Slashdot Mirror
486 Turns 15 Years Old
wooby writes "The 486 processor , introduced in 1989 at 25 and 33MHz clock speeds, is now 15 years old. Intel's simultaneous launch of both the 486, a CISC chip, and the i860, a RISC chip, was a gamble. Remarks Intel's former CEO, Andy Grove: 'our equivocation caused our customers to wonder what Intel really stood for, the 486 or i860?'"
BLURB
Intel's venerable 486 CPU is now 15-years-old. Intel began working on the 486 in the early 1980s, and introduced the chip in April of 1989. The 486 was essentially an improved, modified version of the 386. The 32-bit 486 was initially manufactured on a one micron process, and was introduced at speeds of 25 and 33MHz.
All 486 chips except for the "sx" versions came with a built-in floating-point unit and contained 8 KB of cache memory. The 486 was capable of 20 MIPS performance, and contained certain features (such as pipelining) which had previously been found in mainframes. As a result of these enhancements the 486 was theoretically able to execute one instruction per clock cycle. Today's processors have clockspeeds 100 times faster than the original 486, but the instructions per clock (IPC) of the latest CPUs isn't much better than the IPC of the 486. Intel also decided to release the 32-bit, superscalar i860 CPU, which was specifically designed for scientific applications, in 1989. In Only the Paranoid survive, Intel's former CEO Andy Grove recounts the dilemma of launching two largely incompatible CPUs at the same time:
We now had two very powerful chips that we were introducing at just about the same time: the 486, largely based on CISC technology and compatible with all the PC software, and the i860, based on RISC technology, which was very fast but compatible with nothing. We didn't know what to do. So we introduced both, figuring we'd let the marketplace decide. However, things were not that simple. Supporting a microprocessor architecture with all the necessary computer-related products - software, sales, and technical support - takes enormous resources. Even a company like Intel had to strain to do an adequate job with just one architecture. And now we had two different and competing efforts, each demanding more and more internal resources. Development projects have a tendency to want to grow like the proverbial mustard seed. The fight for resources and for marketing attention (for example, when meeting with the customer, which processor should we highlight) led to internal debates that were fierce enough to tear apart our microprocessor organization. Meanwhile, our equivocation caused our customers to wonder what Intel really stood for, the 486 or i860?
Compaq recommended to Intel that it abandon the i860 and concentrate all of its efforts on the 486. Microsoft pressured Intel to promote the i860, and strongly encouraged Intel to introduce an i860-based PC. Intel decided to emphasize the 486, and ended up selling hundreds of millions of 486 processors. It is intriguing to think of how different the computer industry would be today if Intel had decided to emphasize the i860 instead of the 486.
ask and ye shall receive
I'm a lvl25 Artist in the game of Life (tm)
Later in the year, IBM introduced an upgrade kludge 486 piggy-back board that could be shoehorned into their 386-based PS/2 Model 80s. However, IIRC, these all had to be recalled due to the bugs in the early 486s.
End users didn't get to see a significant number of correctly functioning 486 systems until early in 1990.
BTW, if you ever saw the processor specs for the i860, its byzantine complexity made the x86 architecture look clean and elegant. There's no wonder it never took off.
I remember when it was first released the 486 was billed as the "Cray on a chip." There's just no underestimating the hubris of marketing.
I am Sartre of the Borg. Existence is futile.
I always wanted to have sex with 16 year olds when I was 6. Damn age of consent laws stopped me everytime though :P
I think they were more fun.
:-)
My 486 was a 486SX/25 with 4MB RAM 170MB Conner HDD, later upgraded to 8MB with a 2x CD-ROM drive for the small fortune of just over $1,000 (AU).
Back then, incremental versions of Microsoft products provided actual functionality. Memmaker (bundled with MS-DOS 6.0+) was a godsend for anyone who had sat down trying "loadhigh" (autoexec.bat) "devicehigh" (config.sys) and the ordering of drivers to get more than 600KB of conventional memory free.
Games back then had more depth and bredth - I was a big fan of the Sierra/Dynamix series - Kings Quest 5, 6 and 7, Space Quest, Leisure Suit Larry 6, Quest for Glory 3 and 4, Police Quest 3 were among the games I spent too much of my childhood playing
As a side-note, I remember Kings Quest 6 in particular came in a 12 or 13 (I think) disk set. Included was a note saying that in order to hear the Theme Song (Girl in the Tower), I would have to call a radio station in the USA and request it. For an 11 year old Australian, that bummed me out!
Ahh.. I don't know where it is anymore, but that box se.. oh crap. I left it in the crawlspace in the roof of my old house. C'est la vie.
Intel released the I960 as an embedded chip, expecting to see some military applications. The first versions were the i960KA (without floating point) and i960KB (with floating point). They didn't get all that far in the marketplace. However the i960CA and its followon the i960CF were pretty slick. The i960 had 32 general purpose registers, and a processor-defined function call sequence that always placed a set of 16 on the stack ("caller-owned") and left a set of 16 alone ("args , temp & return values"). The i960CA cached the top 4, 6 or 8 stack frames in on-chip static memory with a 128-bit pathway to the main register set. This gave it amazing function calling and interrupt service performance. We wrote a sample clock-interrupt test that serviced a 100 kHz clock interrupt using only 23% of the CPU. (Remember, this was in 1992...) The product we built (see next paragraph) is still out in the network, switching phone calls.
I remember receiving one of first the 486DX2/66 processors in the city where I live (Columbus Ohio). I was at AT&T/BL at the time, and we were building a product based on a pair of 66MHz 486 and a pair 33 MHz i960CA processors. (Intel sent us a pair of chips for evaluation) We wanted to benchmark them, and I was the only developer whose home system could use the 32-bit capabilities of the 486. The 486DX2/66 was a screamer...
<offtopic>
Being a total geekazoid, I had UNIX (yup, I blew $800 on a "used" SVR3.2 license)! I kept that license current through Novell UNIXware SVR4.2 in 1996, when this new geek-friendly OS called "Linux" had just received BKL-based SMP capability. I tried it, liked it, and kept using it. This "Linux" already had better VM performance (in my opinion) than the traditional UNIX, and semed to me to be on the way to much larger things. I stopped updating my UNIX license, donating it instead to a local high school.
I've been a developer for >30 years and have a clear idea of what I want in a workstation. Linux (and to be honest, including the valuable GNU utilities) provides that set of capabilities better than any other system I've ever used. I don't know about MacOS X, it might be pretty good. But in my experience, Linux has no peer. FYI, this experience includes every Microsoft operating system, every IBM mainframe operating systsem up to VS/ESA, PDP-11 DOS/Batch, RSTS/E, RT-11, VAX/VMS, Data General RDOS, AOS, AOS/VS on the MV4000 and MV8000, classic UNIX on a 68010, UNIX on IBM/Amdahl mainframes, BSD/OS on PCs, SunOS on sparc/2 and Sparc/10, NetBSD and OpenBSD. I also tried out Next and Apollo Domain. Sun and the BSD's came closest to Linux in quality.
Everything else is an also-ran. Finally, at present my day job involves embedded Linux. I've worked with both uClinux (m68k) and real Linux (MPC860 & 826x), (mostly updating and debugging) drivers for both. I have *never* seen a system as robust. Linux itself, the development process that led to its existence, and the ongoing development process that allows it to be such a powerful system, are all major treasures for those willing to recognize them.
</offtopic>
True, you will find that most highend RAID-5 SCSI controllers use a i860 or the i960 to do all the dirty XOR'ing and its doing it fine because of SIMD intructions the chip has to offer.
Ha!
My Athlon 2200+'s fan died on Thursday so my primary machine is a 486. "Still going strong," sure, but let me tell you something.
Microsoft claims that the minimum system requirements for Intarweb Exploder 5.5 include a "486/66".
That is patently false. The only thing that runs good on here is the boot loader.
You could try Opera, it runs alright on slower machines, at least it's a bit more lightweight than Internet Explorer and Mozilla. But you really need a nice, fast Pentium, say 90MHz, to get the best out of the internet. ;-)
10 PRINT "LOOK AROUND YOU ";
20 GOTO 10
You may already know this, but for the benefit of some of our other readers...
When trapped in your room by a live leaner, crack the door open a little bit, then snap it closed. If you do it right, the leaner will be diffused.
Then make sure you find who did it and penny them into their rooms. That's a lot harder to open from the inside ;)
Age of consent is 14 in canada. The sexual partner can be an adult on the condition that both parties are consenting and that the adult is not in a position of autority over the minor, that is is not a parent, a teacher, a civil servant etc... Also, no money may be exchange for the service since that would be prostitution which is illegal.
I
At least some versions of the i860 had the same
MMU as the Pentium. Using the MMU for paging was
horribly difficult though, because the i860 did
not handle faults well. The OS got stuck with the
job of emulating many partially completed instructions.
Intel used the i860 in the Paragon supercomputer,
which ran a SysV UNIX OS.
Mercury Computer Systems used the i860 on VME
boards with a circuit-switched crossbar interconnect that did 160 megabytes/second
(40 MHz, 4 bytes wide) half-duplex to each node.
That's 1.28 Gb/s, many years ago. They sold the
system with a matrix math library for doing
radar and similar tasks.
I think the non-MMU version got used in printers.
Plus those 'old' computers are a lot more durable than ones made today. The old XT keyboards were made from steel. Even into the late 1980s, IBM keyboards still had a steel plate underneath. The IBM PS/2s had steel cases, you could use the case in place of cinder blocks to raise up your car.
If you long this, Matias has build a mechanical keyboard called the Tactile Pro (google it buster). It's simply an awesome keyboard like they used to be. It's based on the same mechanical keys that Apple used to have on it's Apple Extended Keyboard (aka, Mac SE and Mac II era). They had to secure one million key switches from the manufacturer in order to keep them in production.
I'll be in the states in ten days. I'm bringing one of those babies back!
The original Beowulf cluster was 486-based.
It had 16 machines.
> "it could not switch back to real mode without a
> warm reboot."
>
> Bullshit. I guess exiting Windows 3x on a 286 and
> going back to that DOS prompt was a figment of my
> imagination.
That one is actually true.
Officially, once turned on, you could not leave the protected mode on the 286. IIRC there is an undocumented 'loadall' instruction which allows you to do this though. But I doubt Windows was using that one. Instead the BIOS provides functionally to exit protected mode by doing a silent warm reboot (It puts some magic value into the CMOS RAM, causes the processor to reboot and the bootup code checks for the magic value and returns to the OS).
Tobias
I browse the web, do email, a little image manipulation, and run a web server on my box.
I also run NIS, NFS, DHCP, Squid Proxy. I also run mailing lists, tape backup, and a cd burner.
I also run ftp, pop3 and smtp for a lan. Several times a day, the box fetches mail from several hotmail accounts, and alternate POP3. It also fetches and filters data from NNTP. It is also the NTP server for the LAN...
The box? a dual processor PPRO. 200Mhz with 128MB of RAM.
Works fine.
Client side? A 128MB PII 400. Works fine. Maybe one day I'll upgrade, but no reason to now.
Ratboy.
Just another "Cubible(sic) Joe" 2 17 3061
RH 6.2 seems to be the best thing out there. Redhat kept up support until they invented the idea of EOL for their distros, so if you get the ISO then all the updates, you can get a pretty up to date system. 2.2.24 kernel, most of the modern libraries XFree86-3.3.6. You can build dillo, links and have a usable system. Really, I've got two 486 laptops with 20 megs of RAM, I can build the latest PCMCIA release and have full wireless. And those 486 were TINY and cool. Perfect for acting like a remote control for my myth TV system or a vncviewer for my "macho" system.
What about Donkey on the original IBM PC, which was distributed as source? That was written by Gates, in part and was MS software.
Or there's this.
For the Google impaired, here's the Tactile Pro page and a review.
Alex Bischoff
HTML/CSS coder for hire
Not really. The PowerPC that the mac uses now has very little to do with the 68000 series chip that used to be in the mac. The PowerPC is a decendent of the IBM POWER chips found in their high end RISC servers and workstations at the time. The transition from the 68k to PPC was very transparent to the user, primarily because Apple did a damn good job with it, but it was a completely fresh start CPU wise.
-twb
All of this "around here, the age of consent is X" stuff is getting silly, so perhaps we could just look at the great big reference.
It doesn't look like you ever actually used one much. Funnily enough (given the parent article), I had the pleasure of using the Happauge 4860 motherboard which took both a 486 and 860. It was great - I could run UNIX on the 486, and then compile and build 860 programs, running programs on both processors simultaneously. It was good for me, because I could use the 860 simply as a coprocessor, but a very powerful one (it left even a 4167, let alone the 486's onchip FPU in its dust..), never having to worry about actually dealing with an i860 OS (of which there were some ,but I never had one...)
:) And on that note, in typical Slashdot style, the article has a simple factual error: the 486 and 860 were _not_ released simultaneously; the 860 came first.
For all its quirks, I wouldn't agree that the 80860 was a sucky processor. It was fast, but weird. Faster than anything else out there, mind. There were no Alphas back then. It left the 486 in its dust, at least until the very latest 486s (100MHz DX4s, etc)
It was also the first non-RAM million-transistor chip from anyone, ever.
Let's do an overview here:
4004 - first uP
8008 - Intel's first 8-bit uP
8080 - Other 8-bit uP
8085 - CMOS 8080
8086 - Intel's first 16-bit uP
8087 - Math coprocessor for 8086/8
8088 - 8086 with 8-bit bus, meant it could be used with 8080/8085 chipset
80186/8 - Adds many features to 86/88
80187 - 8087 with new package for 186/8
80286 - Adds protected mode
80287 - Math coP for 286
80385 - Cache controller
80386DX - 32-bit uP
80386SX - DX with 16-bit bus, can be considered 80388
80387 - Math coP for 386 (DX and SX versions)
80486DX - Improved 80386DX, 80385, on-board cache (8 or 16k), 80387
80486SX - Minus the improved 80387
80487SX - 80486DX that disables the already present SX chip
Pentium - Even more integrated chip, 64-bit bus, 32-bit internal
Pentium Pro/II/III/M, AMD K6/K7 - RISC internally, later versions have SIMD instructions
Pentium 4 - Adds SMT, will add 64-bit
AMD K8 - Adds 64-bit in all but one version
Performance is increased from 4004 to 4040 and 8008, 8008 to 8080, 8080 to 8085, 8085 to 8086 and 8088, 8086/8 to 80186/8, 8018x to 286, 286 and 386 to 486, 486, Pentium, PPro to P2, P2 to P3, P3 to PM. I didn't include AMD CPUs in that part.