ESTIMATION: The Good Chip Intel is in a critical phase right now. Captain Pat jumped onto the bridge after the ship had begun to float across the Straits of Lateness toward the Rocks of Irrelevance, grabbed the helm, and yelled, “Linux Muckabout Dangerous Ploy Dangerous Time Full steam ahead!”
Although its initial berth at Alder Lake is generally regarded as a return to competitive form, that design was completed well before Gelsinger’s return, and the upcoming missions will still leave no opportunity for navigational errors.
At least one of the course adjustments appears to be a little sour. Intel has always understood how crucial it is to promote open source in order to maintain its processors compatible with Linux.
This necessitates being a little more transparent about the specifics of how its chips operate than in the golden days of Wintel domination, though. The Linux kernel cannot be used to sign an NDA.
Chipmakers are Renowned for their Paranoia
Silicon Valley was founded on mystery and mistrust. Despite the fact that Andy Grove, the legendary CEO of Intel, made paranoia the company’s motto, Intel eventually relaxed.
If they had the option, Qualcomm and Apple would throw you into their piranha pools just for asking questions, but Intel has figured out how to give as well as take. But it can be returning to old unhealthy behaviors.
A recent innovation known as Software Defined Silicon (SDSi), about which Intel has nothing to say, is one of the new things that cannot be discussed. Which is strange given it recently sent the Linux kernel supporting code for it.
The code itself makes no mention of SDSi; rather, it adds a means for controlling whatever it is using a trusted secure token. When the appropriate license is applied, hardware features are essentially unlocked.
It is nothing new. Electronic test equipment frequently has higher performance or extra features, but they are usually deactivated on the base models, with the option for the player to pay for them later. But what can it possibly mean in terms of SDSi and Intel architecture?
It is more cost-effective to manufacture one product that performs everything and charge for unlocking it because having many physical variants of anything is expensive for OEMs and Intel alike. It is a modified version of a method learned by naive schoolchildren in the late 1970s when less-priced Casio scientific calculators shared the exact same hardware as the more expensive model.
Linux Muckabout Dangerous Ploy Dangerous Time Updates
Simply put, Casio didn’t print all the functionalities on the basic kit keyboards. Cores and cache on future Intel processors will likely be disabled until “magic numbers” are reached, and with the SoC future at hand, this might include a variety of IO, acceleration, and co-processing features. In fact, it might already be there.
This is fantastic from the engineering, marketing, and revenue sides. In order to achieve the design, performance, and fab economies that Apple enjoys while also making sense for many OEMs, Intel may create an M1-like SoC that can be instantly adjusted for various platforms. Software upgrades or even subscription-based business models might provide additional money.
Which is suddenly less appealing, DRM on the hardware’s core?
The main issues with this strategy are psychological, at least in the context of consumers. Those who purchase the locked-down models are aware that they have paid for hardware with considerably more potential.
This is grating. Those that purchase the most expensive, top-of-the-line model are aware that they have spent significantly more for what the cheapskates possess. That irritates me as well, and unsurprisingly, people work very hard to figure out methods to game the system.
Much worse in this instance is Intel’s illogical presentation of the enabling code to the Linux kernel. At the very least, it is condescending; at the very worst, it is a declaration of purpose that Linux in particular and open source in general are there to be subverted if it pleases Lord Intel.
This has the appearance of treating open source like a corporate marketing tool. It is beyond regrettable to disguise corporate functionality and add it to the crown jewels of free source.
It’s reckless as well as disrespectful. Security nightmares result from submitting unidentified kernel changes. When there is no hardware accessible and the maintainers don’t know how something works, how can they test it? To do this for a feature that quickly jumps out as a top target for attention is really insane.
Linux Muckabout Dangerous Ploy Dangerous Time Yet again, Intel has failed to understand the psychology of actual users. The hacking groups who have previously attacked and broken through copy protection, DRM, and locks on gaming platforms are heralded as heroes.
To overthrow the Mighty Corporation and grant the modest user the ability to use the object they bought for whatever they choose is a very high-status job. Harsh on the producer, but the reality is not to blame if your business strategy cannot survive it.
All of this is conjecture, but it is good conjecture. Why are you being so indirect? It would have been preferable if Intel had offered this at the same time that it could have provided target hardware to developers, with good use cases, testable systems, and a convincing rationale for why this is advantageous.
The last threat is this one. A lot needs to go right for Intel right now. It requires quality goods. A good roadmap is required. It mostly needs trustworthy friends who are at the core of the business. It no longer needs to be known as a business that plays tricks on its clients and leverages its size to advance its own interests.
Making unreliable friends by tampering with the Linux kernel is impossible, but it is a terrific method to reclaim power over storms that are now much bigger than you. It might just be a squall that Captain Pat can skillfully escape by tapping the wheel.
