In this post I explain why I just spent almost $8K on a new PC (the “MegaBox”), and also built two more – Partly(?) because it turned out to be possible to build one a million times more powerful than my first IBM PC clone for the “same” cost, 35 years later; and mostly, to experiment with what an extremely powerful PC can do that a normal one can’t. I think the computer industry has stagnated and I’m wondering what if anything we’ll do with more computer power. For my experiments, I needed a “regular” PC and a “baseline” PC to do comparisons.

Details:

I’ve been planning to write a first “real” blog entry for over a week now, and haven’t gotten around to it mostly because I wanted it to be a “good one”, which requires more time and effort than I’ve had available.  So I’m going to settle for “anything being better than nothing”, and just write whatever comes to me this morning.

As you can probably tell from the first few photos in my projects section, my primary project lately has been building computers. The last PC I built was in the winter of the first year I was in Colorado, early 2012 – about 8 years ago.  I used to build a new computer about every 3 years. DRB Systems paid for them and Dale always encouraged me to build “hot-rods”, so that PC had the fastest available (at that time) Core-i7 chip, 32GB of RAM (all it could hold, and WAY more than the 1-2GB typical at the time), and a pair of 256GB SSDs, which cost close to $4000 at the time. Even today, it benchmarks almost three times faster than my laptop. Since I’ve retired, I haven’t had a need to replace it. Even a $500 laptop is fast enough to run Word and Chrome. But it was getting old and a little flaky, and I’ve been getting back into programming and using it more, so I decided it was time to replace it.

I ramped up my internet research, found pcpartpicker.com, and started designing a new desktop PC.  What I found was interesting. The highest spec (consumer) CPU you can get today is an AMD Threadripper TR3990X. It costs almost $4000 (half my budget!) but has 64 cores and a huge amount of I/O bandwidth.  Motherboards for this chip can hold up to 256 Gigabytes of RAM – exactly a million times more than that first PC 35 years ago. I was also planning to buy 2TB of SSD, and keep the pair of 4TB HDDs I had in my old computer, so this machine was going to have 10TB of storage space – again, exactly a million times more than the 10MB in than that first PC. Then I looked up transistor counts and found that with a high end video card, this PC would have at least a million times more transistors than the first one in 1985. The price? About $7800, fully loaded. I punched that into an inflation calculator, and got… about $3200 – the same price as that first PC clone 35 years ago.

So, I decided I had to build it. Now I have a “high end desktop” PC at least a million times more powerful than my first one. I named it the “MegaBox”*.  Here is its page on PCPartPicker. (Note: The prices are current, not what I paid at the time.)

The big question is, what am I going to do with it?

In my opinion, the PC industry has stagnated. Even high-end games can’t use more than 32GB – 1/8th the RAM in the Megabox, or more than 8 cores – 1/8ththe number in the Threadripper. In fact, I actually built a machine like that as part of this project – an 8 core (Ryzen 3800X) with 32GB of RAM – and also third one – a 4 core (Ryzen 3200G) with 16GB. With most software, all three machines “feel” about the same speed. In fact, for most purposes they’re not significantly faster than my laptop- a Core-i5 with 4GB.

In the early days, I used to get a new computer every year, because they improved so much faster than they do now.  In 1987, the jump from my 5 MHz 8088 from two years before to an 8MHz 80206 at least quadrupled my computer power. Meanwhile, the software industry took advantage of every speed increase to add functionality and capability. But in the last decade or two, we’ve reached “good enough”. Going from 10 seconds to jump to the end of a document or recalculate a spreadsheet down to 1 second matters; taking the 1 second down to 1/10th is barely noticeable. Now we hardly care about “faster” – thinner and lighter and better battery life matter more.

So now I have a machine with at least an order of magnitude of untapped potential. If I come up with something useful to do with it, maybe it will be something new too. And meanwhile, the MegaBox loads PyCharm almost 8 times faster than my laptop, and if I live long enough, those seconds I’m saving over and over will add up!

* At first I was going to name it the “Box-1M” (Box one million), after Dale Brott’s nickname for my first home-built computer – the “Box-100”. It was a big, heavy box made out of plywood and spray-painted black, containing a Z-80 single-board computer with 64KB of RAM, a pair of 8″ floppy disk drives, and a homemade power supply. But that wasn’t an “IBM-PC clone”, and technically, “box 1 million” would only be 10,000 times more than a “box 1 hundred”, so that didn’t stick, and it ended up “MegaBox” instead.