As Intel attempts to blaze past TSMC and Samsung and sets its sight on becoming the world¡¯s leading chip manufacturer by 2025, achieving hardware supremacy will be just one half of the journey. Hardware plus software is the ultimate winning combination, as they say, something that Intel¡¯s Greg Lavender knows about only too well.
¡°Software is a great democratizer, it creates global opportunity for millions of people and enables our society,¡± says Greg Lavender, senior vice president, chief technology officer (CTO) and general manager of the Software and Advanced Technology Group (SATG) at Intel Corporation.?
We only spoke over the phone, but soft-spoken Greg Lavender¡¯s passion for all things technology is palpable, especially in the software realm. He gets it, given his more than 35 years of experience in software and hardware product engineering across companies like Sun Microsystems, Cisco, Citigroup and VMware. Lavender along with CEO Pat Gelsinger are two battle-tested software industry veterans at the helm of Intel¡¯s push towards a software-first mindset.
¡°The software industry is just an exciting place to be, and the breadth and depth of talent we have here at Intel is just incredible,¡± according to Greg Lavender, who only joined Intel in June 2021. Within that time, he has prioritized some key software initiatives internally like Intel¡¯s oneAPI ecosystem, AI and ML software advancements, while also ensuring all of Intel¡¯s products that come out next year have high quality BIOS and firmware software.
Of course, I ask Greg Lavender about the rise of RISC vs x86 computing and whether Intel¡¯s hardware is seemingly underutilized because of its large software ecosystem, and he has some interesting thoughts on the matter.
¡°Intel¡¯s pretty pervasive,¡± underscores Greg Lavender, ¡°I'd recommend checking up what percentage of servers powering AWS, Google Cloud and Microsoft Azure are x86 architecture and how many of them came from Intel.¡±?
Also Read:?Sneak Peek Into Intel Labs To See What's Cooking In 2021 And Beyond
"There's a lot more Windows and Linux laptops in the world running on Intel¡¯s x86 architecture than ARM or AMD,¡± he continues, without naming Apple. ¡°We like a good competitive fight and competition is good. It sharpens the mind; it focuses the effort and the resources.¡± More than the software it all comes down to the workloads, believes Lavender, which is what Intel¡¯s chasing in a way.?
"We have amazing amounts of data and information about software workloads, how they need to perform, how they need to be accelerated, and our roadmaps take into account that kind of deep knowledge down to the instruction level, to the memory access level for all sorts of optimizations,¡± he says, explaining how a huge swathe of open source runs on backend servers powered by Intel¡¯s chips, ditto that for Microsoft workloads on Windows platform.?
Far from resting on any laurels, Lavender believes Intel¡¯s going to kick on and continue innovating more aggressively than ever in the software space. "We're going to keep innovating and we're going to keep leveraging all the assets we have, including all of our software, recommitting our focus to the developer community, particularly the open-source community,¡± he says, ¡°Pat Gelsinger has hired me as an experienced long-time software guy to lead the charge, and that¡¯s what I¡¯m focusing on at the moment.¡±
One of the main insights provided by Greg Lavender in terms of Intel¡¯s software innovation history is how they¡¯ve been poor at evangelizing some of their behind-the-scenes work, especially when it came to enabling open-source software workloads.
"For a long time, Intel has been the stimulator and enabler of open-source software in the industry. With respect to Linux, we contributed the latest kernel enhancements into kernel.org for our next generation hardware coming up next year, we also recently pushed changes to the GNU compiler system which tickles the advanced vector instructions for accelerated workloads, we¡¯ve worked with Google to get Intel¡¯s oneDNN (deep neural network) accelerators into TensorFlow workloads,¡± elaborates Greg, while stressing on the need to re-engage in a very public way with the open-source community.
¡°From vector neural network instructions which parallelizes machine learning and inference capability in AI workloads, ISVs running their proprietary software on Intel platforms, and crypto algorithm acceleration on x86 architectures, this is just some of the work we¡¯ve been doing behind the scenes,¡± summarizes Lavender. His job is to highlight it and help define the principle of future growth as far as Intel¡¯s software initiatives are concerned, which Greg Lavender insists begins with seeking software and technologies that promote the growth of highly secure open-source cloud platforms.
More machine learning and inference related workloads are moving out to the edge of the network because ¡°that¡¯s where the data is¡± according to Greg Lavender.
¡°Everything from your laptop to smartwatch, your phone and every other device that connects to the Internet has some sort of inference capability built into it based on machine learning models that might have been trained in the cloud. That capability¡¯s increasing very fast because that¡¯s where the real interactions are happening,¡± says Lavender.?
Even though he believes the software workloads at the network edge are still in their nascent stage, use cases are exploding. ¡°The next few years are going to be quite dramatic as compute moves further to the edge in terms of processing power demand,¡± Greg Lavender says. How Intel accelerates these workloads on their processors and accelerators at the edge is a big focus area for Greg Lavender¡¯s team.
In contrast to the edge, everything on the cloud remains an opportunity to do more for Greg Lavender¡¯s team at Intel. Workloads are analyzed and optimized within Intel¡¯s performance labs, Lavender explains, before the gains are passed forward to the open-source communities and ISVs.
¡°From Kubernetes to Hadoop to Kafka, the cloud is full of unique AI / ML workloads and code libraries full of potential, we are profiling the most common open-source software to major ecosystems and tuning them up and putting in our open-source acceleration technology to make sure everything runs extremely well in the cloud on our whole family of Intel processors,¡± assures Greg Lavender.
How can a conversation on software ever be complete without mentioning India, right? I ask Greg about his impression of India's developer ecosystem, and if he has any advice for people who are looking to upskill themselves for software-based career opportunities that are cropping up in a frenzy in the post-Covid world. His response is quite revealing.
¡°India is a hot market in terms of software, it has been for quite some time now, and it¡¯s a very healthy industry,¡± says Greg Lavender. ¡°The talent and innovation that I¡¯ve seen from my teams and companies I¡¯ve worked with in the past in India is incredibly powerful. Without Indian software engineers the world would not be where it is today,¡± Greg emphasizes, before suggesting how we should all be happy and glad to have a strong economic and technical ecosystem in India.
Do people need a computer science degree to work at Intel and in the software services industry? ¡°Your code is your resume,¡± says Greg Lavender. ¡°If you¡¯re an open-source software developer or an upstream committer for some effort in the open-source community, your GitHub account, your product, the code you¡¯ve written, the code you¡¯ve contributed to the project you made, all of that is your best resume.¡±
¡°It¡¯s the power of what you think is the power of what you do. Your ability to work with others, contribute to each other through code is how software is making an impact on a global scale,¡± he says. That¡¯s the best calling card for software-based jobs.
For a lifelong software guy, my final question to Greg Lavender is meant to be an anecdote in the age-old software vs hardware debate ¨C which one matters more? Once again, Greg¡¯s response is quite insightful, bordering on the philosophical.
"Hardware and software together conserve our energy, allowing us to use our minds and our ideas to create interesting things, and one cannot work without the other, they have to go together,¡± Greg Lavender says emphatically, before continuing.?
¡°At the very least, software is a lubricant that keeps the machine going. Software is the soul of the machine, to borrow a quote, and together software and hardware have allowed the human potential to flourish to do amazing things. I've had an exciting life for almost 60 years, and because of my profession I¡¯ve spent two-thirds of it working in the software industry. I¡¯ve never been bored, there¡¯s never been a dull day, and that¡¯s what¡¯s exciting about the software landscape,¡± concludes Greg Lavender.