"Customers must be patient, as the 16nm chip is just nearing tapeout and should be in production in 2H17" (I would have liked to see shareholders in place of "customers".
The above "white paper" had pretty much all good news in regards to X-gene 3 against Cavium and Intel servers.
The last paragraph did include " But AppliedMicro must first demonstrate that its new chip can meet performance and power expectations, as these are the metrics that matter most to customers."
the above "white paper" added to AMCC website on 6/27/16
Last Edit: Jun 28, 2016 20:01:07 GMT -5 by christopher
I think I did post this somewhere, or maybe I didn't, anyway I can't seem to find where I did post it, if I did, so I'll just say "thanks for posting". It is an important enough document to warrant a second posting.
Don't forget that this evaluation does not include the Thunder-X 2. I'm not sure that comparing the hypothetical specs of processors that do not yet exist would be worthwhile, but it might be sufficient to know that the Thunder-X2 compares favorably with the X-Gene 3 in terms of specifications, and they are both going to be entering production around the same time. AMCC may want to think in terms of going directly to the X-Gene 3XL.
The wild card in all this is connectivity. It's what CAVM's scramble has been all about in acquiring QLogic: they need to add connectivity to their portfolio as a product line that is complementary to selling ARM processors. CAVM is paying to acquire this and AMCC already has it in house, in fact someone is paying AMCC to develop 100G PAM4. AMCC is already there.