AMD CPU Settlement

AMD CPU settlement has gained significant relevance in the market in recent years. So much so that it has gone from rumored bankruptcy.

In 2016 to be in the best market status in its history. In this article, we will briefly review the events of the last few years, focusing on the activity of your processor division.

AMD early in the decade

At the beginning of 2011, AMD CPU settlement presented the first time well-known concept of APU or Accelerated Processing Unit.

Which is the same as a processor with a graphics card integrated into the same chip, in the same way, that Intel had done for the first time a few months before.

These processors initially focused on laptops, marked what would later be an essential part of the AMD market. With the first launch of the Series A processors and then with the current Ryzen APUs.

The launch of the AMD FX:

  • The first architecture of the decade created by AMD was the FX. As you will see now, it was widely controversial. And as a result, it failed to get ahead of Intel in terms of performance or market share.
  • The first launch of this family was Bulldozer, in October 2011, with which the FX-8XXX with 8 “cores” (note the quotes) and 4 modules.
  • The FX-6XXX, with 6 “cores” and 3 modules, were presented. And the FX-4XXX with 4 “cores” and 2 modules.
  • Later, the second generation arrived with Piledriver, which belongs to the best-known processor in the range, the FX-8350.
  • This family received improvements in the architecture that allowed an increase in the clock frequencies.
  • And maintained fundamental characteristics compared to its predecessor, such as using a 32nm manufacturing node.
  • Several issues always questioned these processors. We will explain two of the most fundamental: the false cores and the low IPC.

AMD and the “fake kernels”:

  • It could be said that with the AMD FX, it was possible to bring the 8-core processors to the domestic segment. However, in reality, they are a somewhat different story and difficult to explain.
  • Basically, the “cores” are grouped in modules of two “cores” each. Without going deep into the matter, it can be explained basically as follows:
  • We find two cores for arithmetic operations with integers and only one FPU unit for processing with floating-point numbers within each module.
  • AMD considered that this was enough to talk about two cores per module, but as evidenced by the fact that they have lost a millionaire demand for this matter, it cannot be said that this is the case.
  • Suppose we combine this with the fact that there is a level 2—cache per module and not per core and the L1. I cache also shared and many other fundamental parts, such as the jump predictor.
  • We conclude that integer processing cores cannot function independently or the parts dedicated to stages from fetch and decode instructions.

The added problem of the CPI:

  • The quirky and inefficient design of this family of processors added several more handicaps that prevented them from living up to Intel’s offerings.
  • And it is that the IPC (instructions per cycle) of these CPUs was considerably lower than that of the competition.
  • Therefore the AMD FX processors needed a higher clock frequency to approach mono-core performance.
  • As we show you in the image above, an AMD FX-8350 with a 4.0GHz base frequency could not reach a 3.3GHz i3 3220 in its Cinebench performance.
  • The focus that AMD had to take on raising frequencies at all costs negatively influenced the design and power efficiency of these CPUs, yet they were hardly successful.
  • Proof of this endeavor was the 2014 launch of the AMD FX-9590, in a desperate attempt to compete with Intel.
  • This processor reached 5.0GHz frequency (turbo) with an absurd TDP of 220W and could barely beat a much more efficient Intel Core i5-4690K that also had a reasonable margin of overclocking.
  • How were the Financial problems and the salvation of Lisa Su and Ryzen?
  • In 2014, the financial situation of the company was getting worse. The media continually speculated that AMD might go bankrupt or be bought by companies like Samsung or NVIDIA.
  • The company kept falling into irrelevance, and a significant change was needed. Beginning in October 2014, Dr.

The announcement of the gestation of AMD Zen:

  • As we indicated before, the FX processors did not succeed due to their strange architecture. Which could barely compete with what Intel offered.
  • Therefore, it was urgent to present a competent alternative. Developing a brand new CPU from scratch is not an easy process and usually takes at least 5 years.
  • Well, with this objective, the AMD Zen microarchitecture began to be developed in 2012. And its existence was announced to the public in 2015.
  • One of the significant assets promised by the company in its day was a 40% increase in the IPC compared to the last iteration of the architecture inherited from FX processors, Excavator.
  • These new CPUs would come under a platform that would make use of the AM4 socket that we know today, and as expected, they would use DDR4 memory.
  • During this period, what AMD was doing was dripping information about Zen in the form of slides. There was a lot of joke about PowerPoint skills. That the company had in the absence of tangible results.
  • It was fabulous about everything. Would it be “smoke,” or would it be close to how well they were painting it. The moment of truth would come in 2017.

Also Read: Internet Explorer End of Life – Update Information, References, and More