Judge Allows Lawsuit Over AMD's FX Processors to Continue

[MadmanRB]

No they are still 8 core CPU's.... technically.
This judge has opened pandoras box.

 

[Barty1884]

So by explaining how their CPU works, they mislead the customers?

No, by marketing them as "8 core", etc, they did mislead consumers - based on the accepted definition of what constituted a "core" in the same era.

It is called CPU design. AMD FX design work this way, but they do have 8 cores... if you look on the silicon, they do have 8 cores, so it is 8 cores.

That's the point though. They're quasi-cores, in essence.
Yes, in the silicon there are 8 cores - but when sharing resources have them function like 4c/8t opposed to true, independent cores, you potentially have a problem.

The issue is, as no legal definition exists for what constitutes an x86 "core", you potentially see a default precedent being set by the proceeding products in the industry.
Both AMD and Intel marketed "cores" consistently, prior to FX.
Whether enough of a precedent was set by default previously, is what this lawsuit should end up settling.

They did explain how their CPU works and they are 8 cores.

They did.... but that's more of a bait and switch.

Marketing (that the average consumer sees): "First consumer 8 core CPU".
Releasing a datasheet etc, confirming it's not a 'true' 8 core due to shared resources, cannot reasonably be expected to reach the same volume of average consumers that the marketing did -- And I would expect AMD will be hard pressed to argue that it would've.

Only in the US, lawsuits are a national sport.

Yes, the US is more litigious that then rest of the world, that's hardly breaking news. Also doesn't make this less relevant to the industry.
In this instance, the class action is defending the consumer. You could view it either way.

Personally, I think AMD were misleading but there's a shared culpability in place.
Yes, AMD changed the generally accepted definition of a "core" for marketing purposes BUT, IMO, consumers should do their own research before buying into the hype of any given product.


Definitely interested to see how this plays out.

 

[MadmanRB]

Still if AMD loses it could set off a chain reaction and in a bad way.
Meaning that we can sue over mere definitions, isnt the American court system already filled with legal nonsense as it is?
Again this is pandoras box IMHO.

 

[metalcraft218]

Pay attention here, the author hit it on the head and did his research.

"but if sharing resources negatively impacts per-cycle performance, then AMD necessarily has to lean on higher clocks or a greater emphasis on threading to compensate."

I happen to own and am currently on a system with an AMD FX 8370 @ a very conservative 4.3 Ghz and it's overall performance for what I need it to do is outstanding.

Before they came out with Ryzen, this was THE best performing CPU of the era, just under the Intel Core i7-3770 @ 3.40GHz. Hyper-threading was given a bump and the clock speeds were also bumped to 4.0 Ghz...the highest in it's class at the time. Just for performance sake...and I quote:

"As of 2011, the Guinness World Record for the highest CPU clock rate is an overclocked, 8.429 GHz AMD Bulldozer-based FX-8150 chip"...so noone can say it was a garbage chip.

AMD was was consistently outmatched by Intel's multi-threading architecture, so AMD went with more "cores" and higher top speeds to compensate.

What defined cores in that time (since there really was no concrete term that cores had to be individualized if the architecture could support it) was exactly as how the author described it. Each "core" may have shared resources, but they all had their own set of instructions as separate cores and performed their own threaded tasks.

I personally define it as an 8-core and every single program that I have ever installed from Passmark to Futuremark to CPU-Z has always recognized it as an 8 core CPU with 8 individualized cores, and obviously there has to be enough resources to support it and it does. So what's the real problem here? Greed.

People will sue ANYONE today for a little bit of spending cash and for the record, just because the suit was accepted in court by no means declares a win for anyone. Right now, it's all about definition and future legislation of it's translation.

 

[TJ Hooker]

It is so stupid, even an old Cinebench is showing the 8 Core variant having nice scalable score of 90% over the 4 core FX 4100. The 6 cores FX100 is 50% faster than the 4 cores iteration... how can this be if they are only 4 cores.

I think you need to reevaluate your reasoning here. If the allegations are valid (and I am not saying they are), then the 'real' core count of FX 4xxx, 6xxx, an 8xxx would be 2, 3 and 4. Which means you'd expect a ~50% performance increase going from 4xxx to 6xxx, and ~100% going to 8xxx. You'd expect the same relative performance results whether each bulldozer module is two 'real' cores or not.

 

[MadmanRB]

People will sue ANYONE today for a little bit of spending cash and for the record, just because the suit was accepted in court by no means declares a win for anyone. Right now, it's all about definition and future legislation of it's translation.

Indeed, and the worst part is that with so many lawsuits going around it muddles actual cases where the plaintiff is right and the defendant is wrong.
May I remind people of the hot coffee incident?
https://www.youtube.com/watch?v=PAzMMKIspPQ

When this case was ruled in flavor if the plaintiff (and rightfully so) it sadly set off a chain reaction of copycat cases that made the original case seem silly and petty on the part of the plaintiff.

Now i am all in favour of the consumer when it comes to suing over defective products or false advertising but the bulldozer processors do not qualify for such a lawsuit.
Again its suing over a definition and not the actual technology and AMD losing could set off a really nasty chain reaction in the courts muddling the system even further.

 

[Barty1884]

People will sue ANYONE today for a little bit of spending cash and for the record, just because the suit was accepted in court by no means declares a win for anyone. Right now, it's all about definition and future legislation of it's translation.

They will, but a class action typically needs to be a little more substantiated than your run of the mill threaten lawsuit/get a minimal payoff to go away.

While the ruling doesn't declare a win for either party*, the fact that AMD haven't had this thrown out immediately means they don't have a sufficient defense to persuade this particular judge, at this time.

I'd be very, very surprised if this wasn't found in AMDs favour on the lack of legal precedent - and this judge probably doesn't want to be the guy who sets off future debates based on their interpretation.

*Realistically, even if there was a "win" either way right now, there would be numerous rounds of back & forth appeals.

*Edited for brevity*
Indeed, and the worst part is that with so many lawsuits going around it muddles actual cases where the plaintiff is right and the defendant is wrong.
May I remind people of the hot coffee incident?

When this case was ruled in flavor if the plaintiff (and rightfully so)

Now i am all in favour of the consumer when it comes to suing over defective products or false advertising but the bulldozer processors do not qualify for such a lawsuit

Yes, copycat lawsuits do tend to arise once a precedent has been set - that's the risk taken when setting the precedent. Unfortunately, this does impact any given judges thinking. Their name will be attached to any precedent set.

Not 100% sure I understand your logic on the Hot Coffee incident though.
You were in favor of the plaintiff, but believe it should only boil down to defective products or false advertising?
The coffee was advertised as "hot", and was not deemed defective based on industry standards.

 

[MadmanRB]

Yes, copycat lawsuits do tend to arise once a precedent has been set - that's the risk taken when setting the precedent. Unfortunately, this does impact any given judges thinking. Their name will be attached to any precedent set.

Not 100% sure I understand your logic on the Hot Coffee incident though.
You were in favor of the plaintiff, but believe it should only boil down to defective products or false advertising?
The coffee was advertised as "hot", and was not deemed defective based on industry standards.

No I was pointing out that AMD isnt in the wrong here, not saying they are in the right either but we are still having a trial over a definition.
And if that is the case then AMD is in the right as yes they still made 8 core CPU's no question.

 

[Barty1884]

Unfortunately, it's a matter of opinion until a definition is legally defined.

You (among with many others) are of the belief the FX chips meet expectations as an 8 "core".
I (with many others too), would argue that, given the shared resources/dependencies, it's more akin to a 4 "core", 8 thread chip.

I fully expect this'll be settled in AMD's favor though (or out of court settlement, not 100% sure if that's possible with a class action).
Whether that means a legal definition of what constitutes a "core" or not, probably rests more on how good the plaintiff(s) argument is that it's not a "core" than AMD though.

 

[COLGeek]

All about semantics, it would seem, more than anything else. One could even argue to be frivolous. It really doesn't solve or remedy anything. No one was forced to purchase these devices.

 

[donlilli4]

This lawsuit is one of several reason that I don't use or own AMD-based chips. It's bad enough that AMD's technology ends up in certain items like game machines and other controllers that I have to work with or that I own like the Xbox One and Nintendo Switch. Over the last 37 years I have worked with and also built systems using both AMD and Intel chips along with Cyrix and other chip systems. In all those years, I've never had any major issues from Intel, or issues that Intel wasn't willing to own up to and correct. I've had AMD chips burn up on me (Thoroughbred/Thunderbird core CPUs) even when everything is installed correctly. Substandard motherboards and chipsets from VIA, SiS, and other sources that caused unnecessary errors and issues. Cyrix was also famous for their problems too, which is also why they no longer exist.
I don't trust AMD chips, and I likely never will. Intel CPUs are far more powerful and superior. Yes, Intel is more expensive, but they last a LONG time. I have Intel units that I run 24 hours a day and have for the last 20+ years and except for minor routine downtime maintenance have been running continuously. This is not to say I haven't had a failure or two occur, but in the last 20 plus years, I can still count the failures on one hand with fingers left over. THAT is reliability. Unfortunately, I don't know of anyone that has purchased AMD-based systems that have not had a failure. My brothers are AMD fanatics, yet they keep blowing boards and CPUs left and right. Tells you something, doesn't it? At least when Intel says it has 8 cores, the CPU has eight individual cores running their own resources and power, and Intel doesn't hide its specs either.

 

[ctchewy]

Facts is facts. I said when this architecture came out that it was nothing more than hyperthreading on steroids. I still hold that opinion today. I love what AMD has done with it's ZEN products but what the marketing department did with bulldozer was misleading and deceitful. If FX 8xxx are 8 core processors then so are core i7's from the same time frame.

 

[firefoxx04]

You people blow my mind. The die literally has 8 cores on it. The cores just perform terrible due to shared resources.

Thats like comparing a 80s v8 engine to an 80s Group B 4cyl engine. The 4cyl performed better in almost all situations, does that make the v8 less than a v8? No.

 

[drinking12many]

This lawsuit is one of several reason that I don't use or own AMD-based chips. It's bad enough that AMD's technology ends up in certain items like game machines and other controllers that I have to work with or that I own like the Xbox One and Nintendo Switch. Over the last 37 years I have worked with and also built systems using both AMD and Intel chips along with Cyrix and other chip systems. In all those years, I've never had any major issues from Intel, or issues that Intel wasn't willing to own up to and correct. I've had AMD chips burn up on me (Thoroughbred/Thunderbird core CPUs) even when everything is installed correctly. Substandard motherboards and chipsets from VIA, SiS, and other sources that caused unnecessary errors and issues. Cyrix was also famous for their problems too, which is also why they no longer exist.
I don't trust AMD chips, and I likely never will. Intel CPUs are far more powerful and superior. Yes, Intel is more expensive, but they last a LONG time. I have Intel units that I run 24 hours a day and have for the last 20+ years and except for minor routine downtime maintenance have been running continuously. This is not to say I haven't had a failure or two occur, but in the last 20 plus years, I can still count the failures on one hand with fingers left over. THAT is reliability. Unfortunately, I don't know of anyone that has purchased AMD-based systems that have not had a failure. My brothers are AMD fanatics, yet they keep blowing boards and CPUs left and right. Tells you something, doesn't it? At least when Intel says it has 8 cores, the CPU has eight individual cores running their own resources and power, and Intel doesn't hide its specs either.

I have never had an AMD chip fail, ever, I still have 940s, Opterons, and FX's, even semprons and old Athlons etc sitting in boxes not because they failed, just that my needs required new chips. My FreeNAS box is still rocking an FX-8320 it has ran for years almost 24x7 first in my gaming machine, then as an ESX host, now as my Freenas box. I am not saying they never fail, but your personal experience and mine have been very different. All have been overclocked some wayyyy more than others. My old Opteron 165 ran at 3.0 ghz for like 3 years before I retired it. My current Ryzen 1700 runs at a "conservative" 3.7Ghz on an H100 cooler barely breaks 110 degrees.

 

[King_V]

I'm wondering, where does the "shared resources" begin and end?

L1 cache? L2? Other resources? System RAM? The last is, of course, absurd, because it would make EVERYTHING a 1-core.

Don't current chips by Intel and AMD also share resources?

Not asking to play Devil's Advocate, I just really don't know the answer.

 

[kazak9800gt]

When someone says "FX is just 4 cores with hyper threading", then why does it suck so hard in single thread applications? Intel CPUs with HT don't suck in single thread. And by the way, FX is a beast in integer tasks - it encodes x264 pretty fast. If AMD gets sued over this, <ModEdit> "doubled" VRM phases on motherboards should be insta class action lawsuit, because that's real <ModEdit>

<ModEdit>
Language.

 

[Barty1884]

I'm wondering, where does the "shared resources" begin and end?

As I recall, each FX chip had X number of modules... each module consisted of 2 "cores" (for simplicity's sake, I'll use that terminology) with their own L1 cache and shared a single FPU (floating point unit) and L2 cache.

*EDIT*
Actually, not 100% sure on whether L1 was dedicated per "core" or shared in some way.
Some more shared too:
https://www.tomshardware.com/reviews/fx-8150-zambezi-bulldozer-990fx,3043-3.html
https://www.reddit.com/r/buildapc/comments/1e8226/discussion_amds_module_architecture_the_fx_8350/

 

[Eximo]

This lawsuit is one of several reason that I don't use or own AMD-based chips. It's bad enough that AMD's technology ends up in certain items like game machines and other controllers that I have to work with or that I own like the Xbox One and Nintendo Switch. Over the last 37 years I have worked with and also built systems using both AMD and Intel chips along with Cyrix and other chip systems. In all those years, I've never had any major issues from Intel, or issues that Intel wasn't willing to own up to and correct. I've had AMD chips burn up on me (Thoroughbred/Thunderbird core CPUs) even when everything is installed correctly. Substandard motherboards and chipsets from VIA, SiS, and other sources that caused unnecessary errors and issues. Cyrix was also famous for their problems too, which is also why they no longer exist.
I don't trust AMD chips, and I likely never will. Intel CPUs are far more powerful and superior. Yes, Intel is more expensive, but they last a LONG time. I have Intel units that I run 24 hours a day and have for the last 20+ years and except for minor routine downtime maintenance have been running continuously. This is not to say I haven't had a failure or two occur, but in the last 20 plus years, I can still count the failures on one hand with fingers left over. THAT is reliability. Unfortunately, I don't know of anyone that has purchased AMD-based systems that have not had a failure. My brothers are AMD fanatics, yet they keep blowing boards and CPUs left and right. Tells you something, doesn't it? At least when Intel says it has 8 cores, the CPU has eight individual cores running their own resources and power, and Intel doesn't hide its specs either.

I have never had an AMD chip fail, ever, I still have 940s, Opterons, and FX's, even semprons and old Athlons etc sitting in boxes not because they failed, just that my needs required new chips. My FreeNAS box is still rocking an FX-8320 it has ran for years almost 24x7 first in my gaming machine, then as an ESX host, now as my Freenas box. I am not saying they never fail, but your personal experience and mine have been very different. All have been overclocked some wayyyy more than others. My old Opteron 165 ran at 3.0 ghz for like 3 years before I retired it. My current Ryzen 1700 runs at a "conservative" 3.7Ghz on an H100 cooler barely breaks 110 degrees.

FYI that Nintendo Switch runs on an ARM based processor by Nvidia with Maxwell era graphics. Sony also uses AMDs APUs in their console though.

 

[rlarose000]

AMD did not mislead anyone.

Think of it this way, is an Intel 486 SX Processor a 0 Core Processor? Yes or No?

I'd say no. It is a single core processor. Therefore a Core doesn't technically need it's own FPU unit, L2 Cache etc to be a Core.

These features are added bonuses which is why they're advertised to us on the specs sheet.

It's up to us to read up on the specs and make a choice.

If you don't do that, then that's really your fault.

 

[CerianK]

It seems obvious to me that if one buys an Intel CPU that they will get only a single core, as 'Core' (2/duo/i3/i5/i7/i9) is not plural. I see the makings of a new lawsuit claiming that those who expected to receive only a single core from Intel were misled by the name of the processor which 'promised' to only provide one core.

My Intel CPUs tend to be Xeons, so unfortunately I will not qualify to be part of the class.

 

[littleleo]

The FX Core count controversy is still alive. I remember Tom's articles back in the day when Tom's was still trusted. One article would call the FX-8000s models 4-core and the next 8-core. And the beat still goes on....

 

[spentshells]

I always thought of bulldozer as hardware hyperthreading

 

[Shumok]

Unfortunately, it's a matter of opinion until a definition is legally defined.

You (among with many others) are of the belief the FX chips meet expectations as an 8 "core".
I (with many others too), would argue that, given the shared resources/dependencies, it's more akin to a 4 "core", 8 thread chip.

I fully expect this'll be settled in AMD's favor though (or out of court settlement, not 100% sure if that's possible with a class action).
Whether that means a legal definition of what constitutes a "core" or not, probably rests more on how good the plaintiff(s) argument is that it's not a "core" than AMD though.

It's not like a 4-core/8-thread chip. It's single threaded performance is too low vs it's multithreaded performance. It's like an 8-core/8-thread chip.

 

[joeblowsmynose]

Well this seems pretty silly.

Personally I always referred to the bulldozer modules as having two cores (sortof). I always use that (sortof) disclaimer even now when I refer to them.

I do like the argument that Intel must have been producing zero core processors before the had a floating point unit - promising one while getting zero is certainly more concerning. Imagine all those people who were tricked into believing that their CPU had a core, when really there was nothing. Pretty funny.

I can picture how the purchase went that led to this lawsuit ...

"arrhh ... it has a high number!? I want the one where they said a high number in their advertising ... that's all the research ... *hic* .. that I need to make a sound puchasing decision -- whatever hasz a high number, buy me two, okaaay!?? ... *hic* ... what are deeze? Processors? What are those ... it has eight you say? Thanks! .. *hic* I didn't have enough things with the number eight on it ... *hic* ... good purchase!!"

I'm pretty sure people buy processors on more than one factor - if not, your too stupid to deserve any lawsuit winnings anyway.

The lawsuit is just from some dork from Alabama named Tony Dickey who bought two, yes only two, AMD bulldozer processors .. . and then cried like a baby when learned what the architecture looked like on the inside so he thinks his pain and anguish over this is worth $5 million dollars. Idiot.

Apparently understanding the performance of the product you buy doesn't matter, but if a CPU doesn't look on the inside they way you imagined it, that's grounds for a lawsuit ... lol.

 

[Shumok]

Facts is facts. I said when this architecture came out that it was nothing more than hyperthreading on steroids. I still hold that opinion today. I love what AMD has done with it's ZEN products but what the marketing department did with bulldozer was misleading and deceitful. If FX 8xxx are 8 core processors then so are core i7's from the same time frame.

Except Core i7's acted like 4-core/8-thread chips while FX-8000's acted (and still do) like 8-core/8-thread chips. You are just wrong. FX cores are just slower cores.