Sundar Pichai |

Tensor processing unit (TPU) Google I/O conference presentation

At the latest Google I/O developer conference in Mountain View, the tech giant announced a new tensor processing unit (TPU), a processing accelerator unit used in machine learning that moves Moore’s Law forward with almost seven years or about three chip generations.

Google says the accelerator can deliver the highest performance-per-watt order of magnitude, greater than all the commercially available ones.

The new tensor processing unit is designed specifically for machine learning tasks. In fact, as Norm Jouppi, Google engineer confirms, the accelerator has been running in their data centers for more than a year. They have been using it for machine learning actions that required reduced computational precision like object recognition or deep learning.

The name originated in the accelerator’s main application, the Tensor Flow. Tensor Flow is an open-source library software for computing large math datasets and interpretation with visual graphs. This cool software was first developed during a twelve-month academic program specializing in machine learning, neural networks, linguistics and data visualization called Google’s Brain Team.

According to Sundar Pichai, the company’s CEO, CPUs, and GPUs will never be replaced by tensor processing accelerators, but their capabilities can accelerate machine learning with a fragment of the power required by other ASICs.

The drawback of this development comes, however, from the fact that Google’s TPU is designed for highly-specific workloads; the applications are Cloud Machine Learning Alpha and Tensor Flow, which are used to process numerical datasets and allow tolerance for reduced precision.

Only a handful of GPUs and CPUs (see Haswell and later) support this kind of calculations in Tensor Flow.

Such processing mode is useful because it delivers two times the performance of FP32, having reduced memory usage on a neural network that allows, in time, the development of larger networks.

Since the recent slowing of Moore’s Law, the engineers have attempted to defy atomic-level physics and create nanometer level designs. While many of them will have to wait a few good years, Google’s tensor processing unit has brought 2023 performance to present machine learning.

Again, the drawback is, these processors are application-specific chips for deep learning data or little more than that.

The tensor processing unit won’t be available for corporate or enterprise purchase.

Image source: Venture Beat

Artificial Intelligence Is A Lot Stronger

Artificial Intelligence is a lot stronger and we can see it with each passing day. Your mails are put in a precise and accurate order, you can have your voice searches translated for you on the spot and there are many more features which are showing us that artificial intelligence has come a long way. Google intents to make it even smarter, as they have already mentioned that machine learning will enhance its services in the years to come.

Alphabet proceeded to report its own quarterly financial results on Thursday this week and they showed that profit and revenue had gone up compared to last year. Sundar Pichai, Google’s new CEO, participated in the calla bout his first earnings. He did not hesitate to point out how much of an impact machine learning is going to make in the years to follow and he kept talking about it whenever the opportunity arose.

Pichai declared that “machine learning is a core, transformative way by which we’re rethinking everything we’re doing.” This might just be enough evidence that Google is planning on using machine learning in every service that they are already providing, or whichever one they will come up with in the future. We are going to be assisted by artificial intelligence in every moment of our lives.

But we cannot help but wonder if this is the right way or not. As far as we are aware of, human beings’ purpose in the universe is to accomplish as much knowledge as possible about anything. If machine learning is the future, what will become of human learning? Machines will ultimately learn enough to be able to assist us, or even grow their own consciousness. Yet if we are assisted at every moment, why would the majority of us struggle with learning anything at all?

If machines can do it, let them do anything we need. We will just sit around in our comfortable lives and do nothing. Is that “the way?”

We can already see that Artificial Intelligence is a lot stronger. It is only a matter of time until machines will assist each and every one of us in cleaning the house, performing any type of work and even learning. Two pivotal questions arise if we think about this: how far can we go with machine learning and is this the right way that will lead towards human evolution? It’s just food for thought.

Photo Credits staticflickr.com