You are currently viewing SemiWiki as a guest which gives you limited access to the site. To view blog comments and experience other SemiWiki features you must be a registered member. Registration is fast, simple, and absolutely free so please, join our community today!
I'm tentatively posting another SeekingAlpha blog after less than enthusiastic support for that source in a prior thread. I'm not qualified to comment on the business content but it does raise a good point. Where is AMD going? What's the breakout strategy? AMD seems to have been hovering in a twilight, me-too world forever. Getting by certainly but never looking like a compelling #1 any category. But they battle on. Is survival an OK strategy or is there a bigger end-game? (The blog points out that the stock is up 70% but without obvious reason. So yes they were a good buy, but could as easily go back down again it seems?)
I recall that in 2003 AMD introduced the Opteron and Athlon 64 process lines, the first x86 64-bit architecture for consumer PCs, beating Intel. That's about the only technology leadership that I remember for AMD.
I recall that in 2003 AMD introduced the Opteron and Athlon 64 process lines, the first x86 64-bit architecture for consumer PCs, beating Intel. That's about the only technology leadership that I remember for AMD.
integrating first northbridge and later southbridge with CPU leading to so-called SoC. I am not sure here about time-line compared to the ARM ecosystem.
intergrating GPU into SoC.
I do agree though that only in the Opteron time-frame AMD could out-design the technological advantage Intel always had. Intel could keep effect under control using also what has later been catalogued as illegal business tactics. Using these they could delay it long enough to catch up in design space so AMD could not overcome the technological advantage anymore. In that respect the firsts I give can just be seen as trying to make new markets because not really being to compete in the main market.
Personally I think the biggest thing keeping AMD going on the CPU side is that they are NOT Intel. In any other market AMD would be dead with all of the missteps. The good news is that Intel has left the process advantage door open with the delays in 14nm and 10nm so AMD still has opportunity if they work closely with the fabs and execute.
Now that's an interesting theory. The market keeps AMD alive to keep Intel (and nVidia) on their toes. Curious reason for existence, but understandable.
Personally I think the biggest thing keeping AMD going on the CPU side is that they are NOT Intel. In any other market AMD would be dead with all of the missteps. The good news is that Intel has left the process advantage door open with the delays in 14nm and 10nm so AMD still has opportunity if they work closely with the fabs and execute.
Are AMD on Global Foundries Silicon these days? Seems like there is an opportunity to go for better power performance on FDSOI than you'll get with FinFET.
Personally I think the biggest thing keeping AMD going on the CPU side is that they are NOT Intel. In any other market AMD would be dead with all of the missteps. The good news is that Intel has left the process advantage door open with the delays in 14nm and 10nm so AMD still has opportunity if they work closely with the fabs and execute.
"As for RTG’s FinFET manufacturing plans, the fact that RTG only mentions “FinFET” and not a specific FinFET process (e.g. TSMC 16nm) is intentional. The group has confirmed that they will be utilizing both traditional partner TSMC’s 16nm process and AMD fab spin-off (and Samsung licensee) GlobalFoundries’ 14nm process, making this the first time that AMD’s graphics group has used more than a single fab. To be clear here there’s no expectation that RTG will be dual-sourcing – having both fabs produce the same GPU – but rather the implication is that designs will be split between the two fabs."
Source: AMD Reveals Polaris GPU Architecture: 4th Gen GCN to Arrive In Mid-2016
For RTG to split production between both TSMC and GF/Samsung means double the NRE costs, so they better recover their costs with high unit volumes to justify that decision. This type of production split keeps the foundries hungry to win the business, even if it's only part of the business.
AMD has always understood the embedded market better than Intel. The problem has been consistent execution. They have a pattern of running really hard for a few years and briefly catching up or even pulling slightly ahead - anyone remember the K5? - and then falling back as Intel's relentless tick-tock overwhelms their advantage. Had Hector Ruiz executed on Barcelona (the multicore effort @Staf was referring to), AMD might have taken the lead on the server side for a while. As @DanielPayne points out, they were also briefly ahead in 64-bit client parts.
The interesting part about AMD is a hedge between the Intel and ARM community, again a good strategy, not yet stunningly executed.
It's been pretty quiet since their last management shuffle ...
There's also HSA - Heterogeneous System Architecture - which attempts to make programming such processors easier, but I suspect the jury is still out...
AMD has had some bad leaders after Jerry. The same thing is happening at Intel, post Groves. I predict the same thing will happen at TSMC, when Morris is gone.
The model that has shown staying power (rather than leader-dependence) is the conventional conglomerate model that Samsung executes. Like GE, they go into a market, gain the first or second position in market share, and leverage dominant market share to maintain acceptable (although not great) profitability.
"Zen was a clean sheet design that started few years ago. We are in the final figure of executing and the milestone that you want hear us talk about is Zen tapping out, which should be over the next several months. And then putting samples in the hands of our customer and then starting portfolio of revenue in 2017.
And by the way, because we have this reuse approach for cores, you will see us with Zen cores in the high-end desktop first and then the server from our overall products standpoint. But the key is taping out in the next several months, samples and customers for the validation of the product over the 2016 time frame and then the revenue ramp happening in 2016."
So, the 14LPP designs this past summer were some sort of AMD Services parts... either a new customer design services contract or (more
likely) a shrink of the 28nm PS/4 and Xbox One designs. AMD was supposed to deliver in 20nm as a cost-reduction, but had to cancel those projects when GF ceased 20nm planar development -- they took a massive write-off loss (a penalty to Microsoft and Sony?) when they announced they were cancelling their 20nm efforts.
So, we won't see any Zen core-based designs until 1H2017, assuming AMD and GF execute "flawlessly".
There's also HSA - Heterogeneous System Architecture - which attempts to make programming such processors easier, but I suspect the jury is still out...
With their unique APU IP mapped onto leading edge Si, AMD has the potential to become THE engine that drives VR, which IMHO is THE "next big thing," for the very same reason they have a 100% lock on the game-console market.
IMHO, VR will be to game-consoles as Smartphones were to plain cell-phones. I think there will be an analogous, unpredictable, "explosion of utility" that will commence when a "sufficient" VR platform emerges... Just like there was when a "sufficient" smartphone emerged. AMD has the potential to surf that wave just like the ARM ecosystem has surfed the smartphone wave. (Of course there's the question of AMD's ability to actually catch and capitalize on that wave before the larger ARM ecosystem itself does.)
Longer term, AMD also has the potential of, ironically, eliminating the central role of x86 in the "PC" ecosystem, via HSA. If it advances to the point where software performance is determined by how well applications are dynamically accelerated, as opposed to how strong your x86 implementation is, then who will really care about how strong your x86 implementation is?
I think AMD will not be going dual source (TSMC and Global Foundries). NRE's would be high, yes. I think the strategy is to have one part number on one source and another part number on the other source.
