Blue Catalyst
Blue Revolution in VLSI ad System Design
By Swapnil S.,



Tuesday Sep. 08, 2009

Factory model in the Semiconductor World

1. Overview:

Semiconductor Industry is different from the Software Industry in many aspects.Manufacturing, Latest Technological Innovations,Time to Market crisis,understanding the nerve of the customer and speculating the course of technology are few key terms in the Semiconductor Industry, which are different from Software Industry.

Semiconductor Industry in the layman’s terms is indeed the amalgamation of both Software and Hardware flows, processes and practices.

Software Design and Development(in Fabless Companies) is done when one has to actually design a multi million gate chip and verifying it for lowest power consumption,least possible area covering and to identify the speed at which it has to operate, by using some EDA tools. Once the process is done the intermidiate floorplaned and routed netlist(a complex architecture made of digital/analog gates and flip flops) is passed on to the Foundry .

Hardware Manufacturing(in Merchant Foundry) is done taking the input of the netlist and making an Integrated Chip out of it, in a Foundry

Although many companies continue to both design and manufacture integrated chips(achieving efficiency through vertical integration), these Integrated Device Manufacturers (IDMs) are not alone in the marketplace. Economic forces have led to the existence of many companies that only design devices, known as fabless semiconductor companies, as well as merchant foundries that only manufacture devices under contract by other companies, without designing them.

IC production facilities are expensive to build and maintain. Unless they can be kept at nearly full utilization, they will become a drain on the finances of the company that owns them. The foundry model uses two methods to avoid these costs: Fabless companies avoid costs by not owning such facilities. Merchant foundries, on the other hand, find work from the worldwide pool of fabless companies, and by careful scheduling, pricing, and contracting keep their plants at full utilization.
This paper talks about applying the factory model in to the processes, operations, methodologies, flows, human resources, orgaisation structure for the fablesss Semiconductor Industry.

2. What is Factory Model?

A factory is defined as a organization structured such that projects are built in discrete work centers. Work centers generally represent, or specialize in, certain disciplines such as architecture, design, construction, integration, test, maintenance, packaging, release, etc. Much like a true manufacturing facility, software/semiconductor factories require clearly defined product creation and management processes. By utilizing the fundamentals of industrial manufacturing - standardized components, specialized skill sets, parallel processes, and a predictable and scalable consistency of quality – a true Factory is thought to achieve a superior level of application assembly even when assembling new or horizontal solutions. Industrialization of the process can provide benefits in terms of economies of scale, geographic distribution, load leveling, and rigorous product and process control.

3. Why do we need Factory?

  • Eliminate wasted time and resources
  • Build quality into workplace systems
  • Find low cost, reliable alternatives to expensive/new technologies
  • Streamline the creation of business processes that need to be “perfect”
  • Build a learning culture centered on continuous improvement

What if this was the achievable case within the semiconductor industry? What kind of savings/profit would that equate to? Would this sustain the drumbeat of Moore’s Law for years to come?

4. Requirements of an efficient factory

In order to be effective and efficient a "Factory" requires considerable organizational, process and project discipline. These disciplines typically involve: high quality requirements gathering, derivation, management, and packaging; rigorous program management; effective resource management and allocation; proven/reusable components; and excellent project scheduling and control. Effective factories rely upon the use of well balanced and consistent processes and productivity tools that allow existing components, applications, and systems to be easily consumed, integrated, and orchestrated in the construction of software deliverables. Since the factory approach is based on the integration of many organizational and process components, these organizations require well documented and trained processes.

5. Building blocks of a Semiconsuctor Factory :Looking from STRATEGIC APEX viewpoint

Many have adopted this model to their principles of lean in order to build additional rules and principles to be used more generically (away from the automotive industry in general and Toyota in particular). One fundamental difference between Toyota and others is the significant involvement of everyone in the improvement process .if we operate as a lean system, we can have everyone in the organization focused real-time on solving problems and driving waste out of the organization”. Principles of factory production include:
  • Teamwork
  • Communication
  • Efficient use of resources
  • Continuous improvement

Standardized way of thinking at Toyota starts with four rules that have formed the foundation of all of its innovative tools and concepts. The four rules are:
  • Structure every activity
  • Clearly connect every customer/supplier
  • Specify and simplify every flow
  • Improve through experimentation at the lowest level possible towards the ideal state .After one has built the “base” on the rules, the following principles apply (and thus collectively build a factory “house”)

6. Principles in Semiconductor Factory:

  • Directly observe work as activities, connections and flows
    • Structure, operate and improve your activities, connections and flows
    • Understand current reality requires deep observation

  • Systematic waste (overproduction, transportation, motion, inventory,waiting, over-processing, product/process defects) elimination
    • Connect to your customer and always add value
    • Relentlessly pursue systematic waste elimination</ul
    • Establish high agreement of both What and How
      • Standardization is the foundation of continuous improvement; create high agreement and no ambiguity
      • Sustainable change happens only at the systems level – lean is rules, not tools</ul
      • Systematic problem solving
        • Seek every problem as an opportunity to focus on the ideal state
        • Decision-making at the point of activity</ul
        • Create a Learning Organization
          • Create frequent points of reflection – be a learning organization
          • Leaders must be learners and teachers


        7. Factory Impact on the Operating Core

        • Rules-based Code Generation

          There are a number of tools available that aim to improve the software development process by having the developer model the software using the UML. The level of support provided to take the model and turn it into the target implementation varies considerably from no support all the way to full translation of the model and its content. Scripts/programs can be used to generate the templates of the configurations and the job will be to just plug,connect them to modules i.e. bind and port map them.

        • Building executable UML models

          Allows you to verify system behavior early in the development process--reducing bugs and finding bottlenecks early in the game. See how a UML/System Verilog model allows to focus on meeting requirements - and prove they are met through execution. Understand the benefits of translatable UML through the repeatable generation of clean, readable, and highly efficient C/HDL code – tailored to match your real-time constraints, free from common programming errors such as issues with synchronization and threading.

        • Closing the Loop in Testbench Automation

          Existing testbench techniques offer various benefits.However, once a testbench is initiated, it runs open-loop, generating results and then reporting them to an engineer. In turn, the engineer analyzes the results, makes some modifications to the system, and runs it again - and then repeats the process. The process runs open-loop, requires human intervention, and the iterations can span weeks, or even months.

          Most verification engineers consider the "learning testbench" to be a vision of the future. However, a new advanced closed-loop testbench automation system "learns" from both the DUT and the testbench modules during simulations. By combining concepts previously associated with compiler test automation and logic design synthesis, along with some innovative, recently patented technology, directed testing and constrained random testing with algorithmic testing - a learning-based system that actively targets desired results, rather than merely reporting them.

        • Solving the IP Re-use Paradox:

          How to Re-use Module-level Testbenches at the System Level As an alternative to the use of traditional programmatic approaches to design and verification, introduce the use of an automata-based approach. The benefits of such an approach extend significantly into and design and verification of complex system designs.


          Posted by Swapnil S. on Tuesday Sep. 08, 2009 | Add a Comment




RSS Feed

Add this blog to your RSS newsreader!

Archives

About the Author

About me: Swapnil has almost 11+ years of verification experience for System Design, ASIC/SOC and IPs. He has been exposed to various verification and Validation techniques and methodologies like Formal Verification, Assertion Based Verification, Constraint Driven Random based and Coverage Driven Verification, VMM,URM,AVM and now OVM i.e. Open Verification Methodology. Swapnil has been providing consultation to various organizations through his expertise in - System Verilog, Verilog, and VHDL ,ASIC-SOC-FPGA,Memory,Peripheral verification/ and System Validation.Swapnil has worked on the domains like Serial Interface, Wireless, SoC, Avionics and- Processor.Apart from Chip design aspect -Swapnil also work and consults on System Design -Integration, Debuggers, Tools, Scripts, Oscilloscopes, Software Drivers, Linux, DLLs , Applications and Firmware and ROM development.Other than the above mentioned subjects Swapnil writes and thinks about psychology in performance management, astrophysics, metaphysics, string theory, underpinnings of belief systems, mythology, religion, medicines, fashion, oriental sciences and arts ,food recipes, media, advertising, direct marketing, exorcism, politics, society, environment, economy and travel-travel and lots of travel.