On Fri, May 27, 2016 at 04:10:57PM -0400, D. Hugh Redelmeier wrote:
I'm going to do the unthinkable: throw out a working notebook tomorrow.
An NEC Versa SX from 1999. Running Fedora Core 1. Pentium 2 @ 233MHz 256M of RAM 10G HDD [an upgrade from something considerably smaller] Video: Trident Cyber 9388 Screen: 1024x768. No built-in networking (that's what PCcards are for). Swappable bay for CD writer and floppy drive (I have both). Came with a Windows NT 3.5 license (if I remember correctly). but with Win98 installed.
The video is a problem: it is only "TrueColor": 16 bits/pixel. The X driver has a long-standing bug where at each power-on the video might be in a state where the colours are screwed up. The only cure is to power cycle. The fundamental problem is that X doesn't know how to fully initialize the device. Kind of 50/50, if I remember correctly.
I'm playing with it now. The screen is bright and clear. I quite like the feel of the keyboard. There is a constant fan noise.
The machine was quite nice for its day but things are mostly better now. I bought this 17 years ago, lets say 8 doublings by Moore's law. Current nice notebooks vs this:
RAM: 16G (2^6 better)
Well that's hardly high end by todays standards, although probably still higher than typical.
SSD/HDD: 256G vs 10G (2^5 more capacity but a lot faster) HDD/HDD: 2T vs 3.2G (2^10 more capacity and a bit faster)
Might be more fair to compare the 10G against the 2T and the 3.2G against the 256G.
screen: 3200x1800 vs 1024x768 (2^3 better) CPU: 2.4GHz vs 233MHz (2^3 faster clock but more instructions/second too)
But how many cores and how much cache? That's probably where the transistors have gone. After all 4 cores are common, so that alone is 2^2 extra, and a modern 2.4GHz is probably more than 2^4 times the performance of the 233 Pentium, so I would say a mid range quad core these days is easily more than 2^6 times the top of the line Pentium ever made. Not 2^8, but still.
So: Growth is impressive but Moore's law doesn't seem to have applied.
Well, it only applies to transistor density, not performance. -- Len Sorensen