The Red Hill Guide to Computer Hardware

January 6, 2007 11:01 PM

The Red Hill Guide is an amazingly detailed and well-written compendium of desktop hardware old and new, with a focus on PC and x86 compatibles. Look for your first CPU, hard drive or mainboard.
posted by loquacious (37 comments total)

Coral cache here.
posted by loquacious at 11:02 PM on January 6, 2007

With the half-life of hardware being as low as it is I decided early on that I would not concern myself with hardware details.
Years later and glancing at this site I'm glad.
posted by jouke at 11:32 PM on January 6, 2007

Great hardware fetish site--thanks for sharing, loquacious.
posted by retronic at 12:43 AM on January 7, 2007

I remember marveling at the speed of a directory scrolling past when a rich buddy stopped by with his brand spanking new 286...

sighs
posted by Samizdata at 1:01 AM on January 7, 2007

Intel Celeron 633, 666, 700, 733 and 766

The forgotten Celerons, and justly so. These grew up in the shadow of the AMD Duron, still crippled by a stone-age 66MHz bus speed and utterly unable to compete. No-one bought them.

Well, no-one who knew anything at all about computers bought them;

This is a joke, right? Virtually all coppermine CPUs could clock to 1GHz. Running a Celeron 633 or 666 at 133MHz FSB was easy, and a price performance home run! Easily as impressive as the Celeron 300A was, in its day.

Also, and not surprising, they completely fail to mention dual CPU systems. I've never really understood why people failed to see the advantages of 2 CPUs for power users.
posted by Chuckles at 2:13 AM on January 7, 2007

Oh, now I'm going to have flashbacks of discussing the 386SX in WWIV BBSes at 2400 baud circa 1991.
posted by Rhomboid at 2:26 AM on January 7, 2007

I've never really understood why people failed to see the advantages of 2 CPUs for power users.

As I recall, it was because the dual cpu motherboards cost five times as much.

Also, people suck at writing proper multithreaded code. Single user apps often take no advantage of multiple cores.
posted by flaterik at 2:37 AM on January 7, 2007

The main board: The most critical part of your PC

He/she lost me there. Yet the layout is so nice and the topic is so close to my younger years, I can't help liking it !
posted by elpapacito at 2:58 AM on January 7, 2007

Rhomboid:

WWIV at 2400 baud? Now that brings back memories. I ran a seriously hacked one with the blistering 64K program space of Turbo Pascal 2.14 under DesqView...
posted by Samizdata at 3:35 AM on January 7, 2007

As I recall, it was because the dual cpu motherboards cost five times as much. ... Also, people suck at writing proper multithreaded code.

They did cost more, sometimes as much as 2x the price, but often only 20%. When I paid $500 CAD for a P2B-S (in 1998?), I think the P2B-DS might have been pushing $1000. A year or two later, the BP6 was only a little more than any other BX board on the market, and that held true for VIA694 based systems. Now, the AMD dual CPU boards were expensive, perhaps $300 when a single CPU equivalent was only $100, and at that point Intel decided to stop allowing Pentiums to run dual, but those events were late in the history documented by that site.

As for multi threaded code, that is the standard criticism, but it completely overlooks the practical benefits. 2 cpus are better because one processor can be hung up doing number crunching while the computer remains responsive to user input. Dialing down thread priority can help to keep a single CPU system at least a little responsive, it doesn't come close to the miracle of a second CPU.

It was said that gaming performance was hurt by having 2 cpus, due to added overhead, or something.. That might have been a reasonable complaint (although I tend to doubt it was, in practice).

It is all moot now, thankfully!
posted by Chuckles at 4:00 AM on January 7, 2007

Nice site, nice pictures, nice memories.

Although <geez> I couldn't find my first micro cpu-
6809e (CoCo) </geez>
posted by MtDewd at 5:44 AM on January 7, 2007

Hah, while everyone else was limping along at 4.77 Mhz, my first PC had turbo mode, 8 Mhz baby, and dual floppy drives, no HD.
posted by caddis at 6:05 AM on January 7, 2007

caddis

My first computer was a 486 @ 33MHZ from Gateway 2000, which at the time was probably considered the *best* manufacturer.

Me and my brother played a Tetris clone called Doubleblocks on that machine, and he could never end up beating my score... which I, of course, achieved by depressing the Turbo button to slow down the clock-dependent gameplay. :-)

My second and current computer (PIII 933MHZ, purchased in 1999) was also a Gateway, because they were at least still somewhat respectable 'round that time.
posted by The Confessor at 6:43 AM on January 7, 2007

My 486SX-25 still boots. DOS 6.2 for the win. Battlechess, anyone?
posted by BeerFilter at 7:42 AM on January 7, 2007

caddis, my first was a blazing 10MHz 8088. With the maxed-out 640k of RAM and a Hercules monochrome graphics display. Now that was power!
posted by Foosnark at 8:36 AM on January 7, 2007

my first was an atari 130xe, which was already badly outdated at the time ... that was a bit of a joke, but managed to fascinate me anyway

the 286 i got next was ok ... it wasn't until the 486-25 i got that i started to feel like i was really working with "power"

nowadays, i pretty much just web cruise and write in open office, and a 2600+ sempron seems ridiculously overpowered for that ...
posted by pyramid termite at 8:50 AM on January 7, 2007

Eponysterical.
posted by AwkwardPause at 8:54 AM on January 7, 2007

A year or two later, the BP6 was only a little more than any other BX board on the market, and that held true for VIA694 based systems.

Yeah, I had a BP6 at one point. I ran it with one CPU for a couple of years, then threw in another for $30 from eBay, when they got cheap. It was a pretty neat way to extend the life of the computer, although you did run into random driver crashes because no one bothered to test on multi-CPU boxes.
posted by smackfu at 9:30 AM on January 7, 2007

Wow, thanks loquacious for pointing out this site - what a trip down memory lane.

Rhomboid - my first computer was a 386SX16, I was pissed to find out that the SX didn't come with a math co-pro. Heh - RAM didn't come in banks, I remember having to pull out and insert individual chips and soldering pins back on that I accidentally broke off.

I've been upgrading pretty regularly since then, until my current XP1800+ I got, what now, 3-4 years ago?. It's pretty much at the end of the upgrade cycle (mobo is AGP and EIDE) but I haven't really felt compelled to upgrade. Strange.
posted by porpoise at 9:34 AM on January 7, 2007

They mention the 6502 but not the mighty 6809 that powered my precious CoCo? Hmmf.
posted by ROU_Xenophobe at 10:23 AM on January 7, 2007

They did cost more, sometimes as much as 2x the price, but often only 20%....but those events were late in the history documented by that site.

I stand corrected. I started buildng computers around the 486 era, and by the time I was considering dual proc setups it was -much- later. IIRC, the only duals that were available when I was pricing them were early Xeons, which were hideously expensive.

And my critique of multi-threaded code comes mostly from being a developer that writes code which is heavily multi-threaded. I've made plenty of my own mistakes, and witnessed a lot of... interesting... techniques in that arena.
posted by flaterik at 10:31 AM on January 7, 2007

Kickass. I have an old box with a bunch of my own (starting with a p75, which I was eventually able to over clock to 133mhz) On to a k6 and then an athlon. I also have older chips from surplus computers. I think the lowest is a 286.

I got my first computer in June 1995, and I got a 'certificate' for a windows 95 upgrade, which didn't come until like January. I had so much fun reading through old stuff, writing x86 code and dos programs. I find old processors fascinating.

These days people don't have an intimate connection to their computers, because they have so many. iPods, cellphones, pdas, game machines that they'll replace in six months. They never have any desire to get close to the metal like people did (i imagine) when they only had one machine.

By "people" I mean "computer nerds" of course.

I've actually designed a CPU from the ground up from logic gates in CAD software for a class. It's amazing how simple they actually are, or can be. At the lowest level a CPU instruction actually triggers multiplexers directly and 'connects' various parts together like an ADD connection will connect the adding unit to the two registers you want to add.

Of course modern CPUs are far more complicated, and often will decode CISC instructions into RISC instructions in their cores.
posted by delmoi at 11:09 AM on January 7, 2007

My first computer was a TRS-80 Color Computer with 16k of RAM that my dad bought us when I was about six. One of the very first programs I wrote was a BASIC program that drew a swastika (!). My dad, a gentle man of Jewish descent was horrified. It took my mom quite a while to explain to me what was wrong. I remember us hooking up a tape drive from radio shack to load and save programs, and the joy when we got our first 5.25" floppy drive. Our next upgrade was this bad boy. Man, that was a sweet rig, 9" screen notwithstanding.

I'm not sure I understand the logic underlying the assertion from the site that "by definition, a PC uses an X86 CPU." I mean, I understand that "PC" has come to be shorthand for an "IBM-PC compatible," but I would think that a site devoted to explaining the history of the x86 CPU could spend a little time explaining the derivation of the term "PC" as we know it today.
posted by Doofus Magoo at 11:29 AM on January 7, 2007

With the maxed-out 640k of RAM

Hah, I had a full meg on that motherboard, and a driver which let DOS access memory above 640K (excpet for 184K or so used by DOS for overhead).

Anyway, a trash80 beats a 1st gen pc clone for antiquity. I do remember using one at work, yes work kiddies, and saving the program to cassette tape. Uggh. That brand new IBM PC we got that year seemed leaps and bounds ahead and the TRS80 became an instant paperweight.
posted by caddis at 12:09 PM on January 7, 2007

Damn, I'm trying to find the behemonth 1GB drive that my Dad bought. It was a 5.25" drive and black which was really weird at the time especially in the beige case.

And that trusty Sony 1X external CD-ROM. Heh, I still remember the FMV of the girl playing a movie ticket seller "guiding" me through the next generation of entertainment!

I think that damn thing still works, too.
posted by Talanvor at 12:24 PM on January 7, 2007

BeerFilter - Battlechess... wow... that totally rocked my eight-year-old world.
posted by djb at 1:05 PM on January 7, 2007

2 cpus are better because one processor can be hung up doing number crunching while the computer remains responsive to user input. Dialing down thread priority can help to keep a single CPU system at least a little responsive, it doesn't come close to the miracle of a second CPU.

That wasn't really true when the Intel MPS systems first came out. Dual Pentium boards shared the L2 cache which could lead to worse performance if you were running a pair of sequential programs. They fixed that with Pentium Pro, though.

Pre-MPS 80386 and 80486 systems required all sorts of weird proprietary tricks to get around various hardware limitations, and you really only bought them if you had specific parallel processing needs.
posted by cmonkey at 1:23 PM on January 7, 2007

Chuckles writes "I've never really understood why people failed to see the advantages of 2 CPUs for power users."

Because even with the quietest cooling solution twin-CPU systems are insanely loud, perhaps?
posted by clevershark at 2:12 PM on January 7, 2007

My first computer was a single board 8085 that I designed and built myself while in college in 1978. Its only memory was 256 bytes of static RAM. I implemented a front panel like a minicomputer in LSTTL that allowed me to enter programs in binary. The only I/O was the 8 LED output latch, and the single bit accessed by the RIM and SIM instructions. The board was wire-wrapped. I still have it, but I haven't tried to use it lately. The big advantage of the 8085 over the 8080 was that it only needed a single 5V supply.
posted by rfs at 6:07 PM on January 7, 2007

Thanks cmonkey, that's really interesting to know. 5 or 6 years ago I contemplated buying a dual pentium board from a local surplus store, just to see how it would go, but they wanted too much money. Sounds like it wouldn't have been worthwhile at all..
posted by Chuckles at 7:28 PM on January 7, 2007

(I think rfs wins this one.)
posted by caution live frogs at 9:20 PM on January 7, 2007

The Red Hill Guide is an interesting read for your PC nostalgia needs, but it contains a number of outright factual errors, particularly in the early stuff (where Intel is credited for the design of the 6502, for example, and their understanding of the early XT and AT platforms is somewhat confused). As the guide says:

"The only one of us who was out of short pants in 1985 was still a happy Z-80 user long after the 286 was old news."

These guys seem a bit young to attempt to write computer history purely from memory. Although I admire the site as a project, it would be tremendously improved by being made factually correct.
posted by majick at 6:34 AM on January 8, 2007

Where do they credit Intel for the 6502 design? All I see is "A development of the Motorola 6800 by ex-Motorola engineers, the 6502 and its numerous variants..." which is factually accurate, and then they appear to accidentally swap the detail tables for the 6502 and the 8080 above it.
posted by cmonkey at 10:58 AM on January 8, 2007

If those are just swapped detail tables, they're both wrong. Admittedly, the text is completely correct about MOS history and the 6502 being a descendant of the 6800, but even there they appear slightly confused about how the most common 6502 and Z80 parts were clocked.

A bit further down that page, they blame Intel for both segmenting memory -- a legitimate beef -- and some other unnamed architectural flaw somehow mysteriously causing IBM's decision to reserve the upper 384K of the address space for memory-mapped devices. The latter seems to me to be pure hokum.

They seem to think the 68K petered out with the release of the 040, but while the 060 wasn't wildly popular on the desktop it certainly put in an appearance, and was surely used in far more desktop machines than the 186 which merited mention. This isn't so much a factual error as one of omission, but it's certainly noticeable.

They imply that the V20 and V30 were somehow rare or eclipsed by the 286 and that the part was irrelevant in the market. Dirt cheap V20 machines were everywhere at the end of the PC/XT era, when 286 AT-class machines were still relatively high-end desktops and servers. The V20 was used in a ton of low-end no-name clones because was a cheap part (compared to Intel) that lived on cheap 8 bit IO boards. Compared to an 8086 XT, a V20 PC was cheaper to produce, so they were sold by plenty of low-margin manufacturers.

It may be a typo, but they mention the use of a 286-10 in an XT. I think they mean AT.

Don't get me wrong, I find the whole site interesting and a lot of fun to read, and a vast majority of the content there is correct. It's just a shame about the errors I found on the very first page I clicked.

I'm also surprised they don't nominate the Western Digital Caviar 170 for special mention. I don't think an enormously popular drive with a failure rate that high existed before or since, although that very first IBM 1G disk probably comes close. Both disks are very noteworthy as part of PC history, although the WD 170 is especially so. Wasn't there a class action suit over it?
posted by majick at 11:53 AM on January 8, 2007

Where do they credit Intel for the 6502 design?

The table lists Intel as designer and manufacturer, both of which are incorrect, I believe.
posted by caddis at 12:01 PM on January 8, 2007

rfs wins so far...The 286 was definitely an AT.

Anyone remember the Bernoulli box? 10 Mb removable cartidges...back when geeks were actually...well, geeks.

In 1983, I took responsibility for 20 different computing systems as a college TA and computer room 'manager.' Some of the systems (I can't remember them all) were an IBM 5150, Wang Minicomputer, all our (16) new IBM PCs, and Apple IIs, and my pride and joy, an Apple Lisa. I was in heaven. That summer I took delivery of 16 new computers from Apple, the Macintosh - the first 16 shipped east of the Mississippi. I'll never forget my profs face when he saw me with a PC open and apart on a desk one night...

I remember completing a 60 hour semester project in managerial accounting in 3 hours using Visicalc. I was computer God. Chicks dug me. It was great.

Thanks for bringing me back.
posted by sfts2 at 2:06 PM on January 8, 2007

Yes, before we wired our several boxes together in a proto-network (as engineers we were on our own, there were no computer help desks, network admins, or even network) through the printer ports we used Bernoulli cartridges to transfer data between machines and for some back-up. Visicalc was an amazing program, and Lotus 123 even more so. We wrote heat balances using Lotus. It is really a pretty good programming language for engineering. We were able to quickly model heat exchangers, turbines, etc. The dedicated programs of the day were expensive and mostly ran on mainframes. Some of our runs had to be done over night to get them to settle out though (heat balances are iterative processes). The one thing that made it all work was that Lotus was written in assembly language so that it was blazingly fast, even on a slow processor.

Speaking of ancient history, did anyone else use the AIM development system to program 6502s? I remember building control circuits and programming them in assembly language using this. The 6502 was one elegant processor.
posted by caddis at 2:29 PM on January 8, 2007