Blog

Stay updated with our latest news and announcements.

Insights

[Invited Insights] A Journey to New Computers

Cheol Seong HwangCheol Seong Hwang (SNU Distinguished Professor)

When Allan M. Turing wrote his memorable paper entitled “On computable numbers, with an application to the Entscheidungsproblem” in 1936, he assumed the presence of an unlimited amount of memory to complete his conceptual “universal Turing machine”. Because the universal Turing machine is a conceptual prototype of the modern von Neumann computer, an ideal computer requires infinite memory and a high-performance processing unit. As we know, however, an actual computer has limited memory capacity, which severely degrades the performance of the Turing machine, i.e., computer. 

When I entered graduate school in 1987, the first personal computer was built (IBM XT or AT) using Intel’s 8088 CPU, DRAM, and a hard disk. At that time, the CPU clock speed was ~0.2 MHz, and hard disk capacity was 20 MB, which is unimaginable from today’s perspective. Nowadays, the CPU clock is a few GHz, and the hard disk capacity is ~TB, indicating that the CPU performance and memory capacity have increased by ~104 and >105. Therefore, the resulting computing performance improvement is astonishing. The modern cell phone’s overall performance is ~1000 times higher than the mainframe computer, which sent the Apolo11 to the moon in 1969. All these developments could be an outcome of the concerted effort of the semiconductor chip and computer program engineers, who attempted to make the modern computer as close as possible to the original Turing machine. 

Memory used to be DRAM and hard disk, but now it means several other products. Even the term memory indicates DRAM, but hard disk and flash memory are termed as storage. There have been long standing discussions on the future of memory/storage and even new memory/storage class memory. It is time to seriously consider merging memory and processor because of the unacceptably high energy consumption due to the data transfer between them. We already see the emergence of process-in-memory or process-near-memory chips, meaning that the 80 years of Turing machine time is ending. 

RTC should be at the forefront of this new wave, which will lead the technical revolution of not only the SK Hynix company but also the entire computing ecosystem. All these pioneering and revolutionary visions are on the shoulder of the team members. They must be active, brave, and risk-taking while still retaining the promising culture of the company, harmony and collaboration. This is a challenging task but undoubtedly valuable enough to pursue. Personally, I congratulate and encourage the team members to have such a great chance to contribute to the company, society, and human beings.



Popular Insights

Previous Next List