A decent amount of interviews with Jim Keller (Jim Keller) and his associates on the pages of Fortune was related mainly to the Affairs of bygone days, but some things were touched on the current activities of this eminent developer of processors from Intel. The fruits of his labors can be assessed in a couple of years.
In the interview section on the activities of Jim in the state of AMD, is mentioned about his desire to make architecture processors Zen modular design that would allow greater flexibility and reduce costs. How good are the processors generation AMD Zen 2, Jim Keller learned for a long time as an employee of Intel. According to associates of Keller, it should not be assumed that he did not bring it started to end. Jim just interested in basic problems, and the routine depresses him, so he usually instructs subordinate technicians.
Even Intel, as noted in an interview, Keller intends to apply modular design in the architecture of future processors of this brand, which we’ll see in a year or two. Perhaps the rumors about the intentions to use processors Alder Lake-S analog layout big.LITTLE have some basis since they won’t appear until next year. The main advantage of the modular approach – ability to modify individual parts of the processor at the development stage of the successor without any interference to other modules. Does this mean the transition to separate crystals or chipley, to understand difficult, but in the graphics segment of the Intel chip definitely implements the layout at the end of 2021, presenting boosters Ponte Vecchio.
Jim Keller also promises to find cost-architectures that have traditionally been at the heart of the Atom family of processors in desktop and server segments. This again makes us think about the layout big.LITTLE combining productive and cost-effective kernels on the same substrate or chip.
Special attention Keller pays and to improve the functioning of processors, relevant to the operation of systems of artificial intelligence. Finally, it is not a categorical opponent of integration in the Intel logical block third-party development. The principle of “natural economy” for it is not fundamental. By the way, after joining Intel, Jim Keller began to deal with excessive bureaucratic functions that affect operation. In the first days at a new workplace, he was shocked that the meeting to resolve minor problem was attended by at least fifty members, and technical specialists redundantly controlled by management staff having no knowledge in the subject area.