure, it is not unusual for the customer to request engineering changes for the logic, e.g., in an effort to improve the ASIC functionality. In this regard, it can be difficult, after logic synthesis and the initial physical design process, to change the placed and routed gate logic. As a result, incorporating the engineering changes into the ASIC may generally result in a very complicated and lengthy ordeal for the designer/engineer. Moreover, gate logic cannot generally simply be replaced, since changing gate logic can result in different timing characteristics between the gates that must also be considered in the new gate logic.
The underlying obstacles to putting in engineering changes to a chip have for a long time plagued designers. From distilling down the minimal change in a netlist (when the change was made to register transfer logic (RTL)) to rewiring gates on an existing placement the change is always a timely complicated process and does not lend itself to automation.
Functional logic blocks (FLBs) can be used in place of standard cell logic gates in order to enable easy changes to functional blocks with wire, as discussed in commonly owned U.S. application Ser. No. 11/838,929 filed Aug. 15, 2007, the disclosure of which is expressly incorporated by reference herein in its entirety. The function can be implemented with a separate library of simple logic, e.g., NAND gates and NOR gates, where the simple logic blocks are built up from higher complexity block, e.g., AOIs with inputs tied up and down inside the library block to create the simpler logic function. This would enable additional logic that can be easily changed in the existing logic path.
According to an aspect of the invention, a chip design methodology includes identifying engineering changeable logic, and replacing the identified engineering changeable logic with flexible logic blocks.
In accordance with another aspect of the invention, a device for forming an integrated circuit chip includes a device for identifying engineering changeable logic, and a replacing device for replacing the identified engineering changeable logic with flexible logic blocks.
According to still another aspect of the invention, a method for chip design includes receiving a first netlist, receiving a subsequent netlist composed of engineering logic changes, identifying the engineering logic changes made since the first netlist, and replacing at least one of the engineering logic changes with a flexible logic block.
The present invention is described in the detailed description which follows, in reference to the noted plurality of drawings by way of non-limiting examples of exemplary embodiments of the present invention.
The cost of using FLBs in place of simple logic can be quite expensive. Accordingly, the invention is directed to a method of identifying logic to be implemented in flexible logic blocks (FLBs) and a device for implementing the method in forming a chip, e.g., an application specific integrated circuit (ASIC). Further, while starting logic can be from a variety of sources, e.g., RTL for the chip, third party vendor provided RTL or netlists, FLBs can be used with all sources according to the invention.
FLBs are enhanced logic function gates with at least one additional function wired to be initially deactivated, e.g., through extra tied inputs tied to ground. Thus, regions of the chip can be identified in order to arrange FLBs 11 across chip 10 or in the identified region of chip 10 to be interconnected with the normal ASIC logic and yet allow for easy changes to their function and operation. Further, to facilitate the late changes to the design, e.g., after the chip 10 has been placed and routed, these regions likely to require late changes can be identified in order to replace the standard logic with FLBs 11 having used and unused logic gates. Because of the unused or deactivated functions associated with the FLBs, the use of FLBs 11 allow for engineering changes to the ASIC logic, even if not made until late in the design cycle. Moreover, the engineering changes can be effected through wire only changes, thereby avoiding timing problems generally associated with replacing and/or adding logic gates to the chip.
The use of FLBs 11 and their associated reconnection techniques to utilize initially unused logic allow changes to be easily made to FLB function even after the chip is manufactured. Wiring to and from FLBs 11 can stretch around chip 10 or the predefined region of chip 10 to enable easy connections to and from other standard ASIC function placed and wired on chip 10. Further, for ease of modifying the logic, it may be advantageous to build much of the original logic function on chip 10 from FLBs 11.
ASIC chips are generally requested by a customer who provides a “netlist” representing the desired logic for the application to be performed. The manufacturer of the chip can synthesize gate logic from the customer netlist, which can be formed in, e.g., register transfer logic (RTL). As the cost for FLBs can be greater than the cost for the simple logic it replaces, it can be advantageous to identify the logic likely to be subject to subsequent engineering changes.
Moreover, FLBs can also be synthesized, placed and routed in place of normal ASIC gates in areas of the chip identifiable as engineering changeable, e.g., in which new functions, changed functions, new hardware, or bug fixes have been made. In this event, such FLB gates may have some of inputs and outputs wired to perform the functioning logic of the normal ASIC gate and have additional inputs, and perhaps even outputs, that are at least initially appropriately tied off so the FLB functions as the normal ASIC gate it replaces. With this design, should a change, e.g., an engineering change, be required or requested by the customer or designer later in the design cycle, e.g., after placing and routing, the FLB can replace the standard logic so the function of the FLB can be easily changed or effected through wire only changes to the inputs and/or outputs of the FLB.
After a piece of logic or a specific area of the chip has been identified, the specific gates of this identified logic (risky or normal) or area can be isolated and synthesized or formed with FLBs or combinations of FLBs and normal logic gates. As noted above, while FLBs can be very complex functions, e.g., programmable MUX structures, NAND and NOR gates with and/or inverted gates (AOIs) with appropriately tied off inputs, they can change their function through a mere change of wiring. Moreover, it may be advantageous to arrange the tied inputs on metal layers where they can be accessed and, therefore, easily changed.
By way of non-limiting example,
As shown in the illustrations of the replacement FLB in
During design and implementation of a chip, FLBs can be used, e.g., to fix late-found bugs. Further, logic identifiable as highly probable for engineering changes or for replacement with FLBs can be tagged by, e.g., identifying synthesis directives in the source code (RTL); identifying hierarchy in design (named block, unit instance name); start and/or endpoint cones of logic; identifying dataflow and tagging non-dataflow; and tracking changes using logical equivalency checking between a chip netlist and a previous version of that netlist. Cones of logic changed are new function/bug fixes and therefore high-risk engineering change areas.
The tag function can be implemented through a synthesis tool or a post-processing tool. Thus, the logic can be tagged in the source code (RTL) with a direct synthesis tool to implement logic with FLBs until instructed otherwise, or the logic can also be tagged with a synthesis script, e.g., as a table in a post processing tool to replace logic with FLB books.
If no, a determination is made at step 304 whether the logic relates to a new function. In this regard, if the function has not been used before, it may be prone to error and, therefore, necessitate subsequent engineering changes. If a new function, the logic is tagged or flagged for implementation as an FLB at step 305. If not a new function, a determination is made at step 306 whether the designer is inexperienced, because inexperienced designers may be more likely than experienced designers to have mistakes in their logic. Therefore, if designed by an inexperienced designer, the logic can be tagged or flagged for implementation as an FLB at step 307, otherwise a determination is made at step 308 whether it is a late date for design change. This can be significant since the logic may not have been tested or not fully tested. If a late date for design change is identified, the logic can be tagged or flagged for implementation as an FLB at step 309. If not, a determination can be made at step 310 if verification coverage is low, as this can be difficult to test. If verification coverage is low, the logic can be tagged or flagged for implementation as an FLB at step 311. Otherwise, a determination is made at step 312 if bug fixes have been made in this logic. If such fixes have been made, the logic may be tagged or flagged to be implemented as an FLB at step 313. Thereafter, the tagged or flagged FLBs can be ranked according to a priority assignment at step 314.
A priority tag related to a relative importance of the tagged engineering changeable parameter can be utilized to identify which logic is more desirable to target for FLB implementation. Priority can be assigned to various methods for identifying where to use FLBs, e.g., priority may be inversely proportional to verification coverage, since logic areas with low verification coverage are more likely to require engineering changes; designer or verifier experience; interface regions, e.g., asynchronous, between major blocks, between new functions, etc.; dates and days of development and verification; history of the logic function, e.g., range of new function to previously used on other chips; and feedback from logic bug reports, which can be mapped across the entire logic. In the process of designing a chip, simulation typically progresses with the bug fix rate decreasing exponentially. If a log, kept for all bugs encountered and fixed, is mapped over the logic, it could be used as an indicator of entities needing implementation in FLBs and as an indicator of future bug likelihood.
The FLB gates can be synthesized where tagged or flagged and meeting constraints at step 315 or can be synthesized using normal library at step 316 to post process the FLB gates where tagged or flagged and meeting constraints. If the amount of logic targeted for replacement with FLBs exceeds chip area/power/timing constraints, it may be advantageous to back-off replacement of changeable logic with FLBs. However, synthesis would use these priority tags to make appropriate trade-off between area/power/timing and engineering changeability. Thereafter, the FLB netlist is optimized at step 317.
In step 315, the tags/flags can be put on the logic structures in the form of synthesis directives or synthesis script in comments in the RTL source code to synthesize the tagged blocks with FLB gate libraries as the target block, i.e., FLB targeted logic can be isolated into separate logic files, and synthesized using minimal blocks like NANDs and NORs. Further, the synthesis tool can be used to target a source or sink of cones by flop names identified by the logic designer to target the FLB gate library as the target library. With regard to cones, the start of a cone can be a point in the logic where the signal fans out, and the endpoints of the cone can be an output signal or latch. More particularly, inputs that can be cause an output to change can funnel down to a single point or expand from a single point.
In step 316, tagged blocks can be post-processed as FLB gate libraries. FLB targeted logic can be synthesized using minimal logic blocks, such as AOIs or muxes, with additional tied input lines. Alternatively, a targeted block can be synthesized with an FLB gate library as the target library, which allows changes in wire only. Further, the post-processing tool can be used to target flops/POs/PIs as a source or sink of cones of logic to replace simple NAND/NOR logic with FLB gates.
Moreover, according to a further aspect of the invention, data flow, e.g., adders, multiplexers, etc., can be identified, whereby non-data flow (control logic) can be targeted for FLB logic. As control logic is not typically easily identified, the method according to this aspect of the invention can identify data flow, e.g., from a netlist, and all unidentified logic can be considered as changeable logic.
Further,
According to still another aspect of the invention, a non-limiting example of a netlist comparison method is depicted in
When the delivered netlist is not the first drop, a logical equivalency testing (LEQ) tool is run at step 604 between the new netlist and the saved base level netlist in order to identify new function, changed functions, bug fixes, etc. At step 605, new cones for FLB substitution can be targeted, e.g., new function or part of a fix, and non-equivalent cones for FLB substitution can be targeted, e.g., new design (which may or may not be proven logic). A percentage of logic (based on chip constraints, design history, etc.) can be selected in the targeted cones for FLB substitution at step 606.
The circuit as described above is part of the design for an integrated circuit chip. The chip design is created in a graphical computer programming language, and stored in a computer storage medium (such as a disk, tape, physical hard drive, or virtual hard drive such as in a storage access network). If the designer does not fabricate chips or the photolithographic masks used to fabricate chips, the designer transmits the resulting design by physical means (e.g., by providing a copy of the storage medium storing the design) or electronically (e.g., through the Internet) to such entities, directly or indirectly. The stored design is then converted into the appropriate format (e.g., GDSII) for the fabrication of photolithographic masks, which typically include multiple copies of the chip design in question that are to be formed on a wafer. The photolithographic masks are utilized to define areas of the wafer (and/or the layers thereon) to be etched or otherwise processed. Moreover, the process as described above is used in the fabrication of integrated circuit chips.
The resulting integrated circuit chips can be distributed by the fabricator in raw wafer form (that is, as a single wafer that has multiple unpackaged chips), as a bare die, or in a packaged form. In the latter case the chip is mounted in a single chip package (such as a plastic carrier, with leads that are affixed to a motherboard or other higher level carrier) or in a multichip package (such as a ceramic carrier that has either or both surface interconnections or buried interconnections). In any case the chip is then integrated with other chips, discrete circuit elements, and/or other signal processing devices as part of either (a) an intermediate product, such as a motherboard, or (b) an end product. The end product can be any product that includes integrated circuit chips, ranging from toys and other low-end applications to advanced computer products having a display, a keyboard or other input device, and a central processor.
While the invention has been described in terms of a preferred embodiment, those skilled in the art will recognize that the invention can be practiced with modifications within the spirit and scope of the appended claims.
The present application is a divisional application of co-pending U.S. application Ser. No. 12/014,240, filed on Jan. 15, 2008, the contents of which are incorporated by reference herein in their entirety.
Number | Date | Country | |
---|---|---|---|
Parent | 12014240 | Jan 2008 | US |
Child | 13410386 | US |