Sometimes a Great Leap Forward - or backwards - wins
EVERY FEW years some new technology arrives that promises to evolve a previous one or perform an incremental improvement to the previous generation. Sometimes it's a success. Sometimes it's a monumental failure. Here's my Top Ten list, in no particular order.
1. V.92 Modems: I admit I will get a lot of hate mail by putting this as number one, but confess: do you reallycare if a dial-up modem is V90 or V92?. An enhancement of the old venerable 56K V.90 standard, V.92 gave too little, too late to make anyone excited about dial-up connections. The jump from 14.400 to 28.800 bps was spectacular. The one from 56K to 56K with-better-upstream left you scratching your head.
Released seven years ago, by 2000 most people who really wanted it already had broadband or had already intentions of getting it soon, so nobody in its right mind rushed to throw their perfectly working 56K V90 modems to the trash can and raced to buy a "new" V92 one. Described
here as "An enhancement to the current V.90 standard" that " offers a variety of advances in modem technology" it never had its 15 minutes of fame.
Among the new unexciting features it provided "modem on hold" allowing the connected dial-up user to "take an incoming telephone call without losing the modem connection" -I never knew anyone who used it-, a faster "handshake" sequence, shortening the only fun of dial-up connections which was hearing the connection noises-, and an improved theoretical upstream speed of 48Kbps instead of 31.2K of V90 modems.
The latter feature probably allowed a lot of SPAM and chain-letters from AOL users to reach your Inbox faster. Yes if you buy a dial-up modem nowadays [who does?] what you get is probably V92. But really, who cares?.
2. RIP: Probably the only technology that anticipated its demise starting with its name. Before the Internet swallowed the nice hobbyist BBS scene and anything good it had -like its sense of community, very much like the HAM radio folks-, back in 1993 some BBS software makers threw its last few punches to the air, in the form of
RIP technology: Remote Imaging Protocol, or simply "RIPscript". It replaced the outdated ANSI graphics and promised true graphical BBSing using vector graphics. It had several drawbacks: first, modems were too slow (think of it like Flash sites on a dial-up connection, only that by that time modems were at best 28800 or 33600 bps -33.6K). It's been [R]esting [i]n [P]eace even since the graphical interweb began killing the BBS scene circa 1995.
3. OS/2: After Microsoft threw a hissy fit to Big Blue, IBM took it on their own to release the successor of DOS and Windows, making it 32-bit and revamping the dull OS/2 1.x desktop with a new Object-Oriented one dubbed The WorkPlace Shell, or WPS for its friends. For a few years it even looked as IBM had a chance of winning the fight against Windows, as it had the technical edge: multi threading, pre-emptive multitasking, OO-desktop, long file names, then built-in web browser and TCP/IP, and more. IBM finally threw the towel in the fight for the x86 desktop space. Everything I wanted to say about this one I said
here already.
4. DSP-based multi-function boards: At a time CPUs were slow for the amount of work placed on them, hardware manufacturers including IBM and others looked at DSPs -Digital Signal Processors- as a way to "offload" work from the main CPU, by adding "intelligence" to the peripherals. Not only it allowed -in theory- to add new functionality to peripherals -to some degree it did, for instance by allowing some DSP based 14400bps modems to be updated to new standards- but also allowed packing several functions into a single board.
DSP software updates is the same approach that allowed non-standard 56K-Flex and USR "x2" modems to eventually be software-upgraded (via a DSP firmware upload) to the official V90 standard. On the upside, DSP based boards also offloaded the CPU, like a SCSI controller could do Direct Memory Access without taxing the processor.
Perhaps the biggest example of the DSP based multi-function board idea was the
IBM mWave sound card+modem contraption. It worked quite well at the time, but developers quickly found out that the DSP's number crunching power couldn't do anything faster than emulate a 28.800 modem without shutting down the sound card part -or at least that's what I recall of it. Developer Ted Felix has an
interesting page on the History of the mWave card.
The whole idea of a DSP-based multi-function card was a big flop if you ask me, not because the idea was bad, but because the software needs for CPU cycles quickly outgrew the fixed processing capacity of the DSP chips. Nowadays, DSPs are far from dead, and are found everywhere from cell phones to MP3/MP4 players, and have even
moved inside CPUs, like TI's DM320 for instance which power gizmos like the
OSD Linux based digital video recorder and media centre. It's just the idea of the multi-function, DSP-based board that never caught off.
Finally, Intel started promoting just the opposite: overloading the CPU with functionality. In the
words of Byte Magazine "But the (DSP) trend was derailed. First Apple, then Intel, claimed its latest CPUs could handle many DSP tasks natively. Apple's GeoPort Telecom Adapter eliminates the need for a modem by connecting the Mac directly to a phone line while using the CPU to emulate a modem in software. Intel is carrying the concept even further by defining an NSP reference platform for PC vendors and extending the functions to include wave-table audio and software-only video playback. Instead of handing off those functions to dedicated chips and special hardware, Intel's NSP approach uses the CPU to perform audio and video tasks."
Of course, it then meant that in order to use a "Winmodem" which stole CPU cycles from your OS and applications, a 66Mhz or even 100 Mhz chip suddenly wasn't enough... Winmodems, Winprinters... all over-loaded your CPU by making processing work then performed inside the hardware a work of the software drivers, overloading the CPU. That brought the need for faster and faster chips, which not surprisingly is what Chipzilla wanted to sell. I'm still not convinced that overloading the CPU was the right approach.
1/2
Threaded Mode
