John Hennessy’s Journey into Computer Architecture:
Pursuing a math major, Hennessy took a computer course and was captivated by the concept of bringing thoughts to life through computing.

Dave Patterson’s Interest in Microprocessors:
Patterson developed an interest in computing during high school and became intrigued by the potential of microprocessors as a new computing technology.

The Breakthrough Moment:
Hennessy and Patterson’s early work on RISC (reduced instruction set computers) technologies became a pivotal moment. They recognized the potential of simplified instruction sets, leading to higher clock rates and improved performance.

Simplification of Microcode:
Patterson and Hennessy identified the inefficiencies of using microcode in computer design. They advocated for simplifying the instruction set and eliminating the need for additional interpretation layers.

The Energy Efficiency Challenge:
John Hennessy explains the impact of Dennard scaling on power consumption in microprocessors. The slowdown of Moore’s law and the end of Dennard scaling have made power efficiency a primary concern in modern computing.

Transistor Efficiency:
Dave Patterson emphasizes the importance of using transistors efficiently, given the slowdown in Moore’s law. Efficient use of transistors will become increasingly crucial in future computing systems.

Memory Architecture and Storage Needs:
The rapid increase in data storage has put pressure on memory architecture. Hennessy acknowledges the limited success of new memory technologies over the years, except for the transition from magnetic core memory to semiconductor memory.

Flash Memory’s Impact on Storage Technology:
Flash memory has revolutionized storage technology, replacing magnetic disks as the primary backing storage in laptops. Flash memory is a semiconductor technology that offers significant advantages over magnetic disks, such as faster access times, lower noise, and reduced power consumption.

Emerging Memory Technologies:
Researchers are exploring new memory technologies that promise even better performance than flash memory. These technologies include magnetoresistive memory and 3D NAND. The viability of these technologies is yet to be determined, but they have the potential to significantly improve memory performance and capacity.

Challenges in Memory Technology Development:
Developing new memory technologies is a long and expensive process. It can take a decade or more for a new technology to transition from the lab to the market. Large organizations with deep pockets, such as semiconductor companies and tech giants, are typically the ones driving innovation in this field.

Startups’ Role in Memory Technology Innovation:
Startups may face challenges in competing with larger organizations due to limited funding and resources. To succeed, startups need to focus on developing niche applications or partnering with larger companies to bring their technologies to market.

Moore’s Law and the Need for New Technologies:
Dave Patterson highlights the importance of finding an intriguing technology, possibly from university spin-outs, to address the challenges posed by the slowdown of Moore’s Law. Achieving a viable replacement technology for RAM or standard scaling requires substantial funding, estimated to be several hundred million dollars, to demonstrate a first product.

Opportunities for Hardware Startups:
John Hennessy observes the shift in conventional wisdom, recognizing the opportunities for new companies in the hardware sector, particularly in the area of machine learning (ML) accelerators. Hennessy anticipates continued investment in innovative memory technologies if current ML accelerator startups prove successful.

Hardware-Enabled AI and the Role of Thirsty Colleagues:
Rashmi Mohan inquires about hardware-enabled AI, prompting Hennessy to discuss his experience working with the Google Brain Group. Hennessy emphasizes the thirst for computer cycles among AI researchers, coinciding with the slowdown of Moore’s Law. He stresses the value of collaborating with enthusiastic colleagues who demand the highest performance computing capabilities.

New Hardware Fields and Machine Learning:
Opportunities exist for inventing new hardware tailored for machine learning, given the slowdown in computer performance. The rapidly evolving nature of machine learning algorithms poses challenges in designing hardware that can efficiently handle these changes.

Minicomputer Revolution Comparison:
The current wave of machine learning hardware innovation is reminiscent of the earlier minicomputer revolution. Similar to the minicomputer era, a few dominant players are likely to emerge in the machine learning hardware space.

Security Concerns with Edge Computing and Data Collection:
Security becomes a critical concern with the proliferation of edge computing and data collection. Automotive advancements introduce new risks, such as the potential for hacking and causing catastrophic damage.

Challenges in Security:
The field of security has been lacking in progress, enabling criminals to engage in activities such as holding stores ransom. Proprietary instruction sets have limited attempts at security to specific companies, resulting in slow implementation of security features.

RISC-V and Open Source Hardware:
RISC-V, an open architecture, allows for collaboration and innovation in hardware security. Field programmable gate arrays enable the creation of secure hardware implementations that can be connected to the internet.

Need for Security Community to Create Secure Systems:
The security community needs to shift from criticism to creating secure systems. Collaboration between hardware and software experts is essential for progress in security.

Balance Between Security and Ease of Use:
Striking a balance between security and ease of use is a challenge, as seen in password management. Spam and phishing attempts have become prevalent, necessitating improved security measures.

Collaboration Between Experts and Institutions:
Fostering collaboration between security experts, hardware experts, universities, and industry is crucial for advancing security. Open source software has become mainstream, demonstrating the benefits of collaboration and shared development.

RISC-V and Open Source Hardware Collaboration:
The RISC-V community aims to bring the ethos of open source software to hardware development. This approach encourages collaboration among companies, researchers, and individuals worldwide.

Security as an End-to-End Function:
Dave Patterson emphasizes the importance of collaboration across various boundaries to achieve comprehensive security. Security must be considered from hardware design to software development and involve experts from multiple disciplines.

Complexity and the Weakest Link:
The increasing complexity of software systems has created vulnerabilities that can be exploited by attackers. Security breaches often occur due to weaknesses in one component of the complex software stack.

Need for Secure Systems:
John Hennessy highlights the need for creating secure systems from the ground up, rather than attempting to patch vulnerabilities after they arise. Formal verification and hardware formification are promising techniques for enhancing security.

Experimentation and Innovation:
Hennessy suggests experimenting with innovative approaches to security, such as changing the instruction set encoding regularly. Offering rewards for successful attacks can help identify and address vulnerabilities.

Future-Proofing Careers in Complex Systems:
Rashmi Mohan shifts the discussion to technical careers in complex systems. She seeks advice from Patterson and Hennessy on how practitioners can stay abreast of industry changes and future-proof their careers.

Keeping Up with Rapid Changes in Computer Architecture:
The field of computer architecture is constantly evolving, and it’s crucial for professionals to keep up with the latest changes to remain relevant. Online resources, such as courses and tutorials, provide convenient and accessible ways to learn and update skills.

Personal Responsibility for Skill Development:
It’s important for individuals to take responsibility for their own skill development and continuously learn to stay marketable in the field. Employers expect employees to maintain and enhance their skills, as the days of lifetime employment at a single company are largely gone.

Staying Ahead of Technology:
Professionals need to stay informed about emerging technologies and advancements to remain competitive and employable. Massive Open Online Courses (MOOCs) offer affordable and flexible opportunities to learn about new topics and technologies.

Balancing Work and Learning:
Striking a balance between work and learning can be challenging, but it’s necessary to invest time in skill development to stay relevant. Online courses and resources allow for flexible learning, enabling individuals to learn at their own pace and on their own schedule.

Importance of Practice and Application:
Simply consuming lectures and reading materials is not enough to master new concepts and skills. Hands-on practice and exercises are essential for developing proficiency in computer architecture and related fields.

Finding a Collaborative Partner:
Finding a collaborative partner can be beneficial for skill development and career growth. Look for individuals who complement your skills and strengths, and who share your passion for learning and exploration. Serendipity often plays a role in finding the right collaborative partner, but it’s also important to actively seek out opportunities for collaboration.

Finding Compatible Partners:
John Hennessy and Dave Patterson’s partnership was fortuitous due to their shared worldview and values, despite their different backgrounds. They emphasize the importance of finding people with whom you can get along, who are intelligent, interesting, and pleasant to work with. They discourage working with individuals who believe they are the smartest in the room, as such partnerships are often unproductive.

The Synergy of Collaboration:
Hennessy and Patterson’s partnership was cemented early on when they faced opposition in the industry for their innovative ideas. Their initial meetings revealed remarkable agreement on many aspects, leading to a fruitful collaboration. They naturally collaborate and help each other improve, enhancing their individual capabilities and achieving remarkable results.

The Importance of Teamwork:
Collaborative efforts are essential in the computing industry, as complex problems require diverse expertise and perspectives. Hennessy and Patterson emphasize the value of teamwork and the ability to work well with others. They acknowledge that writing their book would have been impossible individually and that it benefited significantly from their partnership.

The Power of Feedback and Collaboration:
Their writing process involved regular meetings where they shared ideas, provided feedback, and refined their work. The collaborative process accelerated the evolution of ideas and improved the quality of their work. They highlight the guilt-inducing aspect of knowing that a partner would review and provide valuable suggestions, motivating them to produce better work.

The Transformative Power of Partnerships:
Hennessy refers to an article in The New Yorker discussing the exceptional effectiveness of partnerships. The article suggests that partnerships can be exponentially more powerful than individual efforts. They emphasize the value of finding such partnerships and the transformative impact they can have on achievements.

Sports and Family:
John is a big sports fan, particularly wrestling and soccer, and loves spending time with his family, including his wife of 52 years and his kids and grandkids. Dave enjoys sports, culture, travel, and the arts, reads extensively, and values spending time with his family, especially as a grandparent.

John’s Book:
John Hennessy wrote a book titled “Leading Matters” about his experiences as president of Stanford University for 16 years. The book highlights his emphasis on learning from the mistakes of others rather than making them himself.

Excitement in Technology:
John is enthusiastic about open instruction set architectures, bringing open source software innovations to open source hardware through the RISC-V effort. He is also excited about accelerating machine learning, considering it a revolutionary technology like the internet and the microprocessor.

Machine Learning Revolution:
Dave agrees with John’s excitement about machine learning, emphasizing the need for new algorithms, innovation in methods, and the computing power to support them. He believes the field of machine learning is experiencing an explosion of progress, similar to the RISC revolution of the 80s.

Fascination with Computer Science:
Both John and Dave find computer science fascinating due to its occasional explosions of progress in specific areas, like the RISC revolution and the current machine learning revolution.