Wednesday, March 07, 2007

What if Opteron never existed?

Earlier today, I was asked that question as well as the following:

  • Would processor prices be going up?
  • Would dual-core technology even exist?
  • Would we all be running Itaniums?
  • Would we need Itaniums to get 64-bit Vista?

My reply:

The first thing would be for me to blame AMD for the current throttling back of processor speeds.

(This is in addition to the loss of income I suffered nearly seventeen years ago from the AMD 80387 math-coprocessor battle, which decimated math-co prices across the board. On the other hand, should I thank AMD for opening up the market for the tons and tons of math-coprocessors I sold when prices were ‘normalized’ by the AMD math-co?)

Prior to the debut of the Opteron, we were on a direct collusion trajectory with Itanium, that all-new be-all of computing. It was going to be the Swiss-knife of computing, allowing us supercomputer power at each of our desktops, and utilizing that most current of computing architectures, the EPIC.

Before that though, we were going to go through several iterations of the Pentium architecture. Personally, I was looking forward to a 10 GHz Pentium 6 without active liquid cooling – no need to have a space heater in my office.

If the Opteron had never existed, it would still be a Pentium world, without a doubt. The mobile Pentium, based on the P-III, would have been slowly ramped up in speed and would have been hitting about 4 GHz right now. As I stated earlier, the desktop Pentium and the Xeon would have been continually cranked up relative to the gains in processing power of the Athlon.

All the while, we would have been inundated with flackware and ink about the coming goodie called Itanium.

Processor prices would have remained stable for mainstream CPUs, with newly introduced CPUs continuing to command ridiculous prices a la Extreme Edition.

It was a very slick way of raising prices or introducing ‘premium’ pricing into the market, disguising it as new product with attendant bells and whistles. In reality, it was just a cranked-up unit just like the mainstream CPUs.

Dual-core? Are you kidding?

Remember, it IS Intel we are talking about!

Remember the 80487?

What was the function of the Intel 80487 math coprocessor? It was to turn off the primary processor in an Intel 80486-80487 CPU-math-co combo.

What do I mean?

Let me jog your cobwebs.

At the intro of the 80486 (i486), the math-co world was still burgeoning, however, due to AMD’s entre into the math-co space with the 80387 for which it (AMD) was still in litigation with Intel, margins were very low. For the 80486, only high-end match-coprocessors were worth stocking for margins, and the Wietek math-co cost over $1,200.00 USD back then. Intel’s solution was a perfectly good integrated CPU/match-co chip sold as being without a math-co. When you purchased the 80487, all it did was to turn off the original 80486 chip and perform all processing itself.

Yes, the only difference was an additional pin on the external connector that shut off the original chip.

Now that we have divulged some of the ‘advances’ developed by Intel, and the tactics used to defraud educate the buying public, let me ask again:

Dual-core from Intel?

Without the Opteron, I don’t think so.

Itanium? Nevertheless, in the 2K7 timeframe, we definitely would not have been running the Itanic Itanium processor. For the following reasons:

  1. Processor yields have been deplorable,
  2. Thermal signature of the chip has remained pretty high,
  3. Performance has remained very, very ho-hum, never exceeding that of the Pentium 5, not to talk about the Opterons. Oops, forgot Intel canceled the P5!
  4. Development of software targeted at Itanic has remained pretty much in the mainframe/alternative OS realm, and
  5. The benefits of Itanic, aka The WIIFM Factor, have never been successfully enumerated to both the public and developers.

64-bit Vista?

It would have been pie-in-the-sky without the Opteron, IMO.

What is very ironic is that just prior to the intro of the Opteron, Intel had been sabre-rattling about the IP contained in its 64-bit platform, not knowing that AMD was going to extend the much-maligned current computing architecture, the x86, and create the AMD 64-bit platform.

It was a brilliant move, and superbly executed, and a fitting ‘Exit Stage Left’ moment for Jerry Sanders.

I can safely say that the Opteron ushered in the modern 64-bit desktop and server processor era.

It revolutionized servers, created the market for x86-based blade servers, and redefined data center architecture, especially in terms of thermal output and cooling requirements.

My 2¢

© 2007, John Obeto II for®

No comments: