For some Sunday nerd fun, I asked Grok to take the performance of the fastest super computers from 1955 onward, then graph how long it took PCs and mobile phones to catch up. It came up with this chart.

For example, it took about 28 years for PCs and 54 years for mobile phones to catch up to the most powerful computer in 1955, and it took about 10 years for PCs and 20 years for mobile phones to catch up to the most powerful computer in 2000.

Remember the Cray A&M bought in 1989? They replaced it with a PC 12 years later.

I admit that I don't know much of anything about the current state of supercomputers. But I have a hard time believing consumer PCs are as powerful as supercomputers from <10 years ago? I mean just staying within the desktop world, I am sure a 2025 Ryzen crushes a 2015 Skylake (a recent PC swap I've done for myself). But the difference isn't exponential or probably even more than 2-3x on a CPU benchmark.

SJEAg said:

---

I admit that I don't know much of anything about the current state of supercomputers. But I have a hard time believing consumer PCs are as powerful as supercomputers from <10 years ago? I mean just staying within the desktop world, I am sure a 2025 Ryzen crushes a 2015 Skylake (a recent PC swap I've done for myself). But the difference isn't exponential or probably even more than 2-3x on a CPU benchmark.

---

I really don't know enough about any of the newer hardware specs to comment but I did laugh pretty good because when I typed in "2025 Ryzen vs 2015 Skylake" into google and the AI response was literally

Quote:

---

A 2025 Ryzen CPU is exponentially more powerful than a 2015 Skylake processor due to a decade of advancements in architecture, core counts, and manufacturing processes.

---

Haha, ok fair enough.

I'm just saying I can be using the Skylake today as a very functional Windows machine. Looking up benchmark comparisons, its maybe +50-150% in real world performance. That's significant sure, but not some revolutionary leap in technology like I would mentally picture overtaking a "supercomputer" would be.

Anyway, maybe supercomputers aren't advancing due to the lack of need. That could explain it. Until the quantum computer thing is fully figured out anyway.

Very insightful indeed!

SJEAg said:

---

Haha, ok fair enough.

I'm just saying I can be using the Skylake today as a very functional Windows machine. Looking up benchmark comparisons, its maybe +50-150% in real world performance. That's significant sure, but not some revolutionary leap in technology like I would mentally picture overtaking a "supercomputer" would be.

Anyway, maybe supercomputers aren't advancing due to the lack of need. That could explain it. Until the quantum computer thing is fully figured out anyway.

---

The "supercomputer" really has just changed how it's implemented. it used to be a single computer system, albiet a massive one, nowadays high performance computing is clustering thousands of servers. Supercomputers are definitely advancing, they've just changed form from an entire room to an entire building.

The need is largely for laboratories for crunching data, and they are years long custom projects to design and build. You'll likely start seeing something similar with AI processing, though it will take a different form as hundreds of thousands of GPUs being the primary driver of compute power.

Quantum computing is a totally different animal, with zero similarities to silicon computing. There is no comparison of compute power, only of time to compute specific use cases.