AI/ML

Can specialized AI chips solve AI product challenge?

Posted on

Xilinx organized a session titled: Can specialized AI chips solve AI product challenge?

Nick Ni, Director of Product Marketing, AI Software Ecosystem, Xilinx, said that there is projected growth of $30 billion approximately by 2023, as per Barclays Research Co. There is an era of domain-specific architecture. AI, deep learning, etc., are far from mature. There are DSAs all around, such as FPGA, Zynq, ACAP, etc. DSA in AI has custom data path and custom precision.

Today, AI apps are everywhere. They are in classification, object detection, segmentation, speech recognition, recommendation engine, anamoly detection, etc. AI is also evolving rapidly. There are the AlexNet, GoogLeNet, and DenseNet. The AI innovation is now pushing for new types of DSAs.

There are trends toward depth-wise that will need flexible and large device memory DSAs, with fixed and limited device memory as GPUs and typical AI chips. You can also have custom layers with dedicated hardware accelerators, which lead to host CPU. These are just two examples. You can use sparse neural network (NN) for visual search. The face search is generally dependent on latency. There is also the binarized neural network. DSA with 1b can fit order of magnitude more MACs than 8b.

There is the use of adaptable hardware with DSA. You have the world’s most responsive AI in MLPerf. Xilinx is just ahead of Intel in performance. Xilinx also achieved the peak TOPS in MLPerf. The power of DSA is immense. AI application can help detect things, etc. You have to take on hundreds of camera streams and decode. You have to take on the computer vision, as well, to decode. In terms of ADAS/AD, you have to take into action motor control. GPUs and typical AI chips fail to accelerate the full system. DSA is needed to power AI.

As an example, Pony.AI is using Xilinx AI with DSA for autonomous driving. Subaru is using it for ADAS. SK Telecom is using it for speech recognition. Ouster is using it for the LIDAR.

Xilinx is also enabling AI scientists. Vitis AI is a deep learning acceleration stack. There are 70,000+ downloads of Vitis and Vitis AI tools since launch in November 2019. Xilinx has had 100+ design wins. Vitis AI is also integrated with Microsoft ONNXRuntime for enabling AI inference. Xilinx has the Adaptive Computing Challenge going on with 1,000+ developers and start-ups registered.

AI productization requires the DSA to meet the production needs for performance, latency and cost. Only adaptable hardware can achieve 100 percent hardware utilization by building DSA for the exact workload, proven with MLPerf submission in v0.7. AI app is much more than AI-> DSA for non-AI acceleration. Now, an adaptable hardware solution enables the AI and software developers with Vitis, supporting TensorFlow, Pytorch, etc.

This entry was posted in AI/ML, global semiconductor industry, Semiconductors, Xilinx and tagged , .

Building up industry knowledge! For good!!

Posted on Updated on

This is an old post from 2007. I am re-posting it, as per some requests. I have made minor changes. It is actually very useful to anyone who cares to read.

Fate has been most kind to me. Fate has also been most kind to me in terms of bringing me in touch with several good folks who appreciate whatever little talent I possess.

That’s me, in front of the JFK Memorial, in Boston, USA, Sept. 2015.

One of my good friends, Alfred Cheng, in Hong Kong, suggested that I speak to the sales team at Global Sources regarding building up the industry knowledge. It was an idea, floated later, by Claudius Chan too. I actually ended up taking a session with the team! 😉

May I also add that two of my very, very dear friends, Kevin Ho Lau Fai and Jo Yashan Kuo, felt great pride after reading my presentation, and later, thanked me profusely.

Well, it’s all worth it when you have friends such as these four, who take the time to appreciate all of your efforts. They make my day, my life! Thanks guys!!

Here’s what I have to say regarding building up industry knowledge.

‘Speaking’ the supplier’s language

  • From experience, suppliers open up to those who can share information on their (suppliers’) industries/verticals.
  • From experience, suppliers open up to those who can share information on other markets.
  • From experience, suppliers open up to those who provide good, useful, suggestions.
  • Suppliers DO NOT open up easily, if you cannot provide insights about specific verticals of their interest.
  • Suppliers would respect you even more, if you could provide meaningful suggestions.
  • Suppliers always look for leads; so, develop a habit of providing those.
  • Suppliers also look for trust; provide that by exchanging relevant information, when possible.
  • If covering electronic components, try and get an insight on the industry. It certainly helps!
  • If covering electronics, computer or telecom, know the industry background a bit; see how a supplier laps up all the information that you may have.
  • Convert the supplier into a trusted ally. How do you do that?
  • Suppliers always look for leads; develop a habit of providing those.
  • Suppliers look for trust; provide that by exchanging relevant information, when possible.
  • Try and understand what the buyers need; it helps in understanding the suppliers needs.
  • From experience, suppliers are keen to know what’s happening in India and China? Can you provide that information?
  • Do not provide statistical information, as suppliers already have access to that! Instead, give them information you may have gathered from other sources.
  • China and India markets are very strong cases. Develop yourself and know more about those.

Most of all, keep learning all the time.

Educate yourself!
No one can say, “I’ve a good knowledge about an industry.” Absolutely, no one! There is no end to learning!

Everyday, new things are happening. How do we keep pace with so many new things happening in the verticals? The best way is to educate yourself at work!

Education-at-work programs
Follow these religiously:

  • Bookmark relevant Web sites – visit them often.
  • Sign up for newsletters and news – at least, read the headlines.
  • Flip through technology and other magazines you may get your hands on – there’s always something to learn.
  • Build your network and get help.
  • Visit Web sites of leading research firms, there’s lot of information out there.
  • Visit Web sites of industry associations.
  • Visit Web sites of industry events, look at the live coverage.
  • Browse presentations of industry events; that’s an excellent way to track trends.
  • Read relevant articles in print/on Web.
  • If you don’t understand a technical term, do search on the Web/dictionary. You don’t need an expert, nor do you need one to tell you what the term means.
  • Listen to what the suppliers say – you can pick up lots of points.
  • Make it a point to talk to your seniors, especially in sales and editorial – they actually have a lot of information to share.
  • Look up competitor Web sites and magazines – they always have relevant stuff.
  • Read flyers and memos from sales and marketing departments.

DEVELOPING SOUND INDUSTRY KNOWLEDGE TAKES LOT OF EFFORT AND TIME! BUT, IT’S ALL WORTH IT!

PS: A leading manufacturer has launched next-gen ultracapacitors. While I won’t divulge who, it certainly took me back to my earlier posts. Again, if interested, read here.

This entry was posted in Automotive, EDA, Electronic Components, Electronics, Industry 4.0, IoT, renewable energy, Semiconductors, Solar PV, Telecom and tagged , , , , , .