News AMD Settles FX Bulldozer False Advertising Lawsuit for Roughly $35 a Chip

TheSecondPower

Honorable
Nov 6, 2013
22
9
10,515
This seems pretty rediculous. There are way bigger exaggerations in other product markets. And if people get at most $35 per chip, they gain almost nothing; just a waste of their time. The consumers lose, AMD loses, everyone loses. Except the lawyers apparently.
 

GetSmart

Commendable
Jun 17, 2019
173
44
1,610
Nopesies, the consumers win not from the financial standpoint but from the use of the term core which cannot be simply applied to CMT (clustered multi-threading) clusters. Thus no more misleading advertizing or marketing of multi-threaded CPU core designs.
 
Nopesies, the consumers win not from the financial standpoint but from the use of the term core which cannot be simply applied to CMT (clustered multi-threading) clusters. Thus no more misleading advertizing or marketing of multi-threaded CPU core designs.

Pretty much this. AMDs Bulldozer uArch was just CMT, which is a much more resource inclusive version of SMT. Both ideas came from the same place technically, DEC. Intel went the SMT route but marketed their chips accordingly as cores and threads. I think if AMD did the same, stating that it was 4 modules and 8 threads instead of 8 cores they would have avoided this altogether.

I am not surprised that they settled. This happens most often as it is probably cheaper to settle and pay out than to drag it out in years of court to potentially lose and have to pay even more.

I doubt we will see more CMT designs in mainstream with more and more cores coming to mainstream though.
 

GetSmart

Commendable
Jun 17, 2019
173
44
1,610
Pretty much this. AMDs Bulldozer uArch was just CMT, which is a much more resource inclusive version of SMT. Both ideas came from the same place technically, DEC. Intel went the SMT route but marketed their chips accordingly as cores and threads. I think if AMD did the same, stating that it was 4 modules and 8 threads instead of 8 cores they would have avoided this altogether.
Well, IBM's Power also marketed their SMT as it is (like Intel). That "module" definition was AMD's own marketing term as well. In reality the "module" is the actual core, while the "cores" are actually clusters.
 
Well, IBM's Power also marketed their SMT as it is (like Intel). That "module" definition was AMD's own marketing term as well. In reality the "module" is the actual core, while the "cores" are actually clusters.

Agreed but even IBMs Power SMT is more like Intels than like AMDs.

Their marketing team screwed them really.

Im sure almost nobody will pay attention to the part where you had to buy it from AMDs website (or california?) to quality for the payment.

I wonder if that will stick. How many people ordered it directly from AMD? I doubt the vast majority did.

But this does open them up for more lawsuits for other states. I doubt they will nor do I think they should. I really think this one is stupid TBH but its how the world works these days. Obviously they guy had major losses from this. Yup. Major losses from a CPU that didn't quite perform as well as modern 8 cores.
 
  • Like
Reactions: bit_user
So how about Ryzen "cores" that don't work without the I/O die?

Not really the same as none of the parts taken from the CPU and put onto the I/O die are part of the core itself. They took basically the USB, PCIe and IMC and put it onto the I/O die and connected the CPU cores to it. Think of it as the old school Northbridge but vastly faster, fast enough that you don't lose performance.

Before Intels Nahelm or AMDs K8 uArchs the CPU cores were just cores and were connected to the Northbridge which had the memory controller and some I/O. This is essentially what AMD did to offload some power and heat from the cores themselves but it is not comparable.

Bulldozer had 2 threads per module but unlike say an Athlon X2 or Intel Core 2 Duo it shared certain resources that the latter had for each core individually.

Basically all this covers is the Bulldozer and Piledriver uArch.
 
Feb 14, 2019
40
12
35
As usual it's the lawyers in class action lawsuits that get anything out of them. It's advantageous for them to agree to settlements, they are guarenteed a large sum of money. The people not so much. It's a rare case where they do.
 
  • Like
Reactions: MasterMadBones

waltc3

Reputable
Aug 4, 2019
420
223
5,060
Litigation lawyers--what a racket....;) They probably got half of it. Everyone knew what Bulldozer was--AMD copiously explained it at the time. Reminds me of the ill-fated hard-drive "What's a megabyte?" lawsuit. People complain of taxes often without realizing how much litigation lawyers add to the price of the things they like to buy in hidden costs, etc.
 

parkerthon

Distinguished
Jan 3, 2011
60
54
18,560
I agree with OP. This is mild stuff at best. Consumers only win if they started applying these kind of common agreed upon industry language standards to about a thousand other consumer electronic products out there. Thinking TV's, Wifi, and smart phones for starters. But they won't. Review sites are very important for a reason. Who buys something and throws it in to a build without first seeing how it performs per professional reviewers? I think unless there's some deep flaw that only emerges with time(thinking more failure and defects), I don't see a class action being that big of a benefit to any consumer that actually care to do a little research.

Lawyers def made bank here. Not much else.
 
Last edited:
  • Like
Reactions: Soaptrail
Personally, I disagreed with the lawsuit from the start. There are 8 CPU cores, but 4 FPU cores, in 4 Modules. And as far as I know there was not a standard definition of what constitutes a 'core' at the time. So I saw no intent to mislead consumers.

And even though I have purchased 6 AMD CPUs for myself, friends and family, that would fall under this settlement...I will not be filing a claim.
 

GetSmart

Commendable
Jun 17, 2019
173
44
1,610
Personally, I disagreed with the lawsuit from the start. There are 8 CPU cores, but 4 FPU cores, in 4 Modules. And as far as I know there was not a standard definition of what constitutes a 'core' at the time. So I saw no intent to mislead consumers.
The "modules" was AMD's marketing invention as with the term "integer cores". Technically the "module" is actually the core, while the "integer cores" are actually (integer execution) clusters. That is how CMT or clustered multi-threading topology works. Heck that "integer core" was written as cluster in one of AMD's early patents. Thus it is actually misleading, hence AMD's decision to try to settle this as soon as possible.
 
What makes you think I need an explanation? I already understand their design, in detail. I simply don't agree with your interpretation.

I see people applying the current understanding of CPU cores retroactively to an old, innovative design.

You need to be able to point to some industry standard that was established by a vendor neutral group of experts from that time period, and show how AMD knowingly lied as to how their product did or did not meet that standard, and then you can say they were misleading. Provide that source information and I will agree with you. Othewise, this is nothing more than opinion and I do not agree.

Settlements like this are rarely about actual wrongdoing and more about preserving a positive brand image.
 

GetSmart

Commendable
Jun 17, 2019
173
44
1,610
What makes you think I need an explanation? I already understand their design, in detail. I simply don't agree with your interpretation.

People are applying the current understanding of CPU cores retroactively to an old, innovative design.

You need to be able to point to some industry standard that was established by a vendor neutral group of experts from that time period, and show how AMD knowingly lied as to how their product did or did not meet that standard, and then you can say they were misleading. Provide that source information and I will agree with you. Othewise, I do not agree.
You can look at AMD's own patents like this one. Clearly that "module" was originally the core while "integer cores" were originally clusters.
BD-Core-1.jpg

Anyway "integer cores" is misleading. All CPU cores mainly operate with integer operations. Thus the "integer" part is in fact redundant. If you are wondering what is really a "cluster" in CPU design then look at this example..
st_block_diagram_s.jpg

Basically an integrated collection of execution units (including ALUs and AGUs). AMD's Bulldozer had two clusters, which can only execute the decoded instructions fed from the single shared front-end. The clusters themselves cannot fetch the instructions individually as in a real CPU core. The only difference is that AMD's simply added two back-ends to each cluster (thus creating two output threads).
 
  • Like
Reactions: TJ Hooker
What makes you think I need an explanation? I already understand their design, in detail. I simply don't agree with your interpretation.

I see people applying the current understanding of CPU cores retroactively to an old, innovative design.

You need to be able to point to some industry standard that was established by a vendor neutral group of experts from that time period, and show how AMD knowingly lied as to how their product did or did not meet that standard, and then you can say they were misleading. Provide that source information and I will agree with you. Othewise, this is nothing more than opinion and I do not agree.

Settlements like this are rarely about actual wrongdoing and more about preserving a positive brand image.

AMD knew they were stretching it with their designation of 8 cores. They could have easily just matched Intel as a 4 core 8 thread product and it would have never happened but their marketing team wanted to make it sound more impressive than it actually was as 8 cores sound better than 4 cores.

That said I would say an easy way to judge is to look at the CPU before Bulldozer and the CPU after (K10 and Ryzen). Both have cores that are designed very similarly besides maybe cache sizes and pipeline design but they are otherwise similar.

AMD settled knowing that dragging it out presented the possibility, and most likely, of losing the case and having a judgement vastly larger than the settlement.

Now in the end it is a stupid lawsuit as there was no harm to the consumer. I think a consumer should be somewhat responsible for researching what they are buying as marketing is never accurate.
 

TJ Hooker

Titan
Ambassador
I had a FX4170---how do I show I did. It is long ago.
Some of these Class Actions are lax---so, I wonder.
Interestingly enough, from what I can tell the class defined for this class action suit only contains those who bought 4 module bulldozer CPUs (FX-8xxxx and FX-9xxx).
Settlement Class Definition: The Settlement Class is defined as follows: “All persons who purchased one or more of the following AMD computer chips either (1) while residing in California or (2) after visiting the AMD.com website: FX-8120, FX-8150, FX-8320, FX-8350, FX-8370, FX-9370, and FX-9590.”1 Agreement § 1.28.2
https://regmedia.co.uk/2019/08/27/amd-eight-core-settlement.pdf

I'm no lawyer, but does this settlement not pave the way for other suits on behalf of people who bought 2 or 3 module bulldozer-derived processors?
 
Interestingly enough, from what I can tell the class defined for this class action suit only contains those who bought 4 module bulldozer CPUs (FX-8xxxx and FX-9xxx).

https://regmedia.co.uk/2019/08/27/amd-eight-core-settlement.pdf

I'm no lawyer, but does this settlement not pave the way for other suits on behalf of people who bought 2 or 3 module bulldozer-derived processors?

I would imagine so. It also paves the way for other states if this is indeed only in California. I think AMD should have just bit the bullet and covered their entire lineup based on Bulldozer and upped the settlement amount as well as covered more than just a state.

Hell if it went here I can imagine countries with stronger consumer rights would have a much easier time winning.
 
Now in the end it is a stupid lawsuit as there was no harm to the consumer. I think a consumer should be somewhat responsible for researching what they are buying as marketing is never accurate.
Not to such a degree,if you have to look up a patent to figure it out it should be something stated on the box.
Just like with cheese analogue and similar stuff, if they just wrote cheese on the package it would be fraud because what cheese is is common knowledge that everybody shares.
The term core was also universally considered to be a certain thing at that time (mainly a core that has full access to all of its resources, all of the time)