Computing Event of the Year
The Magic Kingdom Cracks It was a bad year for Internet anarchists. The results from it started coming in, and the real world started saying, “No.” The real computing product of the year was spam, spam, spam, spam, accompanied by a healthy side order of viruses.
The first shots in the IP Wars were fired as an empire began to strike back, and the first casualties came in.
This is going to be a long, long war, lasting years, on a worldwide scale, more likely decades, and only a fool would think otherwise.
Ironically, the vast majority of participants in this war won’t even really know what it’s about. It will be virtual reality vs. real reality. It’s really about bringing the real-world and real-world order and laws and rules to cyberspace; taming the Wild, Wild CyberWest.
The end result is historically inevitable; it must and will be tamed, just like the Wild West was tamed. There will always be a criminal element, just like there is in any society, and that element will be heavily flavored with young, rebellious men, just like in any real-life society.
But from this point on, it is going to be growingly called what it is, and order, not anarchy, will be considered the norm.
The Slump Ends: PC sales in North America finally ended a long slump. It boosted Intel’s profits and cut AMD’s losses. It came in the nick of time to keep AMD from going into a death spiral of continuing $200 million losses, and buys it time to get Hammer established.
The bad side of this evidenced itself in component manufacturers trying to keep the prices of high-end components high. Video cards led the way, others appear to be following.
Intel Hits A Wall: The causes aren’t entirely clear, but Intel couldn’t get Prescott out the door in 2003. Some of the reasons appear mundane, but for the long run, Intel doesn’t seem to have new answers to the old question of how to make their CPUs ever faster. The appears to baffled as to how shook off the doldrums and came out with a series of CPU and motherboard products that were as good or better than the competition.
Overclocking Product of the Year
AMD Thoroughbred B Processor: The first generation of this line was lame, but the next generation ran like hell on very little hay and became the best bang-for-the-buck CPU of all time, with prices as low as $50.
Intel Northwood 2.4C Processor: This was the top performing overclocked CPU for most of the year, but was overshadowed by the extraordinary price/performance of the TBredB.
Western Digital Raptor: In an industry dominated by “me-toos,” Western Digital led, and jumped into the world of 10K IDE drives.
Western Digital JB Hard Drives: This was the premier IDE hard drive for most of last year. It was the premier mainstream IDE hard drive for most of this year. A classic computer product.
Dog of the Year
nVidia FX 5800: In a word, Dustbuster.
Winner of the Year
ATI: This is a repeat win for the Canadian company. Last year, we said, “ATI put itself in position to stand toe-to-toe with nVidia over the entire video card range in 2003.” This year, they did it.
While ATI’s market share didn’t skyrocket overall (it did go up), its market share of the high-performance/overclocking audience (which is often a leading indicator of the overall market) did. Our surveys during the year indicated nVidia buyers are few and far between.
If ATI three-peats next year, nVidia will go the way of Voodoo.
Loser of the Year
nVidia: They made some dubious design decisions, and tried to whine and code-fiddle their way out of the less-than-stellar results. It didn’t work, and left the impression the company was a bunch of babies tossing temper tantrums. While there is good reason to expect 2004 to be a comeback year for the company, if nVidia doesn’t execute in 2004, they’ll be executed.
Futuremark: They released an unfocused benchmarks, then proceded to get their credibility shredded in the video card ethics wars.
Slick Willy Award
AMD and the FX-51: They realized at the last moment that they had blown it by not having a dual-channel desktop solution for Hammer, so they took the cream of their Opteron crop and called it the Opteron That Isn’t An Opteron (aka FX-51) to buy them time for a real product. The FX-51 generated few sales, but much positive media talk.
Intel Sees FX-51 And Shrieks EE!!: Intel took one look at what AMD did with the FX-51, and said, “If you can’t beat them, join them.” Out of the blue, they hosed out a Xeon, and called it a PIV Extreme Edition. They sold even fewer of those than AMD sold FX-51s, but it at least took the edge off the media’s praise of the FX-51.
Trends To Watch In 2004:
The DVD Burner Cometh By the end of 2004, dual-layer DVD burners will become near-standard if not standard equipment on PCs, and Hollywood will start hyperventilating.
AMD And Hammer: Some of you may wonder, “Why wasn’t the Hammers considered Products of the Year?” The primary reason is that few Hammers were made in 2003. 2004 will be the make-or-break year for Hammer.
Given an upsurge in PC sales, and what looks to be a year of relative stagnation by Intel, Hammer ought to do well in 2004. However, there is reason to believe AMD could snatch defeat from the jaws of victory by going from one extreme to the other in valuing their processors.
What AMD needs to do is get somewhat more money from its current users and attract millions of new ones from the Intel ranks. What they are in danger of doing is overpricing their products, getting neither, then be forced back into the bargain basement again.
XPensive PCs Rise Again: The performance PC got commercialized this year, as evidenced by the rise of high-end PC OEMs promising all of the performance with none of the effort. I recently saw news of an upcoming $80,000 PC in Maximum PC a few weeks ago (OK, that one is supposed to use RAM rather than a hard drive).
It’s hard to argue that the emergence of a luxury section of high performance computing hasn’t been responsible for products like the FX-51 or PIV Extreme Edition.
What is unclear at the moment is how the existence of that luxury section will affect the rest of the high performance market. Will the luxury segment essentially throw tons of money to get a little more out of reasonably priced “best-equipment,” or will mainstream-priced equipment be downgraded to justify the price of the top-end products?
In other words, will the existence of the luxury sector spawn over-the-top products like 4-GPU video cards, or will it ensure that you’ll get a decidedly second-rate video card if you don’t pay over-the-top prices for one?
Slowdown In Circuits The delays in Prescott are not in themselves terribly important, simply because AMD hasn’t been in a position to capitalize on them due to its own delays. Further delays and/or a lackluster Prescott generation could boost AMD’s sales, but probably not more than a few percentage points towards the end of the year.
What could be far more important in the long run is why Intel has had such problems. To oversimplify, is the old policy of “shrink and ramp, shrink and ramp” hitting a variety of walls?
Put another way, how hard will it be to make a 10GHz CPU? Can you even feasibly make a mass-produced, affordable 10GHz CPU following the traditional rules and procedures, or will it be back to the drawing board for some radical rethinks?
The issue isn’t “Can you make a 10GHz CPU any way?” but rather “Can you make a 10GHz CPU the old traditional way that fits into the current PC CPU parameters?”
You probably can’t.
The key term is “current PC CPU parameters.” Could you make a 10GHz CPU the old fashioned way? Sure, if you also made something like a $500 Prometeia or better required equipment. That’s not going to happen in the typical Joe Sixpack machine.
That doesn’t mean you won’t see the equivalent of a 10GHz processor. It does mean at best it will be a much different beast than what we have today, and it will probably show up later rather than sooner.
At worst, it could mean that the average person will prefer a cheap, small, portable, good-enough 4GHz to an extremely expensive 10GHz, and the CPU market will bifurcate into really cheap and really expensive CPUs.