Intel patents the AMD Zen architecture

[qualverse]

Summary

Intel has been granted a patent for their 'Ocean Cove' processor architecture that appears to be a direct copy of the AMD Zen 1 architecture.

 

Quotes

Quote

Since the beginning of 2018, I had been following the work of the Hillsboro team, looking forward to have access to the first patent of the disruptive new architecture that was being developed... And in 2019, this patent was finally published.

 

Upon analyzing it, I was absolutely shocked by what I saw and was certain that this patent would never be granted in the state in which it was presented. However, yesterday I received notification that this patent had been granted.

 

My thoughts

The evidence appears pretty damning for this. I'm really not sure how or why something like it could happen. Is Intel actually planning on making a chip with the Zen 1 architecture? I really doubt it, so this is probably some sort of strange legal move. Perhaps AMD failed to obtain protection over Zen 1's design themselves?

 

Sources

https://twitter.com/Underfox3/status/1511697371020902401

https://www.freepatentsonline.com/11294809.pdf

[Fasterthannothing]

Ok AMD is screwed dang how could they be so stupid as to forget to patent their technology?? 

[Levent]

Twitter thread is giving me the same vibe BUT knowing how anal patent institutes are, I bet this is not what it seems.

[Jeppes]

22 minutes ago, SlidewaysZ said:

Ok AMD is screwed dang how could they be so stupid as to forget to patent their technology?? 

You cant patent something that has been used in the market for years. And even if you can by some miracle find someone incompetent enough, good luck enforcing the patent.

[Fasterthannothing]

28 minutes ago, Jeppes said:

You cant patent something that has been used in the market for years. And even if you can by some miracle find someone incompetent enough, good luck enforcing the patent.

I don't understand your comment. Intel literally just patented it. And they have enough money they can keep throwing lawsuits at the wall till something sticks if they wanted.

[tkitch]

4 minutes ago, SlidewaysZ said:

I don't understand your comment. Intel literally just patented it. And they have enough money they can keep throwing lawsuits at the wall till something sticks if they wanted.

What gets done and what is actually following the rules are not always the same thing.

A product from a competitor that's been on the market for years cannot be patented by another company, and be enforced.  As soon as AMD Takes it to court with evidence showing "We developed this tech in-house and have been using this for the last 6 years" The judge is gonna look at the patent and light it on fire.

 

[porina]

1 hour ago, Levent said:

Twitter thread is giving me the same vibe BUT knowing how anal patent institutes are, I bet this is not what it seems.

Based on the diagram comparisons shown, it does look like a copy of the Zen diagrams. The rest of the body text is beyond my comprehension and I'll leave the analysis/comparisons to those better placed to do so. 

 

The Zen diagrams are dated August 2016. The patent listed was filed August 2018, apparently as a continuation of an earlier patent filed December 2016. So it is very possible Intel copied AMD's homework. I dislike that the Tweeter is using marketing tactics in presenting it as a granted 2022 patent, which while technically not wrong, is giving the impression this is a current action as opposed to something started years ago. Patent process isn't exactly fast.

 

41 minutes ago, Jeppes said:

You cant patent something that has been used in the market for years. And even if you can by some miracle find someone incompetent enough, good luck enforcing the patent.

Not an expert on patent law, but I have gone through the process before with US office. To my understanding, the patent office looks through existing patents as basis of whether to grant a new application or not. Granting it just means, within their remit, they don't have a reason to reject it. Information that is disclosed before filing may be used to invalidate a filing. Ryzen wasn't launched until after the original patent filing, but I don't know how continuations work here.

 

Edit: the older patent seems to be for a specific single feature. The newer patent sounds like it is for an architecture. Again, I don't know enough about how to interpret this further. It really would take deeper analysis to see what exactly is claimed and if that is Zen or something else that looks similar but otherwise different.

[leadeater]

1 hour ago, SlidewaysZ said:

Ok AMD is screwed dang how could they be so stupid as to forget to patent their technology?? 

They did

[porina]

 

This is kinda making a bit of sense now. The older patent was for a cache function. The just granted patent which is a continuation is for a few features. It does not sound like Intel is claiming AMD's whole Zen. Apparently that might be copypasta used for context, just a bit of a gaff whoever wrote it used AMD material as a reference. So lower your pitchforks, Intel is not claiming Zen.

[leadeater]

Nothing to see here

 

 

Edit:

@porina2 minutes faster than me, booooo lol

[LAwLz]

ITT: Some Twitter user tries to drum up controversy by making Intel look evil and AMD look like an underdog. Then he begs for money.

Meanwhile, the subject is too difficult for most people to understand so they just blindly believe the tweet and gets mad. Because who is on Twitter besides people who are looking for things to be mad at?

[leadeater]

2 minutes ago, LAwLz said:

ITT:

Phew glad it wasn't LTT this time 

[leadeater]

Also if I had to guess this is to address Spectre/Side Channel vulnerabilities. Edit: Oh haha Ian literally said that, oh well didn't read his post that far LOL

[porina]

5 minutes ago, leadeater said:

@porina2 minutes faster than me, booooo lol

You found a nicer presented version of it at least  

 

3 minutes ago, LAwLz said:

ITT: Some Twitter user tries to drum up controversy by making Intel look evil and AMD look like an underdog. Then he begs for money.

Meanwhile, the subject is too difficult for most people to understand so they just blindly believe the tweet and gets mad. Because who is on Twitter besides people who are looking for things to be mad at?

I wonder how many "news" sites will repeat the claim without checking, simply to get the clicks?

 

I don't think it is uniquely limited to Twitter, this can happen anywhere discussion takes place openly. 

[AnonymousGuy]

122 pages...holy balls that must be a very specific feature being patented.

 

I skimmed the beginning and end of the patent document where it sort of summarizes things anyways: the whole thing seems to be describing a very specific scenario of one core zero-ing out the cache lines of other cores through some sort of coherency mechanism over an interconnect.  I don't really understand the significance of this.

[Fasterthannothing]

Ok after reading through this it makes more sense. But 120+ pages good grief that's insane. No wonder bad conclusions were drawn myself included. You almost need to be a engineering patent lawyer  to understand the full thing.

[porina]

8 minutes ago, SlidewaysZ said:

You almost need to be a engineering patent lawyer  to understand the full thing.

Probably for historic reasons, the language in patents is in a world of its own. See it as more of a legal language than a technical one, even if it is trying to describe technical stuff.

 

I contributed to a patent in past employment. What ended up in the patent was a struggle for me to comprehend. We had someone to translate it to patent speak but we still had to sign off it represented what we had. Those meetings were painful.

[LAwLz]

34 minutes ago, SlidewaysZ said:

Ok after reading through this it makes more sense. But 120+ pages good grief that's insane. No wonder bad conclusions were drawn myself included. You almost need to be a engineering patent lawyer  to understand the full thing.

Let this be a lesson that things are not as simple as they might seem, and a lot of people online want to drum up controversies to influence your opinions.

[Zodiark1593]

4 hours ago, porina said:

. So lower your pitchforks, Intel is not claiming Zen.

Dammit, I finally got my delayed shipment of Pitchforks to sell. 

[porina]

3 minutes ago, LAwLz said:

Let this be a lesson that things are not as simple as they might seem, and a lot of people online want to drum up controversies to influence your opinions.

There's also the other possibility covered by the saying: "never attribute to malice that which is adequately explained by stupidity" 

 

8 minutes ago, Zodiark1593 said:

Dammit, I finally got my delayed shipment of Pitchforks to sell. 

I'm sure there will be plenty of opportunities to use them, just not in this instance.

[Vishera]

5 hours ago, leadeater said:

Nothing to see here

 

 

Edit:

@porina2 minutes faster than me, booooo lol

I knew that there is something off with OPs post.

It just doesn't make any sense.

 

 

[TetraSky]

Now this is just ridiculous. How was this even granted. And how come AMD didn't patent their shit!?
 

[poochyena]

7 hours ago, LAwLz said:

ITT: Some Twitter user tries to drum up controversy by making Intel look evil and AMD look like an underdog.

uhh, and LTT user

[RejZoR]

No wonder patents are such nightmare to deal with if a minor architectural detail requires 122 frigging pages of documentation...

[Mark Kaine]

13 hours ago, porina said:

I wonder how many "news" sites will repeat the claim without checking,

ALL of them.