Sam - I'll try to answer a few of your questions.
1. "no rational consumer will delay buying a computer..."
For some, MMX will be a non-issue. Lower prices will be in effect with the non-MMX machines and many rational people will think that this a good choice for them.
2. "When do you expect MMX computers to be available ..."
Several suppliers - Compaq, Toshiba, NEC - will announce systems on January 8, 1997 as part of Intel's well orchestrated MMX introduction. 150 MHz & 166 MHz Pentium Notebook MMX devices will be a focal point. Bear in mind, the enlarged L1 cache (32 KiloBytes) in the Pentium MMX chip will provide a performance boost on existing (non-MMX) applications. Notebooks that perform "nearly" as fast as desktop machines are always in high demand.
3. "What pricing do you anticipate for the MMX chip?"
I expect the MMX chips will sell at the same price (or a very slight premium) as the same-speed non-MMX chips sell for today. These non-MMX chips should be re-priced at lower points to make them attractive and help blow out the inventory in the channel pipelines.
By not charging a real premium for MMX, Intel will help induce the early adopters to go for the MMX, if this is what they really want.
4. "...does this imply that a businessman who is contemplating purchasing a notebook computer primarily for making sales presentations ..."
My suggestion is "Try Before You Buy". Bring your software demos to the dealer from whom you will be buying a notebook computer and run speed/compatibility tests on both MMX and non-MMX machines.
As a personal note - if you are displaying computer graphics on a notebook computer to a potential customer - CONCENTRATE ON THE DISPLAY - this is what a customer sees. THIN FILM/Active Matrix displays are night & day better than the passive display DSTN (?) and don't suffer from severe angular fade out, vanishing cursors, and dim/washed out contrast. This may be a more overriding concern than the underlying hardware unless you are demoing CPU intensive software such as CAD/CAE applications. In that case, BOTH speed AND the display are critical.
The higher price of the AM thin-film display will pay for itself!
5. "Dell Direct tells me it is not expecting MMX to be available..."
From what I have read, Dell is not going to offer notebook MMX machines for several months until the MMX pipeline gets filled up. I recollect that they will offer desktop MMX systems shortly after January 8, 1997.
6. "...liquidate Pentium inventories rather than any bugginess of MMX1, as has been speculated..."
Intel has to properly handle the conversion from non-MMX to MMX from a production standpoint as well as a credibility standpoint.
The effectiveness of the MMX instructions will be a hot topic for debate in the media starting at 9:00 AM on January 8, 1997 (after the reporters have consumed their free doughnuts and coffee at the rollout). Benchmarks will be flying fast & furious, as well as bug reports due to "alleged" incompatabilities of MMX hardware with different software.
How all this sorts out will be hard to predict. Intel will address the issues in a professional manner, having improved on this sort of PR behavior after the Pentium FDIV debacle of two years ago. I hope they have an MMX Swat Team in place to handle the issues as fast as they come up.
I believe the real focus, however, will be with the ISV's - independent software vendors. It is the applications that sell hardware and the demonstration of a compelling software product, making use of the MMX instructions, that will draw people/customers to this technology.
From my own standpoint, I use a single add-on card with a DSP chip that performs many functions - CD-ROM control, 28.8K Modem, MWAVE recording/playback, MIDI recording/playback, and Voice-mail. The MMX, with the help of an appropriate Analog-Digital/Digital-Analog converter, could easily replace this card.
If such an implementation is offered in a notebook form factor, with minimal cost increase, the utility of this machine could be greatly enhanced without the worry of plug-in card compatibility. This exploitation, in my opinion, will take months as PC and software manufacturers ramp up the MMX learning curve , learning what is possible and doable with this new chip.
Paul |