Slashdot Mirror

Back in the 1970s, 80x25 characters was the common resolution for a terminal display. Each character was 9x14 pixels (or less commonly, 8x8 pixels). That meant the terminal displayed 720x350 pixels = 252,000 bits of data, or 31.5 kB (or 16 kB for the 8x8 fonts). How then were cheap terminals able to support these resolutions when RAM cost $50 per kB?

Clever engineers came up with text mode. The bitmapped fonts were hard-coded into the display in ROM (which was much cheaper than RAM). So if the terminal supported ASCII, it held 127 9x14 characters in ROM (2000 bytes worth). The computer would then send the display a list of characters to display, rather than a list of pixels. The display would look up each character in ROM, and use that bitmap to display that character (or rather, a row of 80 characters). Then move on to the next row, and so on, then repeat again from the top. This allowed the terminal to display 16 or 32 kB worth of graphics using only 2 kB of RAM (either built into the screen or streamed in real-time from the computer). The caveat being that only text could be displayed.

Came here to same just about the same thing.
Even brought along some facts: http://www.jcmit.com/memorypri...

Price per mb at the end of 2012: $0.0037/mb
Price per mb Sep 13, 2014: $0.0085/mb
Price per mb May 15, 2015: $0.0056/mb

Sure, it fluctuated, but it wasn't a big drop, and definitely not a historical low.
The better question, is why isn't it going down further (especially on larger modules)?

Last time it was above $1/mb was in 2000.
In 2002, it hit a low of $0.19/mb - THAT was a drop.
First time it dipped below $0.05/mb was 2007 (got as low as $0.024/mb that year).
It still hasn't hit another 1/10th the price drop ($0.0025 has never hit).

I'd like to get some more memory, but the last time I got 2x8gb, it was cheaper than it is now. Makes it hard to justify... I've expect it to eventually go down in price, and if I wait long enough, I'll have to get a different format - probably worth waiting at this point anyway (ddr4 instead of ddr3).

NT required a much more powerful machine, not like today where you can get a usable computer for $250 new.

RAM pricing history:
http://www.jcmit.com/memorypri...

Early in 1995 it would have been several hundred dollars cheaper to run Win95 this all but evaporates circa 96 and beyond.

Requirements for NT 3.5.1 workstation:

12MB of RAM
90 MB free drive space
VGA level video support
Keyboard
IDE, EIDE, SCSI or ESDI hard drive
386 or 486/25 processor or better
CD-ROM, floppy or active network connection

Requirements for Windows 95:

Personal computer with a 386DX or higher processor (486 recommended)
4 megabytes (MB) of memory (8 MB recommended)
50-55 MB

He was saying that the SS10 could handle 512 MB in 1992, at time when the best PCs were maxed out at 32 MB or so.

The SS10 takes proprietary memory, and I know there was a firmware update that allowed it to use larger (32 MB, I think) sticks at some point. Ultimately, I don't think there was any way to put 512MB into a SS10 in 1992, even if the machine did eventually support it. I think 128 MB was more likely, though even that's very good for a desktop box back then.

As for 128MB simms in 1992, I have my doubts. This chart doesn't really try to list *everything* that was available, but even so -- it doesn't list 128 MB sticks until 1999. (It doesn't mention 64 MB sticks until 1999 as well, so clearly, it's missing some stuff.)

According to this, there were 64 MB SIMMs available in 1995 for a massive price -- $2600 each. (I didn't try to find the ad itself, however.)

Prices fell more like 40%: http://www.jcmit.com/flashpric...

It all depends on the numbers which we don't have. The problem is the use of relative terms like "wide range" and "higher humidity". For example, High humidity in the tropic is much different than high humidity in the desert. If BR media can handle a wider range of temperature and a higher humidity level there will be savings in HVAC.

So in five years, you may be able to get 20GB for what a 4GB HDD costs today.

That is an assumption and I bet that Facebook has looked at what is coming down the pipe. It is quite possible that these price decreases will slow. By the way HD prices have not come down as fast as you seem to think. In 2010 a 2TB Seagate drive sold for 0.0000550 $/MB. In 2014 a 3TB Seagate sold for 0.0000367 $/MB That is a 38% drop in four years. If it followed Moore's law (cut in half every 2 years) it should be 0.00001375 $/MB or a 75% drop.

PS. You probably meant TB not GB.

Around 1994, you'd spend close to $300 on 8MB of memory, a 486 CPU could be around $250 (being generous). I've blown your $500 budget, and you've still got the motherboard, sound and video hardware, floppy drive, hard disk, power supply, and IO peripherals before you have a functional computer. Sound and CD-ROM alone would cost around $500. If you're buying enough to build a whole computer from scratch with retail-priced parts, you're looking at an easy $1500+ for a machine that isn't particular top of the line. Swap meet prices would lower that somewhat, granted, but it's hard to imagine it would've been an over 70% discount.

And anyhow, the guy you replied to didn't say anything about computer prices. His parent post did, but you replied to the wrong post.

8 meg of RAM was the minimum requirement to run Win95, not typical, which I would say was closer to 32 meg.

The price of 32 MB of RAM in 1996 --- as advertised in BYTE was --- $550. Memory Prices (1957-2012)

$756, adjusted for inflation (2010) The Inflation Calculator

I would be enormously surprised if the mass market Win 95 PC averaged more that 8 MB of RAM.

1993 price bump was barely 20%, 1999 price bump was >300%
http://www.jcmit.com/memoryprice.htm

He is also doomed to look like a fool, as it is supposedly his technical savvy that lead to market dominance by Microsoft.

Bill did study history. He studied Xerox and their inability to extract profit. And he probably cocked a snook at Adam Osborne. I expect better from a primordial UID.

This is classic motive substitution. Bill has never been about beauty or progress. He's all about the tsunami of monetization. Innovation never had much to do with their business model. That's why you can hardly find three paragraphs from the mouth of any Microsoft executive in which the word "innovation" doesn't appear. Innovation is Microsoft's FTC shibboleth.

In information technology, innovation is more generally a code word for "those who won deserved to win". Also known as "history is written by the victor" or its corollary "those who study history are doomed to rewrite it".

The ugly truth is that when you an exponential technology curve so uniform over thirty doubling cycles that a Tour rider would select a fixed gear bicycle for weight reduction, technological innovation is not what separates the fit from the fallout. What separates the fit from the fallout is expertise in non-linear business methods and continuous disruption. Anyone who studies the naked Napalm children of the PC revolution with a gun-steel gaze will recognize Microsoft for expertise and innovation in both of these areas.

Telling me that PLATO is a viable present day reference is like discovering that some ancient civilization had a system for loaning out graven stone tablets exactly like a modern day library right down to the papyrus library card.

Here's where innovation enters the ancient picture. Some prescient dude observes "if only we carved the same symbols onto gold leaf, the whole concept would take the demographic leap".

Memory Prices (1957-2010)

The oldest memory chip I recall is the 2102, which cost $50/kB back in 1975. I learned how to fat finger on a board with 8 of these chips. We're presently in the vicinity of $10/GB. Mesopotamian libraries with books hammered out of gold leaf is not much of a mental stretch for me.

But somehow if I skipped the introductory chapter on Mesopotamia 101, I'd be doomed to repeat some classic mistake != coming to market with a technology pounded out of gold leaf.

Little known fact: Jacques Cousteau spent the majority of his career in secret pursuit of an overdue gold-leaf copy of a Mesopotamian page turner entitled "Roots". According to legend, the author complained bitterly over it being "passed around".

The short history of windowed operating systems:

Memory Prices (1957-2010)
1985 $300/MB
1990 $100/MB
1995 $32/MB -- cartel warning !!!
1996 $5/MB -- and the wall came crumbling down
1998 $1/MB

Even uglier if you correct for inflation.

I couldn't stand *any* of the early Windowing products. Too cramped. My "fat" Mac was no better than anything else. Spent more time dragging Windows around on that small screen to see what I was working on than getting anything useful accomplished.

Had two floppy drives, but Jobs had instilled a miraculous ability into the OS to pop out the wrong floppy when it needed a file on a different diskette, the one you knew you'd have to put back in two seconds later. There was no way to override or preempt this. Jobs knows best. Burned into my amygdala so deeply it will twitch on the autopsy table at the sound of automatic floppy disk eject.

If I had been willing to upgrade with a hard disk, that system might have become borderline usable. I priced the drive upgrade at roughly on par with buying a turbo XT with a hard drive and monitor from scratch.

From that day forward, I learned to tolerate MSDOS, and had a work flow that got enough done. What I realized in retrospect is that my work flow discouraged experimentation. No Carmack for me. New ideas were added straight into the production code base. Just to keep the number of contexts to a minimum, with no multitasking to help out. I was shaped by my tool, and not in a good way.

At one point we had a spare machine and I tried out Coherent. That experiment ended immediately when I discovered that the bundled C compiler supported K&R, but not ANSI. All of my own code was portable ANSI. Game over. I didn't have access to the internet yet, so it wasn't easy to download Linux, which was still pretty green.

My long sentence under DOS ended when I jumped to Windows NT 4 circa 1996 on a brand new Pentium Pro. Had my first cable modem within the year. Good times.

By 1999 I had several large monitors, a KVM, OpenSSH, and *finally* enough system memory. I was no longer shackled to picking one primary work environment. I could use whichever system best matched the requirements. Divorce rocks!

For me 1999 was an inflection point in the merit debate. It was like the sexual liberation of the late 1960s. Choosing your system environment in 1985 was like watching a movie made in the 1950s about communicating emotion. You weren't operating in a regime of easy choices. Much consumer loyalty in the PC space 1985-1999 was born of cognitive dissonance. After you married the damn thing, you couldn't bear not to defend it.

Most people think of getting a driver's licence as a milestone of independence. For me, a decade of suck ended the day in December 1995 when I discovered AltaVista through my crappy dial-up service. "Damn, this rocks!" I thought to myself. I've never give ten brain cells to a press release for the rest of my natural days! Finally a liberation worth having.

For my desktop, I wanted a window that opened outward, not a set of prison bars to tile a display that was too small to begin with. This was largely driven by the limitations of the CRT and the cost of memory. Pricey to have a desk full of bit-mapped workstations prior to 1996.

After my sour Mac experience, had it been available in 1986 for a competitive price, I sure would have enjoyed a 24" text mode video panel (say 80 rows by 200 columns, and a couple of grey levels). My window manager could have been a frame enhanced Emacs running under MSDOS, and I would have been happy as a pig in poo.

I never found much use for micro-managed gun turrets with more screen area devoted to window cruft than document content.

Considering that a 1.35GB hard drive cost $1800 in 1993, and a 1.5TB (over 1048x the storage) is only $90 right now, and as 2MB is ~.15% of the 1.35GB disk, and 2GB is ~.13% of the 1.5TB disk, I'd say he could like it less and it'd still be an improvement.

Given what memory cost in 1970, I suspect that using 64-bit time would have been an expensive decision.

A lot of gross little hacks look like (and are) great ideas when hardware costs a fortune and you don't yet know how persistent legacy stuff is going to be.

Wow. Here's a proper link. It's well worth a look, especially the graph. The first data point from 1957 is particularly astounding: 411 million dollars per megabyte. Although the computers of the time would probably have run out of address lines long before you had installed all the 10000-bit flip-flop arrays.

I have copies of advertisements I made for my computer store circa 1987 (3 years after 1984). 1MB of ram was $350 then at the retail level. I suggest checking out the graph at the following site for an insight into memory prices over time http://www.jcmit.com/mem2002.htm.

When it was release 128 mb of ram would cost $671,088,640.00. memory costs.

this list has it at $1,229,575, or $34,000/month rental.

This was also pointed on Saturday's Slashdot Story

A mere $US5,000 would be something of a price sensation by the standards of current large capacity SSDs, whose prices aren't dropping nearly as quickly as are those of magnetic media.