July 3, 2006 -- When the IP industry emerged in the mid nineties, chip designers had high hopes that reusing highly complex IP would let them close the gap which had opened between the gate capacity of modern process nodes, and the design complexity state-of-the-art design tools and methodologies were able to produce. After 2000, faster product cycles and the more aggressive time-to-market requirements for new products, more narrow market windows, as well as limited resources in most companies doing chip design increased the motivation of chip designers to speed up their design process by reusing IP, especially in areas outside of the core competencies of the design teams.
It quickly turned out though that successfully integrating IP into a chip was far more complex than putting a component on a board or using a standard cell library or a memory. Quality issues in complex IP purchased from 3rd-party providers in many cases ended up seriously jeopardizing the timeliness of chip tape-outs if the issues with the IP were found too late in the design process. In cases where designs with faulty IP had taped out due to the high mask costs of modern process nodes, entire companies found themselves at risk if their design ended up not working and a re-spin of the design was required. Complex legal issues had to be sorted out between the 3rd-party IP provider and the company using the IP in these cases.
The industry quickly realized that quality was the single most important issue hindering wide-spread IP adoption. A means for assessing the risk of IP which is being considered for use needed to be found. In fact, quality and associated risk started to be considered more important criteria for IP selection than the traditional IP selection criteria such as features, performance and price.
As the IP industry matured, competing IP providers started to offer similar types of IP, especially for commodity IP. Even though more choices is a good problem to have, it brought up the issue of how to select the right IP. Whereas characteristics such as performance and features are relatively easy to compare between different types of similar IP, getting insight into the differences with regards to quality and the risk associated with the IP from the different vendors proved to be far more difficult.
Users of IP found themselves spending a significant part of the development cycle evaluating IP before making a decision on which IP to purchase and adopt for their designs. Even if several IP from different vendors were being evaluated in parallel, there was no guarantee that hidden quality issues could be spotted during that evaluation period. Also, IP providers are typically very hesitant to hand over a significant part of the full IP deliverables just for evaluation purposes without having a commitment for purchase from the potential customer.
The VSI Alliance (VSIA) recognized the need for a standard for IP quality and developed the VSIA QIP Metric, which systemizes and standardizes quality aspects of IP. QIP builds on top of the Reuse Methodology Manual and OpenMore issued jointly by Synopsys and Mentor.
The QIP Metric is essentially a comprehensive set of questions the IP provider is expected to comment on. Each requirement has a weight to distinguish more crucial from less important requirements. Based on the answers given by the IP provider, a score is built, which is the summary of all requirements that have been met. Using a standardized set of questions across different IP and generating scores for each of them makes IP quality objectively measurable. IP Scores from different vendors can be compared against each other to identify the most promising candidates before making a decision on which IP to evaluate in more detail.
The VSIA QIP also makes sub scores for areas such as Ease of Reuse, IP Maturity, Documentation Quality, Verification Quality, and others available. The scores, sub scores and other information are presented in a summary sheet, which makes it very easy for IP users to digest the quality situation for a particular piece of IP they’re considering using, often within a few minutes. While the scores themselves provide an indication of the general quality of the IP, the QIP is most useful as a communication mechanism between the IP provider and end user.
Based on the answers provided in the QIP, the IP user can immediately identify where the IP being assessed might have quality weaknesses and can then drill into the details of the particular requirements which are not met to determine whether this needs to be considered an issue. Some quality aspects may not be show stoppers in certain end applications, but being able to quickly spot areas where the quality of IP might be less than perfect can be the starting point for follow-up discussions with the IP provider. This can help clarify what has and what has not been done in particular areas to ensure high quality IP, and the ramifications on its intended use. The IP user can then use this information to plan for back-up activities on his/ her side to make up for quality issues the IP might have. This might be a valid option in situations where a certain type of IP is only available from one vendor and where the only alternative to not using the IP from this vendor would be to design and verify this functionality in house. Walking away from using IP from a vendor due to quality issues in certain, isolated areas highlighted by the VSIA QIP might in reality not be an attractive option, especially if this functionality is not the core competency of the IP user, and a reasonable mitigation plan can be developed.
Recently, VSIA issued rev. 2.0 of QIP. Based on feedback from users and to help promote widespread adoption of the metric, the VSIA made significant changes to QIP. The QIP requirements are now presented in a format which makes it easier not only for the IP user to quickly assess the quality status of IP, but also for the vendors to complete the metric for their IP. Previously existing requirements have been put into two main categories, IP Integration and IP Development. Requirements in IP Integration are directly relevant for the successful and efficient integration of the IP into the chip context. The IP user will typically be interested in the specific answers for each requirement in this section. Requirements in IP Development contain guidelines for IP developers to help ensure high quality IP. Even though following these guidelines will result in higher quality IP and ease its maintainability, the IP user will typically not be that interested in the specific answers to individual questions in this area, but care more about the overall summary or score.
With QIP 2.0, a new category of questions and requirements was introduced called Vendor Assessment. Whereas the requirements in IP Integration are of most interest for the engineer integrating the IP and the requirements in IP Development are of most interest for the engineer designing the IP, the requirements in Vendor Assessment will be most relevant for managers of companies considering the IP purchase. This section will help determine which IP vendors to engage with before investing significant time and resources into performing an in-depth evaluation of the IP.
The requirements in Vendor Assessment look at a company providing IP or an IP organization within a larger company as a whole. That’s why Vendor Assessment will typically only be filled out once per company or organization, whereas IP Integration and IP Development will be filled out for each IP. The requirements in Vendor Assessment are at a higher level than the requirements in the other parts of QIP and address aspects such as processes, verification methodologies, quality assurance, support and distribution infrastructure, consistency, and liability. Vendor Assessment even touches on more subjective requirements such as Vendor Confidence, which tries to assess how likely it is that a particular vendor will survive over the years, be able to give support once parts are being manufactured, and offer an IP roadmap for future product generations.
Future revisions of the VSIA QIP Metric will also address hard IP, Verification IP and will have a stronger focus on deliverables for IP, which are only implicitly covered in QIP 2.0.
In summary, VSIA QIP 2.0 is a powerful tool to objectively measure the quality of IP. It creates visibility into various relevant aspects of IP quality and enables informed discussions between IP provider and user before making a purchasing decision. VSIA QIP 2.0 can significantly reduce the time required to evaluate the quality differences in IP offered on the market place. The newly added Vendor Assessment requirements help the IP user to quickly identify IP providers worth engaging with for more detailed discussions, evaluations and ultimately a purchasing decision.
By Robert Bethge and Kathy Werner.
Bob is with LSI Logic Corp. and Kathy is with Freescale Semiconductor, Inc.
Go to the VSI Alliance (VSIA) website to learn more.