A new way to design custom silicon
Tthe success of Arm Flexible Access in providing easier access to a wider range of Arm IP, meant a more effective way for a broader range of users to understand how Arm IP works, the key benefits, and how the different IP components work in an SoC design - speeding up the design process (which can take up to 12 months) and enabling more innovative chip designs to be created. I was lead product designer and business architect working with domain experts and small development team on a ground=breaking customer discovery project to understand the SoC design process, customer journey, needs, pains, and opportunities to create new innovative enterprise software & tools to help significantly reduce (from 12 to 3 months) the SoC design process whilst increasing the effectiveness of the designs.
ROLE: LEAD PRODUCT DESIGNER
DESIGN PHASES: DISCOVER + DEFINE
RELEASED: APRIL 2022
CLIENT: ARM
PLATFORM: MOBILE, WEB
DESIGN TOOLS: ADOBE + FIGMA
PROTOTYPING: ADOBE + ZEPLIN
VISUAL DESIGN: ADOBE XD + FIGMA
Arm Flexible Access enabled easy and quick access to many more of the latest Arm IP products all via a subscription and membership model. This allowed startups and smaller OEMs to use their own silicon, realizing a products full potential whilst protecting IP, reducing costs, and increasing security. This was most evident in the automotive industry where advances in new technology adoption, and drive for autonomous vehicles, has seen a growth in new startup investment, and with access to greater CPU/GPU power rapid improvements in powertrain, connectivity, and ADAS autonomous systems were possible.
DISCOVERY
A key part of the project was formulating the design methodology and framework to achieve the best outcome. As this was looking at new innovation around existing way of working I decided to use Outcome Driven Innovation (ODI) and Jobs-to-be-done framework to identify unmet needs, and design thinking methodology to create humanistic concepts.
To collect the required data meant gathering sufficient quality information to build a detailed picture of the end-to-end SoC design process. This required ethnographic, observational, and industry research conducting 100's of interviews with the many different specialists involved in the process of SoC definition, design, build, test, and implementation. Using this data I created a JTBD framework which then allowed me to analyse the data along with customer needs to define at set of desired outcomes. From this I could segment the market around the unmet needs to create a segments of opportunity map with focus on differentiated and disruptive strategy, and a clear set of problem statements for each of the key personas.
Using data from the user research, JTBD, and ODI matrix I could pull this together into a detailed customer journey map; presenting the technical challenges involved, and highlighting the moments that matter, pain points, and unmet needs. A key paint point and opportunity (unmet need) was identified in the ear/y stages of SoC definition and design where the IP to be used in the design was selected. This often took place with limited understanding of the IP configuration, potential alternatives, and impact on system performance of any configuration/design changes. This means SoC designs were often over or under engineered and any virtual performance evaluations that we done were both costly and time consuming, so they rarely tool place.
DEFINE
Using the outcomes of the discovery phase; JTBD, personas, ODI segment matrix, journey map I facilitated a mini (3-day) design sprint to focus on generating ideas, theme, and concepts around the identified pains points and unmet needs in the pre-study > requirements > initial investigation phase of the SoC design process.
The culmination of the design sprint and additional sketch-hack workshops was a new product concept I named; 'IP Architect' where the design is created using a cloud based tool which would allow, even those without SoC design skills, to select, build, test, and evaluate IP options with a range of template designs to act as a starting point. This required a completely new cloud based approach in bringing all the key tasks (Jobs to be done) into a user flow that could accommodate both expert and novices needs.
The concept was then tested with both subject matter experts and novices to establish the concept desirability and viability, and then refined to create a lo-fi prototype with a set of wire-frames around the key tasks. The prototype was tested in task based 1-2-1 user interviews and the corresponding feedback and analysis applied to the designs. In parallel the technical and security requirements around delivery of a cloud based solution were defined, and requirements for a Proof-of-Concept (PoC) software product were created. Key wire-frame designs were built and tested to support the PoC which would help to establish the feasibility of the concept.
DEVELOP
The PoC software product was able to establish the technical feasibility of designing IP in the cloud and form the basis of an MVP. The prototype testing enabling the concept to be refined and interaction design solutions to be created and tested using new design system components. This resulted in a series of design mock-ups and a HiFi prototype that was used to create both internal and customer awareness (video demo & promo) by showcasing the concept at key events, collaborating with potential early partner customers, and support the MVP development process. On successful feedback to the PoC and design concept from customers, partners, and the IoT / Automotive business area, the project was transitioned into the Scaled Agile Framework Environment (SAFE) process that all committed IT development projects are managed and deployed. Within SAFE the specific product development teams are organized into squads that form tribes working on specific areas of the business. As result I became the product design lead for the IP Architect squad. After iterations of the MVP and successful customer beta trials, the first phase of the solution 'IP Explorer' was launched in 2022.
Tap / Click to view prototype demo
GALLERY
–
