In this blog post, we’ll focus on “The Emergence of Multi-Core Processors and Their Challenges: Why Doesn’t 1+1=2 Hold True in Computers?” to naturally summarize the background, main content, and key insights and impressions revealed throughout the text in a single sentence.

The Emergence of Multi-Core

Terms like “dual-core” and “quad-core” frequently appear in recent computer advertisements, and these are collectively referred to as “multi-core.” Here, “core” refers to the processor that performs the computer’s core functions (computation and control). In the past, since there was only one processor in the CPU, “CPU” and “processor” were used interchangeably; however, with the advent of multi-core technology, the term “CPU” has come to mean a collection of processors. But if multi-core technology is so good, why didn’t it appear in the 1990s? To understand the reason, we must examine the background around 2005, when multi-core technology truly took off.
The factors determining a computer’s performance are integration density and clock speed. Integration density affects the ability to execute complex instructions, while clock speed determines the computer’s calculation speed. The clock acts as a start signal for computer calculations; when the signal is generated, a single calculation is executed and completed. For example, a 3.0GHz CPU can be described as performing 3 billion calculations per second.
CPU clock speeds increased rapidly from 8MHz in the early 1990s, and by the end of 2004, the arrival of 4.0GHz CPUs was anticipated. However, in 2005, Intel announced that it was abandoning the development of 4.0GHz CPUs. The reason for halting development focused on increasing clock speeds was the heat generated at high frequencies. As the clock speed increases, electrons move very rapidly, generating a significant amount of heat; at speeds above 4.0GHz, adequate cooling becomes difficult with air-cooling systems alone. Specialized cooling equipment is required for normal operation, but such equipment is not cost-effective.
To use an analogy, the approach of continuously improving a single carpenter’s skills had reached its limits, and the return on investment was diminishing. Therefore, the strategy shifted toward hiring multiple carpenters to divide the work—and this is the concept behind multicore processing.

The Challenges of Multi-Core

So why wasn’t multi-core technology adopted sooner? If multi-core had been used before heat issues arose, we could have gained performance benefits immediately. The reason is that adding more than one core leads to the law of diminishing marginal returns. Diminishing marginal returns refers to the principle that when an additional unit of a production factor is added, the marginal product decreases.
To illustrate with a carpenter analogy, hiring another carpenter of equal skill might seem, in theory, to cut the work time in half, but in reality, this is not the case. For example, a task that took one person two days might take not one day but about one and a half days when done by two people. In other words, even when using two processors with a dual-core design, performance often does not exactly double. This is an example in the computer field where 1+1 does not equal 2.
Why does this happen? When two carpenters are building a house, if there is a shortage of tools or if certain stages of the work must be performed sequentially, one carpenter must wait until the next stage is completed, even if they have finished their own task first. In other words, while some tasks can be processed in parallel, others must proceed sequentially, which limits the overall speed improvement. In computer science, this principle is known as “Amdahl’s Law.”
For example, when calculating “(3+5)*(9+1)+3” on a dual-core computer, the first processor (A) can compute 3+5 while the second processor (B) computes 9+1. However, when combining the results of the additions to calculate 8*10 and 80+3, one processor must wait for the other’s result; therefore, the total execution time does not exactly halve compared to a single-core system. In this example, while a single-core system took 4 units of time to compute, a dual-core system took 3 units of time.
The table below shows the speedup achieved based on parallelism and the number of cores. As shown in the table, even with 16 cores, a parallelism of 50% yields only about a 2x speedup. Therefore, increasing the program’s parallelism contributes more to speedup than simply adding more cores.
However, it is very difficult for someone accustomed to sequential thinking to design parallel programs. Because of the significant industrial demand in this field, it is currently an area of active research in computer engineering. However, as seen in the example above, there are ultimate limits to parallelization. No matter how much you program in parallel, you cannot completely eliminate sequential parts.
Consequently, multicore technology may also eventually reach its limits, and research aimed at a “device revolution” is gaining attention as a way to overcome this. Just as the transition from vacuum tubes to transistors enabled further development, there is a view that introducing new devices—such as bio-devices—will be necessary to surpass current limitations.

Summary

In summary, as speed improvements in single-core processors hit a wall due to heat issues, multi-core processors emerged as an alternative. However, the actual performance of multi-core processors depends on the degree of parallelism in the program, and improving this parallelism presents significant challenges. In fact, with the shift from single-core to multi-core, the pace of CPU speed improvements—which had been sustained until 2005—began to slow down.
From the user’s perspective, there may be a sense that upgrades have become less frequent in recent years, as it is difficult to perceive that computers are slower than before. Consequently, Intel faced a crisis of declining profits and sought countermeasures such as entering the smartphone market. As competition for market share with ARM—which originally dominated the smartphone market—has intensified, attention is now focusing on the computer sector.