It's simply a function of their history. They used to have professional level graphics cards ("Nvidia Quadro") using exactly the same chips as their consumer graphics cards.
The BIOS of the cards was different, enabling different features. So people wanting those features cheaply would buy the consumer graphics cards and flash the matching Quadro BIOS to them. Worked perfectly fine.
Nvidia naturally wasn't happy about those "lost sales", so began a game of whack-a-mole to stop BIOS flashing from working. They did stuff like adding resistors to the boards to tell the card whether it was a Geforce or Quadro card, and when that was promptly reverse engineered they started getting creative in other ways.
Meanwhile, they couldn't really Open Source their drivers because then people could see what the "Geforce vs Quadro" software checks were. That would open up software countermeasures being developed.
---
In the most recent few years the professional cards and gaming cards now use different chips. So the BIOS tricks are no longer relevant.
Which means Nvidia can "safely" Open Source their drivers now, and they've begun doing so.
I think companies should stop trying to segment customers. It's a long term loss, despite a short term gain.
[1] https://www.joelonsoftware.com/2004/12/15/camels-and-rubber-...
While I dislike market segmentation as much as anyone, it seems like it worked out ok for Nvidia.
It let them keep a (very) profitable segment of the business, which in turn financed other developments and let them become the GPU leader for a very long time.
No idea how it'll play out in the end of course, but we're definite past the whole "short term" time frame.
1) NVidia drivers had a lot of secret sauce to give high performance.
2) NVidia for machine learning still has a lot of platform lock-in (although fading gradually), and cross-compatibility doesn't help them
3) Quite often, if you've licensed something from a third-party, you can't legally open-source. Proprietary codebases sometimes get... messy.
I'm jumping ship as soon as Intel drivers are good enough. I don't trust AMD to have anything working -- too many bad experiences -- but Intel has a good track record. Arc A770 gives 16gb for <$300. That's as much as I have on my >$1000 NVidia card. I don't need maximum FLOPS. So long as deep learning models run, and 3d apps are accelerated, I'm happy.
I'm remembering this, though it's possible some things might have escaped the rename:
Violating patents is one thing, as you're only violating the concept/idea, but the implementation is still up to you meaning it will still be clean room design, whereas this guy also blatantly copied the source code and design files which is a slam dunk lawsuit, hence why no company ever wants to have competitors' IP on their systems.
The concept/idea is not what is patented. The patent is (or should be) for the specific execution of the idea. Competitors are free to implement their feature using methods other that what is covered by the patent, even if the end result gives the exact same functionality.
Have you ever seen patents? They rarely cover the implementation details, or at most they're intentionally super vague about that, most of the time it's just the general idea on how the widget would work and what it does, but not how to implement it technically.
https://www.theverge.com/2017/11/6/16614038/apple-samsung-sl...
It expired in 2017 but for the period it was in force, Amazon collected millions in licensing fees.
if you can’t figure out how its implemented you’re looking at an invalid patent or an application
> they're intentionally super vague about that
yes that’s how claims work
> most of the time it's just the general idea on how the widget would work and what it does, but not how to implement it technically
you’re looking at an invalid patent or an application
The implementation or execution of the idea usually takes the form of some Verilog or some C++. That is covered by copyright.
The patent is for the idea. Which is part of why I'm so opposed to patents, not just in software. In other fields, like medicine, patents are perhaps for discoveries, which are IMHO similarly valuable as the execution. But ideas aren't that valuable, or shouldn't be.
IP lawyer here (EDIT: not yours, of course): That's a considerable (and potentially-dangerous) oversimplification. What matters is whether what you do comes within the claims of the patent.
(For a more-detailed explanation, written in pseudocode-like terms, see a 2010 post I did: https://www.oncontracts.com/how-patent-claims-work-a-variety....)
Tangentially, it gets difficult in software because a lot of patents are .... maybe overbroad in their wording of claims. Lot of ambiguous looking landmines.
This is somewhat similar to business method patents (which were curtailed a little by the SC a decade ago, but were already known to be kinda sketchy for decade+ before that). Can't patent a pure algorithm, for example.
Doesn't that mean that in general it is also a really bad idea to ask an engineer questions about a particular piece of tech that they patented at a previous employer, even though the specific information is a matter of public record by virtue of being explained in the patent?
"Willful blindness" can be a danger (according to the Supreme Court, albeit in a different context).
Possibly a bigger danger: Your product gets kicked out of the market by an injunction (a court order to stop making, using, selling, etc.)
I did similar for medical self-symptom before ACA prevented "pre-existing condition" scam.
I feel like the granularity of patents is defined more so by where the frontier of knowledge is for a given domain than the patent office (i.e. what is hard but also valuable). But, I also haven’t spent a lot of time with patents
Unless you already want to count intel, not sure if they count as actual products or are still in early access.