In a nutshell: When it comes to the FOSS / FOSH friendliness, should one prefer Intel over AMD, or vice versa? Which is the way to go considering the degrees of freedom in hardware?

Sorry for the inconvenience, I am still a novice on Linux and at the moment, buying my first pre-installed Linux laptop. I was told that the good people here at LinuxQuestions.org might help me with my quest.

In the beginning, I'd like to emphasize that I am not looking for "absolute freedom down to firmware level" nor "fully free hardware" (e.g. Purism, Raspberry Pi, Olimex, BeagleV, or any other single-board computer out there on the market). I mean, it's not that those things wouldn't be important. I just want to say that in my opinion, certain binary blobs can be acceptable. Anyway, in this thread I'd like to focus on the comparison between the two commercially predominant processor manufactures, AMD and Intel, and the small steps they have taken to approach the developer community and its aspirations of free hardware / firmware.

Now, I know that neither Intel nor AMD is a great option in terms of freedom per se. That's why I am emphasizing the course of progress and all the little steps that have taken place.

I've tried to make my research on the topic, and — to make a long story short — AMD seems to have been rather generous with releasing source code in the past, and it even used to be the choice of preference by Libreboot / Coreboot project.[1][2] In the same breath, Intel has been lashed out for allegedly releasing "distracting, irrelevant pieces" of source code, that have been merely to lead developers astray. There was also a pretty outspoken statement on LinuxInsider (24.09.2012), calling AMD "being a LOT more FOSS friendly" than Intel, and therefore to buy AMD chips "if you want to support FOSS, be it server, desktop, or laptop":[3]

That quote is quite old, though, and things might have changed ever since. For example, Intel has been said to be a major contributor to the open source technologies nowadays. In turn, it is said that Google is working towards to excise (to get rid of) the "binary blobs from the x86 part of Zen CPUs" in AMD.[4] Well, these are not — of course — a guarantee of anything, but at least small gestures in the direction of playing together with the FOSS / FOSH developer community.

I am also aware of the different brands that offer pre-installed Linux laptops (System76, Tuxedo, KDE Slimbook ... just to name a few). I am sure both AMD and Intel run just fine on those. Should there be some problems, though, I am ready to live with that. That's just not really my main interest at the point. What I am interested in, however, is the somewhat philosopical perspective on things, and what steps have those chip manufactures taken / what have they taken not to collaborate with the FOSS / FOSH community.

Thanks a lot in advance for your time, expertice and goodwill! Cheers!


Yours Sincerely,

Jayaguru-Shishya

Ps. If you think that there's not much different between the two chip makers, that's a perfectly fine answer as well! Sorry if I left an impression that I am forcing one to choose between the two!

I must confess to being a little less enthusiastic than you are for Open Source Hardware

What's important to a user is that hardware works. Hardware has firmware these days, because most chips have internal industrial cpus. So the source code in a low level language is no use to you, because unless you have the circuit and the cpu instruction set, you can't figure out what's going on.

In the old days, there used to be eproms for the PC BIOS, and in Industrial devices. One company I was with had to replace those and it was a huge cost. Hence the use of E²Proms which are electrically erasable. Any security hole can be plugged by a firmware update.

I'm biased against Intel, because when they could rip folks off, they did.

On OSS, I don't frankly see a lot to choose between Intel & AMD. Both ship graphics drivers with Mesa, although AMD has to watch over it's shoulder at potential patent lawsuits from Nvidia. Mesa tries to handle everything, so AMD does closed source drivers as well, probably needing to tweak libGL*.so for it's own ends like Nvidia does. What matters imho is not the aesthetics (pure FOSS or FOSH) but practicalities (e.g. does it work?).

Tails & Heads are a case in point. Tails ship firmware, and things work. Heads is like Tails, but doesn't ship any 'closed source' firmware, so nothing works. It's all well and good sticking to principles, but it's not good if you can't even get online to make your peripherals work, because there's no firmware.

Except for one thing, to me it's six of one, half dozen of the other. That thing is Intel's Israel component. When I buy new, I buy Intel. I don't buy new often, so I'm using both. My two most recent are Intel, 3 years old. Two of the next four are AMD, 6-8 years old. Older are all Intel back to around 2002. Older probably shouldn't be counted, the tail of the 32bit era. I'm typing this on 7 year old Intel.

Both Intel and AMD understand the importance of open-source software. As hardware manufacturers, these companies need to make all software run best on their hardware.

Other hardware manufacturers have not figured this out. I am looking at you, Nvidia and Qualcomm.

PowerVR was an anomaly on Intel SoCs because it was licensed from Imagination Technologies.
Ed

Hardware guys know that if their circuitry gets out, their product is commercially useless to them. They are inclined to be neurotic about any details, software included. That's why FOSH is a joke. It's not unusual, for instance, to find multilayered circuit boards, with few tracks on the top or bottom layer. Everything is on the hidden layers, one of which is an earth layer. So even if you shine a bright light through the circuit board, you can see nothing.

In my long time in Hardware, only a handful of industrial products released schematics. In every case, the design was so awful, that they hoped guys like me would fix some, and lessen their losses.

If a design gets stolen, it will appear as a product from the Far East next month.

As far as the parts they BOTH make, it's a wash. The processors, gpu's, and chipsets are both as open source as they can get without having fully open hardware (ie - fully open source drivers while still supporting proprietary hardware). Intels wireless is more open (currently) than the "AMD" wireless (rebadged MediaTek) is. HOwever, the "AMD" wireless is getting open source linux drivers, which will bring it up to on par with Intels.

My last 2 laptops are AMD because currently they offer better performance/$$$ (8-core/16-thread 15-watt Ryzen Pro & 6-core/12-thread 15-watt Ryzen Pro less than the competing Intel 4-core/8-thread 15-28 watt laptops can muster), the 2 before that were Intel (4-core/8-thread i5's/i7's easily outperforming AMD's 4-core/8-thread Ryzens).

There's no way you can tell about the hardware from the software. You might grasp things like 'it has over 4 cores' but that's it. You sure won't spot the value of a current limit, or any of the details thyat matter. I'd make 2 exceptions
1. Where someone has routines for handling some protocol or handshake in the driver.
2. Nvidia, who are streets ahead hardware design wise may expose pointers allowing some to infer

Another benefit of OSS is that there's always folks looking for bug bounties, and contributing patches.

I don’t know about CPUs. I know about GPUs and 3D graphics APIs.
There is a lot of effort that goes into writing drivers and I can understand that AMD and nVidia want to protect their drivers.

There was the case of the texture compression method. S3TX.
Someone has to pay S3 some royalties. MS has deep pockets so they payed them and integrated S3TX into Direct3D and they called it D3TX.
For Windows OpenGL drivers, the S3TX goes into the nVidia and AMD drivers so these 2 companies pay S3.
For Linux OpenGL drivers, I guess it is the same. These 2 companies pay S3.
But S3 is getting payed. Why not open source the driver?

As for that HDCP none sense. What’s the point? People have been pirating for years now.
For DVD encryption, again, what is the point? it was hacked from day 1 or 2.

Some things should be open sourced. Some things shouldn’t.
Some things should not be encrypted.

Very interesting post. My comments were solely related to deducing hardware details from software.