• Skip to main content
  • Skip to search
  • Skip to footer
Cadence Home
  • This search text may be transcribed, used, stored, or accessed by our third-party service providers per our Cookie Policy and Privacy Policy.

  1. Blogs
  2. The India Circuit
  3. CDNLive India Keynote: Qualcomm On 5G And More
Madhavi Rao
Madhavi Rao

Community Member

Blog Activity
Options
  • Subscribe by email
  • More
  • Cancel
5G
artificial intelligence
CDNLive India
CDNLive
IoT
machine learning
Qualcomm
mobile
7nm

CDNLive India Keynote: Qualcomm On 5G And More

13 Sep 2017 • 4 minute read

CDNLive India concluded last Friday and what an event it was! With 87 paper presentations, six keynotes, an exhibition area, photo booth, Cadence technology pods and more, CDNLive once again proved to be one of the premier industry conferences.

Both days of CDNLive started out with high-powered keynotes, and my next few blogs are going to focus on the guest keynotes. This blog is about Venugopal Puvvada’s keynote which took place on Sep 7, the first day of CDNLive India. Venu, as he is known, is vice president of engineering at Qualcomm.

Venu’s keynote talked about designing a 7nm 5G chip in India. Qualcomm is, of course, at the forefront of 5G wireless technology innovation. Venu started by saying that while there many 5G SoC design challenges, it is mobile that’s driving some of the latest technology nodes, 7nm in particular.

And that, he said, is an area chip firms need to work together with EDA partners to crack, with special attention paid to machine learning (ML). Talking about how Qualcomm will design a 7nm 5G chip in India, Venu remarked that with over 7 billion people in the world, there are 3.5 billion unique mobile users, of which 2 billion are smartphones users – all quite staggering statistics.

Of the broad portfolio of technologies that will drive the future of mobile – power management, Bluetooth, modem, CPU, GPU to name a few – the newest entrant everyone is talking about is machine learning.

AI, ML and DL

Calling it an “exciting domain”, he described what he believed is the difference between artificial intelligence (AI), machine learning and deep learning.

AI is about how intelligence is simulated using a machine – the overarching domain of AI. In the case of ML, Puvvada observed ML can tell us the difference between an orange and an apple by looking at multiple pictures, using multiple algorithms. But deep learning is something that will tell if the apple is from Kashmir or California, the orange from Nagpur or Singapore.

According to Venu, ML should be done at the source, not in the cloud. He said, “If it’s a mobile, do it in the mobile, if it’s a self-driving car, do it at the car level, rather than at the cloud. Otherwise, you really increase the load on the cloud. Instead you can move a lot of this to the source.”

As we know, a future where millions of devices are talking to one another is already here. Already a massive amount of data is being collected and the future will be about handling all this data.

3G vs 4G vs 5G

Venu called 5G a unifying connectivity fabric which addressed enhanced mobile fabric, mission critical services and massive IoT.

A bit more about 5G and mission-critical services: as Venu saw it, the difference between 5G and 4G is that 5G is about Gbps data speed and mission-critical services. Such services are about high reliability – such as in automotive when your self-driving car is talking to another self-driving car, causing the need for very high reliability along with a good amount of data to be shared.  So handling and supporting mission-critical devices is the critical differentiator for 5G, especially with the massive IoT wave that we are expecting.

SOCs for mobile and IOT – Designing at 7nm

Talking about mobile and IoT is talking about hundreds of millions of devices, necessitating utilizations in blocks and chips, a challenge well understood by physical design engineers. Thermal power performance efficiency is another issue. Yet, everyone wants to be the first to market, with differentiation. Doing this, to him, is the biggest factor to keep in mind when doing chip design!

Puvvada pointed out to multiple challenges for the mobile SoC - cost, power, performance, time or schedule, heat (thermal issues), the technology nodes to be used, what mode gives the best cost options and the best power-performance tradeoff, and wondered aloud about what would be a strategy to tackle this multitude of challenges.

To his mind, the first is SoC and system-level expertise, looking at things from a system-level perspective. The next is world-class IP in the building blocks, whether they be standard cells, memories, or anything else, because this is the key differentiator. World-class EDA tools are also essential, he added, saying that Qualcomm had an excellent partnership on with Cadence in this regard that helped it get ahead of competition.

An interesting change happening in the last couple of years is that mobile is driving new technology node adoption. “In the previous years we use to be very reluctant to get on to the new technology nodes in a big way… we could wait one chip and then do it. But now we are straight away jumping into that,” he said.

The last point he touched upon was automation, of which ML is the most interesting facet, but which, in his opinion, has seen indifference from the chip industry.

 Venu concluded with a call to action: “I know [ML] is changing in a big way, we’ve suddenly realized we’ve lost time and need to jump onto it. We need to join hands and crack this problem. Embrace machine learning and deep learning.”

© 2025 Cadence Design Systems, Inc. All Rights Reserved.

  • Terms of Use
  • Privacy
  • Cookie Policy
  • US Trademarks
  • Do Not Sell or Share My Personal Information