[Beowulf] Register article on Epyc (Brian Dobbins)

[mathog]

On 22-Jun-2017 15:05, Greg Lindahl wrote:
> On Thu, Jun 22, 2017 at 12:27:30PM -0700, mathog wrote:
> 
>> Recall that when the Opterons first came out the major manufacturers
>> did not ship any systems with it for what, a year, maybe longer?  I
>> vaguely recall SuperMicro going in quickly and Dell, HP, and IBM
>> whistling in a corner.  Something about contractual obligations to
>> Intel, or a desire not to piss off Intel.
> 
> Uh, it's more complicated than that. There's a ton of proprietary info
> related to shipping a motherboard and BIOS with a given chip, and iirc
> Intel insisted that anyone who worked with their proprietary info not
> have access to any AMD proprietary info. Once you add that to the fact
> that many of the big companies didn't want to bother with high-end AMD
> servers until they were a proven success, and you get a year-long
> delay.
> 
> I don't think it hurt AMD that much in the end.

I disagree.

AMD had a product that Intel could not match at the time but was unable 
to capitalize on that advantage due to non-market forces.  (Ie, the game 
was rigged.) That effectively put a lid on AMD's ability to grab market 
share and gave Intel enough time to develop and produce their own 64 bit 
x64 options.  The loss of potential market share had follow on effects 
for AMD, like not gaining economies of scale, which would have helped it 
compete with Intel going forward.

Prior to the Opteron Intel's plans rested squarely on the Itanium, and 
we all remember the huge sales predictions and the tiny number of 
systems which were actually sold.  The Itanium sales forecasts went into 
a nosedive, never to recover, as soon as the Opteron came out, when it 
became clear to most what the future was going to hold.  In the two 
years during which Itanium didn't have Opteron competition (2001, 2002) 
it didn't make much of a dent, even with all the hype and even more of 
Intel's patented non-market based competition, in this case having 
convinced Digital to axe the Alpha, SGI to drop MIPS, and HP to give up 
PA-RISC.  IBM and Sun wisely avoided drinking this Kool-Aid.

Regards,

David Mathog
mathog at caltech.edu
Manager, Sequence Analysis Facility, Biology Division, Caltech