Hello everyone and welcome to ZD Tech, ZDNet’s daily editorial podcast. My name is Guillaume Serries and today I will explain to you by what miracle artificial intelligence enters mainframe computers.
One could think of the alliance of carp and rabbit. Or the Ford T and the Tesla. But it is nevertheless true. The new IBM z16 mainframe is equipped with an in-house Telum processor that enables artificial intelligence to analyze transactions entering the computer in real time.
Certainly proof that mainframe computers remain as relevant in 2022 as they were in the 1960s.
The Telum processor
The Telum processor is a new 5.2 GHz dual-processor chip consisting of 16 cores. It can perform 300 billion deep learning inferences per day, with one millisecond latency.
In short, it’s fast. Much faster than what the cloud computing technology has been offering us for a few years.
“The z16 has 20 times the response time, with 19 times the throughput, compared to an x86 cloud server,” says analyst Patrick Moorhead. Above all, a cloud server offers an average latency of 60 milliseconds. We are therefore a long way from the millisecond of the mainframe.
What are the differences between mainframe and cloud computing?
So now, I will explain to you why we arrive at such differences in performance. The mainframe, known as a central computer in good French, or macro-computer, is a computer with high processing power, which serves as a central unit for a network of terminals.
In short, it is the exact opposite of cloud computing. Because the cloud is an architecture in which distributed servers are placed in one or more datacenters. So yes, the latency is more important than for a mainframe.
Because the mainframe is an architecture with a single, very powerful computer, located in one and the same place. Its advantage in addition to latency? It is very reliable and secure. Its downside? It is not very scalable, and its operating and maintenance cost is high.
Consequence: mainframes are used only in very large companies such as banks, insurance companies, or even airlines.
What are the differences between a mainframe and a supercomputer?
Come on, we have a few minutes left so I’m going to dig a little deeper. Because you have certainly heard of supercomputers, like the Japanese Fugaku.
Well know that supercomputers and mainframes are different. Yes, they are both very large computers. But their architecture is very different.
Simply put, a supercomputer has tremendous computing power. And so, supercomputers are used to solve scientific and engineering problems. Specialists then speak of high-performance computing.
Mainframe computers are very, very good, not in computing power, but in transaction processing. They are therefore capable, for example, of quickly modifying huge databases.