maybe I've figured out the AMD Bulldozer strategy

In my prior post, I came down pretty hard on AMD's recent release of their Bulldozer cpu.  I had concluded that AMD was poisoned from the inside - that they put all their money on an engineer's long-shot idea.  That still might be right - but after giving this more thought, there is another possibility - and I hope I'm right.

I'm a big believer in approaching problems laterally (my first post discusses a lateral approach to goal setting).  What if AMD, instead of tackling Intel head-on, is taking a lateral approach?

Let's face it, AMD and Intel are in a longstanding war.  Intel clearly has more troops (staff), better equipment (their own fabs), and have honed their talents in frontal assault (making insanely fast chips).  Being smaller and without equally good weaponry, AMD has to think laterally.

As a frontal assault, the Bulldozer wasn't a total flop.  AMD made a strong effort - not as strong as the market leader - but strong nonetheless.  Now, remember the big picture - the Bulldozer is just one volley in a longstanding war.  AMD, we have to assume, has something up it's sleeve.

The market for cpus goes beyond the desktop.  In fact, many argue the desktop is a dying breed.  Where are the fastest growing trends for cpus?  Massive servers (many cores, low power) for virtualization, cloud computing, and databases.  (Not to mention tablets and smart phones.)  In AMD's case, guess what?  They actually kick ass in the server space.  Check out the CPU Mark for multi-cpu systems - AMD has three systems in the top five.

Battling over the desktop is like battling over a huge land mass that is becoming less-and-less tactical for the overall war strategy.  My prediction is that AMD will, within the next year, deliver a brutal blow to Intel in the server cpu space.  Intel, will be caught off guard, because their troops are largely amassed for maintaining desktop supremacy.

Do you believe in the future that many pundits predict, of smart phones, tablets, and browser based PC's?   If so, then complex computing is moving more and more to back-end servers.  And if that's the case, then AMD's server strategy is a pretty darn good one.

the "wish and hope" strategy

As a self-proclaimed AMD fan-boy, I have to add my commentary on their recent Bulldozer release.

I'm really disappointed.

It isn't just a matter of being disappointed in the chip.  The Bulldozer failure is a reflection on the engineers and management at AMD.  They must have adopted a "wish and hope" strategy - and that's unfortunate.  They hoped that setting a World Record overclock would matter - but it doesn't.  They hoped having eight cores would be a strong differentiator - but it isn't.

I imagine there was an AMD Bulldozer engineer who convinced his Product Marketing what direction to take the chip.  This engineer's approach was academic - focusing on architectural perfection - instead of what the market wanted.

All we wanted was a new cpu that would put a beat-down on the Core i7 - and AMD didn't deliver.  In my opinion, AMD should have held the Bulldozer release.  Would BMW release a new M5 that couldn't beat Mercede's latest AMG?  No way!!  I think I speak for many when I say I want to be proud of the cpu I choose for my system.  I've been taking a beating from my friends for years for continuing to support AMD.  I knew I would be able to dish it right back at them when the Bulldozer came out... but no... I'm still the dummy.

You goofed up, AMD.  You lost sight of what was important, you rushed a product to market, and you "wished and hoped" we'd be okay with it.  Left with the same decision, I would have simply moved the Phenom to 32 nm and made it an 8-core.  That'd be a true 8-core, none of this shared floating-point pipeline kind-of-but-not-quite-hyperthreading Bulldozer mess.  Instead, you're end-of-life-ing the Phenom.  What?!

Here's my message for everyone trying to bring a new product to market:  don't "wish and hope" that your potential future customers won't notice if you release something that isn't the best - because they will.  Instead, work with your customers:  learn how they'll benchmark your product, and learn what qualities will distinguish your product as a great one.  Then focus like hell on these "qualities of greatness" - and you won't have to wish or hope.