Replies to post #20367 on Advanced Micro Devices Inc (AMD)

[drjohn]

Granted, the 1.8G Dothan will have a 2M L2, but 6% clock increase in 10 months does not look promising.


As has already been pointed out, when you have no competition and are making loads of money whats wrong with making more loads of money.

[Mysef]

Petz,
re:Granted, the 1.8G Dothan will have a 2M L2, but 6% clock increase in 10 months does not look promising.


I'm convinced Dothan will lose every 64 bit benchmark against the Athlon64, but remember Intel has been devoting resources to a 64 bit processor that isn't named Itanium. Keep your eyes on the ball.
:)

Mysef

[subzero]

"The Pentium M roadmap here http://www.xbitlabs.com/news/mobile/display/20031211124114.html looks disastrous."

In a way, you are correct.

The Pentium M roadmap is disastrous for AMD - becasue Intel's Pentium M has zippo competition from AMD.
Instead of going after the lucrative high ASP 2 million mobile CPUs per month business, AMD is chasing a bottle rocket for the 10,000 unit/month Opteron business - which is chewing up all of AMD's fab capacity with their huge 193 sq. mm. chip.
Remember AMD's inverted pyramid slide that Dirk laid on the analysts?

[wbmw]

Petz, Re: 1.8 GHz Dothan won't be available until February 15, 2004 and 2 GHz won't be available until sometime after April and before Q4 (unspecified).

There's also a 2.13GHz Dothan before Q4, and it looks like it uses a 533MHz FSB. Should give Prescott and Athlon 64 a run for the money in the mobile space. That doesn't sound too disastrous to me.

Re: Granted, the 1.8G Dothan will have a 2M L2.

It will also have additional IPC improvements. That with the added cache should add another 5-10% in performance. The 533MHz FSB should also add another 5-10%, and the Alviso chipset probably has a faster memory interface. Add 5% for that. Lastly, the frequency boost from 1.7GHz to 2.13GHz is roughly 25%. By the end of next year, Dothan will have 40-50% higher performance than Banias today. That's more or less capable of keeping up with AMD's most optimistic roadmaps in 2004.

[chipguy]

Granted, the 1.8G Dothan will have a 2M L2, but 6% clock increase in 10 months does not look promising.

Clue 1: the comments of the Banias design team in ITJ that
they didn't have time to implement all their new ideas on
how to improve both power efficiency and performance.

Clue 2: the unexpectedly large (compared to Banias's 83
mm2 that is) die size for Dothan given Intel's stated SRAM
cell size in 90 nm.

It may be quite foolish for either AMD or Transmeta to
assume that Dothan is just a 2 MB Banias that can clock
a little faster. ;-)