Mar 112005

Last week Allen and I popped into the Intel Developer Forum held in San Francisco to see what Intel has been up to on the hardware front.  Intel certainly has a lot of interesting stuff in the pipeline, which will continue to push more raw processing capacity into the market at ever decreasing price (per MIP) levels.  After oohing and ahhing over spiffy new technical bits like multicore silicon, though, I found the consistency of Intel’s messaging and product positioning over the years to be as remarkable as the tech talk itself.

Taking in Intel’s multicore chip messaging and market rationalizations, I was suddenly reeling in deja vu.  Somehow, I could see Intel CEO Andy Grove delivering nearly the same message.  But when?  And where?  Hello, Google!  Andy Grove delivered the keynote address at the 1993 BorCon in San Diego.  Al Stevens mentioned it in his C++ column in Dr Dobbs Journal (requires login). The subject of Dr Grove’s keynote was: the then-new Pentium processor and the need for new software to take advantage of that new power.  My recollection of that keynote (12 years ago!) is that it contained a thinly veiled (if veiled at all) subtext along the lines of “People don’t need a lot of processor power to read email.  Complacency is bad for business.  So, to sell bigger and better processors, we (Intel) need you (developers) to build killer (read: bigger, better) apps to create end-user demand for bigger, better processors.”  (paraphrased)

Fast forward to last week, 2005.  To fully utilize the processor capacity of multicore chipsets, software needs to be (re)written to use multiple threads of execution.  To take advantage of 64 bit chips, software needs to be rewritten and rebuilt (don’t give me that “just recompile” nonsense) for 64 bit data and execution.

What puts the smile on my face is when they insist that all of this is needed to respond to user demand.  Where exactly do they keep these users who are demanding multistream video conferencing with VoIP delivered to cell phones?  I’ll concede that most of the nontechnical computer users I know of do suffer from sluggishness in their home PCs, but in my experience that’s due to “system bit rot”, not due to their demanding anything new from the box.  Disk fragmentation, registry bloat, zombies, viruses, spyware, anti-virus scanners, and poorly designed software are what cause these poor old machines to degrade from their original snappiness over time.  Buying a new, more powerful processor will solve that sluggishness!  Then again, so would reformatting the old system HD.

Tired of looking at all that junk lying around in your house?  Dirty dishes?  Dirty laundry?  We have a solution!  Buy a new house!

Don’t get me wrong – the Intel hardware is great stuff.  Top of the must-have, my-Hz-are-bigger-than-your-Hz gadget list.  And the Intel spin is no different than it’s always been – Intel can’t survive on passive “build it and they will come“ market demand.  Intel must create market demand.  Intel needs consumerism.

There is one significant difference between the current Intel messaging and years past:  inclusion of the consumer market.  Historically, Intel has looked down its nose at the consumer end of the PC hardware spectrum.  Every Intel processor since the 8086 has been introduced as a high-end $10k engineering workstation or server part.  With the notable exceptions of the 80186 chip, every CPU in the x86 lineage has been a mass-market success, even with Intel’s best efforts to deny the consumer market.

That may be changing.  Not only is Intel rolling out the sexy new chip technologies in consumer-oriented processors along side the high-end processors, but the Intel demand creation engine has its sights set on the end-user market as well.  Intel may be finally waking up to the idea of marketing CPU power to end-users.

Kudos to AMD for realizing this three years ago.