This site may earn chapter commissions from the links on this folio. Terms of employ.

Llano-Die

Three and a half years ago, a group of investors filed a lawsuit against AMD. The grouping argued that the company'south decision to write downwardly the value of its inventory of Llano — the first Fusion APU — by $100 million, after pumping it in bullish quarterly meetings and annotator days, constituted an endeavor to defraud investors by keeping disquisitional information about the production from the investor community. Yesterday, AMD agreed to pay $29.v 1000000 to make the lawsuit get away for good.

A fleck of historical context is in lodge. When AMD bought ATI, information technology sold the buy to investors and consumers as a functional necessity given long-term trends in the CPU market place, and work on the Fusion concept began immediately in mid-2006. A shipping product, however, proved elusive. AMD repeatedly hit 'reset' on the concept and changed their messaging on which architecture- — Bulldozer or Phenom Two — would bulldoze the CPU side of the last pattern. Past mid-2010, AMD's then-CEO, Dirk Meyer, had made the decision to pull Bobcat into market earlier and to filibuster the introduction of Llano into 2022.

AMD's transcripts from this time flow reveal that the company was fairly bullish on Llano once it began to hit the market, just to reverse class and take a abrupt writedown of $100M in Q3 2022 equally a result. The company had admitted that its problems were partly due to a strategic misalignment between Llano inventory and channel demand, and that this resulted in a sharp reduction in earnings. The bad news was piling upwardly for AMD at the same time; Piledriver clearly wasn't going to magically fix Bulldozer'southward issues (though it did improve the situation modestly), and its Trinity APUs didn't move the power consumption or raw functioning metrics nearly as much as AMD might've liked.

Did AMD defraud investors? That'due south a lot harder to prove. The PC market place began to pass up in 2022 and the decline became precipitous in 2022:

pc-sales-in-2016-were-the-lowest-theyve-been-in-a-decade

PC sales began declining in 2022 and cratered in 2022.

Granted, AMD was seeing the touch on of this strategic misalignment on its own sales before CY 2022 had begun, but manufacturers tend to identify orders for new hardware based on how the old equipment is selling. Regardless of whether AMD'south public statements rose to the level of beingness legally actionable, the visitor was conspicuously tired of the case following it around. And it's definitely true that Llano'due south introduction was delayed by poor yields. The trouble was almost certainly on the GPU side of the equation — GlobalFoundries had plenty of experience building CPUs, but AMD had never before attempted to put a GPU and CPU on the same slice of silicon. When Intel rolled out its beginning CPUs with GPUs on-packet, information technology kept the chips split for a generation before integrating them into the same die with Sandy Span. AMD skipped this footstep, hoping to save time, but information technology cost them in the long run.

Llano ultimately faded out of the market, replaced by other APUs, and the entire 'Fusion' concept and the HSA capabilities AMD once trumpeted have largely vanished too. Information technology'southward not clear how tightly AMD volition tie Zen to this previous exercise. 1 of the paradoxes of the HSA/Fusion situation was that accelerating workloads by using the onboard GPU was disquisitional to improving the performance of AMD's older hardware — but now that Zen finally offers a solid CPU core to friction match with AMD's on-lath GPU, the demand for that dispatch is significantly reduced.

Let the states know: How likely are you to recommend ExtremeTech?