Gee, 1987, I was at IBM's Watson lab as a researcher in their effort at the time in artificial intelligence.
So, at the lab we had an ambitious computer services group, and they had for our general use six mainframes, I believe all single processor machines, running the operating system software Virtual Machine/Conversational Monitor System (VM/CMS).
For access from home, I had a PC/XT with some special software to use a standard dial-up land line telephone line to access the mainframes. As I recall, the data rate was about 30 characters per second.
So, the mainframes were likely water cooled. As I recall, the last of the water cooled mainframes had a clock speed of 153 MHz.
In contrast, the last desktop I plugged together has an AMD (American Micro Devices) processor, FX-8350. It has 64 bit addressing and 8 cores. The standard clock speed is 4.0 GHz. To buy the processor, at Amazon I paid their usual price, quantity 1, ~$100. Tiny little thing; came in a tiny little box. The processor has been fine: It is the basis of my startup.
So, let's compare, counting clock ticks:
The 6 mainframes had best I can guess at most in total
6 * 153 = 918
million ticks per second. The AMD has
8 * 4.0 * 1,000 = 32,000
million ticks per second.
Then we can take the ratio
(8 * 4.0 * 1,000) / (6 * 153) = 34.858,387,799,6
Call it 35. Soooo, counting clock ticks my desktop with the $100 processor from the tiny box is 35 times faster than all six of those mainframes combined.
With speculative execution, out of order execution, pipelining, etc., there is on average some number of instructions per clock tick. So, first cut here, and maybe a gift to the IBM mainframes, for the two processors count the average number of instructions per clock tick as the same.
So, this fantastic, we're talking not just world class but unique in civilization class, improvement in price/performance, is much of what enabled Google, Facebook, Amazon, Apple, Microsoft, AMD, TSMC (Taiwan Semi-Conductor Manufacturing Corporation), smart phones, solid state disks, and essentially all of the Internet.
E.g., for communications data rates for connect from home, now with 1 Gbps Internet versus then, 1987, 300 bps, ..., to the 56 Kbps of V.90 of 1995.
So, how the heck in 1987 was a computer science student or early career person to anticipate and actually believe in this fantastic, all-time unique step up in price/performance of computing and communications along with associated infrastructure?
This progress blows away wood, wheels, stone, copper, iron, steel, steam, electricity, and broadcast radio and TV.
Now, what are we going to do with this suddenly cheap, powerful computing, communications, and infrastructure?