AMD Agrees To Pay Out $35 Per Chip Over FX Marketing Lawsuit

[pas008]

5 hours ago, mr moose said:

I think you might have misunderstood me then.  There seems to be a lot of people questioning the judicial system on these forums, except when the judicial system finds in favor of their opinion.  So my question is not why did they come to an agreement, but why does it seem people think it changes between just and fair and ignorant depending on the outcome?

 

Then drop the irrelevant criticism of my posts.  If you think I am being irrelevant, repetitive or ion some way deteriorating the forum, then report the post, otherwise go away.

 

 

No,  I have said nothing of the sort.    My post was about the inconstant way in which this forum seems to treat the judicial system depending on the outcome.

 

It's always a case by case thing.

 

The courts and those fighting in them call for expert witnesses all the time.   

 

 

yup its funny

look at the 970 settlement nvidia should pay pay pay oh no not my amd this is ridiculous

 

its performance was exactly right behind the 980 consistently

and did have the 4gb just slower .5

 

but I dont care either way I got my 2x in on that think in the end i paid 220 for each 970 lol,

[Brooksie359]

12 hours ago, mr moose said:

So when the courts find in favor of consumers on tech related stuff we all congratulate them, but when they find something like this all of a sudden they are ignorant?

 

I was under the impression the reason they had expert testimony was to inform the judge of the technicalities from a knowledgeable standpoint, thus the judge doesn't have to be an expert, they just have to know if the expert is being reasonable given the counter arguments etc.

 

 

I dont know how much you know about the US court system but from what I know "expert testimonies" is fancy talk for someone paid to say things that support one sides argument. It is very routine for one side ot pay for an expert to come and testify on a topic and they just so conveniently end up testifying in favor of the one that paid them to come in. Saying the judge was well informed by biased experts with ulterior motives is just not true. 

[Mira Yurizaki]

48 minutes ago, TechyBen said:

I think it was more a different way of doing it to SMT, and it just being a different type of benefit, one that was not universally beneficial. AMDs Buldozer example, meant you always knew how much compute each "thread" had (1/2 ) for each core, and could scale accordingly. In a SMT system, it depends on what the other thread is doing, and how much resources it uses up, it would be "between 3% - 30%". Where as AMDs was "always 15%" (as a made up example).

 

So AMDs example failed, because having the boost much higher some of the time, turns out better for general compute (and I guess most server farms too), than having a small increase all the time.

I think that's what AMD was trying to achieve here: consistent performance. But they had to face an established way of doing things and if you tell people that you need to make specific considerations for your processor, people aren't going to nudge all that much.

 

Plus every time some other radical CPU technology tried to eke its way into the market, the result was always the same: most programs that people run on CPUs almost never play nice to this thing called predictability, which is what you need for consistency.

 

4 minutes ago, pas008 said:

its performance was exactly right behind the 980 consistently

and did have the 4gb just slower .5

That's what always bugged me. In most cases, the GTX 970 performed pretty much as you'd expect. It's just in the rare (for the time) scenarios you'd run into the 3.5GB limit. If anything, NVIDIA should've been sued for not delivering on the promised driver fix that should've mitigated or fixed the way memory was handled. Though then again, how do you prove that it worked as intended?

 

But any way, in this case, I don't think NVIDIA's marketing necessarily lied about there being 4GB of VRAM. Plus the GeForce GTX 660 Ti also had a similar partitioning and nobody seemed to bat an eye at the time or retroactively sue NVIDIA for this.

 

I mean if anything, the alternatives to the GTX 970 scenario were either a card with 3GB of VRAM or lower yields trying to get that last ROP to work, which begs the question if that last ROP would've made a difference or not.

 

5 minutes ago, Brooksie359 said:

I dont know how much you know about the US court system but from what I know "expert testimonies" is fancy talk for someone paid to say things that support one sides argument. It is very routine for one side ot pay for an expert to come and testify on a topic and they just so conveniently end up testifying in favor of the one that paid them to come in. Saying the judge was well informed by biased experts with ulterior motives is just not true. 

Do you actually have proof of this?

[Brooksie359]

1 minute ago, Mira Yurizaki said:

I think that's what AMD was trying to achieve here: consistent performance. But they had to face an established way of doing things and if you tell people that you need to make specific considerations for your processor, people aren't going to nudge all that much.

 

Plus every time some other radical CPU technology tried to eke its way into the market, the result was always the same: most programs that people run on CPUs almost never play nice to this thing called predictability, which is what you need for consistency.

 

That's what always bugged me. In most cases, the GTX 970 performed pretty much as you'd expect. It's just in the rare (for the time) scenarios you'd run into the 3.5GB limit. If anything, NVIDIA should've been sued for not delivering on the promised driver fix that should've mitigated or fixed the way memory was handled. Though then again, how do you prove that it worked as intended?

 

But any way, in this case, I don't think NVIDIA's marketing necessarily lied about there being 4GB of VRAM. Plus the GeForce GTX 660 Ti also had a similar partitioning and nobody seemed to bat an eye at the time or retroactively sue NVIDIA for this.

 

I mean if anything, the alternatives to the GTX 970 scenario were either a card with 3GB of VRAM or lower yields trying to get that last ROP to work, which begs the question if that last ROP would've made a difference or not.

 

Do you actually have proof of this?

The first time I heard about it was while watching the Joe Rogan experience when he had a defence attorney named Mike schmidt on as a guest. He talks about it during the podcast and other corrupt things in the US court systems. My mom happens to be a lawyer and I asked her about it and she said it was fairly accurate.

[The King of the Undead]

My familly might get $35 for one of the dumbest things in history.

[pas008]

3 minutes ago, Brooksie359 said:

The first time I heard about it was while watching the Joe Rogan experience when he had a defence attorney named Mike schmidt on as a guest. He talks about it during the podcast and other corrupt things in the US court systems. My mom happens to be a lawyer and I asked her about it and she said it was fairly accurate.

but they are experts in their fields with credentials along with it

and of course they are paid because they have to look at the info to view it for or against the party

thats why both have their "experts"

and many times they dont want them to testify because how easily it can be slipped up because another expert can tell the other lawyer what to ask in cross

[Brooksie359]

1 minute ago, pas008 said:

but they are experts in their fields with credentials along with it

and of course they are paid because they have to look at the info to view it for or against the party

thats why both have their "experts"

and many times they dont want them to testify because how easily it can be slipped up because another expert can tell the other lawyer what to ask in cross

Yes but they are still paid by one side to come in and they want people to pay them to testify so they are going to want to make it look good for the side that brought them that way people are more likely to pay them to testify in the future. It's pretty clear that they have a conflict of interest. 

[pas008]

14 minutes ago, Mira Yurizaki said:

That's what always bugged me. In most cases, the GTX 970 performed pretty much as you'd expect. It's just in the rare (for the time) scenarios you'd run into the 3.5GB limit. If anything, NVIDIA should've been sued for not delivering on the promised driver fix that should've mitigated or fixed the way memory was handled. Though then again, how do you prove that it worked as intended?

 

But any way, in this case, I don't think NVIDIA's marketing necessarily lied about there being 4GB of VRAM. Plus the GeForce GTX 660 Ti also had a similar partitioning and nobody seemed to bat an eye at the time or retroactively sue NVIDIA for this.

 

 

that last little bit of vram did shit

980 dropped too because you are playing at unreasonable setting as is

but the fact remains its performance was right behind 980 in all scenerios and technically had 4gb of vram lol

but ill take it lol

[pas008]

10 minutes ago, Brooksie359 said:

Yes but they are still paid by one side to come in and they want people to pay them to testify so they are going to want to make it look good for the side that brought them that way people are more likely to pay them to testify in the future. It's pretty clear that they have a conflict of interest. 

of course they do that but the other side gets to cross them too so its still equal playing field and punch a hole in that if its bunk

[Brooksie359]

28 minutes ago, pas008 said:

of course they do that but the other side gets to cross them too so its still equal playing field and punch a hole in that if its bunk

I wouldn't say that is an even playing field. Having the expert looking like they agree with your side means alot in terms of other people's perception. Again to say that a judge is well informed because of an "expert's testimony" is sorta flawed because they are going to give biased information.

[Mira Yurizaki]

18 minutes ago, Brooksie359 said:

I wouldn't say that is an even playing field. Having the expert looking like they agree with your side means alot in terms of other people's perception. Again to say that a judge is well informed because of an "expert's testimony" is sorta flawed because they are going to give biased information.

I'm under the impression that the US court system allows for a lot of information to be open to both sides of the playing field so much so that if the other lawyer couldn't find some way to have someone there to verify the "expert testimony", whether it'd be themselves or another expert, then that's a problem on them and not a problem with the system.

 

Also who's to say a judge's opinion isn't flawed all the time? They're always going to be given biased information and opinions because at the end of the day, it's two sides trying to win an argument.

[Curious Pineapple]

7 hours ago, porina said:

I'm not US (or anywhere else) legal expert. Please correct if wrong, I understood that in the US there isn't a "loser pays winners costs" system like in UK? If that is correct, then even if you win, you're still out of pocket on lawyer fees and time taken. It makes sense to settle if that is going to be cheaper and get rid of a distraction faster, letting you get on with more important business.

Find the smallest reason to go to court and make a claim and people will. If you take a company to court and loose, they have lost out on defending thsemselves and you have paid pretty much nothing.

[pas008]

41 minutes ago, Brooksie359 said:

I wouldn't say that is an even playing field. Having the expert looking like they agree with your side means alot in terms of other people's perception. Again to say that a judge is well informed because of an "expert's testimony" is sorta flawed because they are going to give biased information.

thats why theres experts on both sides to present the case and both sides trying to punching holes in the other's side

debating

judge or jury decide from there on which is more relevant

thats why legal system is costly

 

fyi there have been many cases where an expert testified for both sides in same case

why because they are pointing out factual information for ones case point

and why experts/witnesses get recalled to testify

19 minutes ago, Mira Yurizaki said:

I'm under the impression that the US court system allows for a lot of information to be open to both sides of the playing field so much so that if the other lawyer couldn't find some way to have someone there to verify the "expert testimony", whether it'd be themselves or another expert, then that's a problem on them and not a problem with the system.

 

Also who's to say a judge's opinion isn't flawed all the time? They're always going to be given biased information and opinions because at the end of the day, it's two sides trying to win an argument.

agreed

[Quadriplegic]

$35 per chip? That's more than current value of FX chips on second hand market....

[79wjd]

6 hours ago, porina said:

 

 

Courts make rulings based on the law and facts presented. What we see as the headline reason for the case is not necessarily the same as the legal argument. Without digging in the detail, how this might have gone will come down to exactly what AMD promised, and how it is interpreted in a legal context. 

The law isn't quite that black and white. There are many ways things can be interpreted and the outcome can often vary quite drastically between different judges.

[Curufinwe_wins]

Good. It wasn't just about the cores, it was about the blatantly false statements AMD continously used to market a patently inferior product as if it was (in absolutely terms) better than the competition (ALL competition, not just at same price).

 

And for those of you who are worried about AMD. Newsflash, class action suits like this are tax deductible, so the net cost to AMD is pennies on the dollar here.

[79wjd]

 

2 hours ago, Mira Yurizaki said:

Do you actually have proof of this?

Expert witnesses are generally hired by each side and each aim to format their argument to best support the case they've been hired to testify for.

 

Some expert witnesses might perjure themselves, but most will just focus on facts/arguments that support their side and conveniently try to ignore ones that don't.

 

Of course, there is usually a back and forth of some sort where the expert on the opposing side can raise counter arguments.

[leadeater]

5 hours ago, TechyBen said:

Yeah. I really don't like the current trend of "4GHz CPU" when it's multi core, 1 or 2 cores hit 4GHz, the rest 3.5... IMO it should be (on the packets/packaging) "3.5Ghz + 2x4Ghz" or some similar method of min/max/average, or the number of cores hitting it/each.

 

I still agree, AMD said "8 cores", and it was. They did not say "64bit" when it was 32, or "8mb cache" when it was 2mb... The cores were slow/shared/smaller than Intel's, but they were still cores.

 

If however, the OS treats them only as threads, in a similar way to HT "cores", then on a useage case, they would not perform as cores. If it was purely a math/compute limitation, then IMO, they were slower compute, and that is down to the consumer to check (same as a Pentium 4 2 core (yes they made those!) at 3GHz is not quicker than the latest i3 with 2 cores running at 2.9Ghz...).

A big reason why the performance wasn't very good was AMD chose to share the wrong thing, everything was shifting to FP and vectorized code which meant a CMT design with a 1 FP and 2 INT was at a severe disadvantage. If you were doing any kind of heavy INT work the performance was far closer to 2 tradition cores.

[mr moose]

10 hours ago, TechyBen said:

Averages are not individual points. Individual points are not averages. We can all swap back and forth over "but they said, they they said" on a forum with multiple opinions and factual viewpoints (yes, facts can have multiple valid solutions, but the choice of each one could then be an opinion, but both valid).

Doesn't matter. I am not trying to individualize this forum to the one person.  I am addressing what I see as a trend.  That is quite a common thing people do around here.

 

10 hours ago, TechyBen said:

A lot of people get annoyed/offended here when someone posts "you always say that" as if this forum is a unified creature or person.

The difference is I am actually addressing the forum as a whole when I point out a trend I think is worthy of discussion and not talking to one direct person.

 

10 hours ago, TechyBen said:

Again... how and what is this communal person in this forum? Why keep treating it like one single person? What?

I don't know why you keep trying to turn the whole forum into one person. If my comments don't address anything you've said then ignore them.  They are addressed at many people not just one person.

 

10 hours ago, TechyBen said:

Yeah. But those that reach the news, always seem to say strange things. Like they had ulterior motives. I guess that does not happen in smaller cases, with less money/fame/pride being thrown around.

Then I would suggest not using the news as your only source.  It happens in all cases larger and small.

 

10 hours ago, Trixanity said:

Because people have opinions. People have knowledge. They're questioning the decision making of courts when the very outcome can be determined by where and who preside over the case.

They criticize the merits on which some rule in these cases because we've seen people who get expert counseling that still don't grasp the subject on even a basic level.

The problem is not general criticism, but that when people claim the courts (namely judges) are ignorant and never understand the technology thus make the wrong decision, then that is an absolute, it's not the sort of claim you can turn on and off depending on the outcome of any specific case.   

 

10 hours ago, Trixanity said:

I can't and won't report a post when it's not against the rules; I've told you this already. However just as you're free to comment on them, I'm free to comment on you. Ultimately we're accomplishing the same: nothing.

 

Trolling is the other name for that.     You are criticizing me and not the my opinion on the topic.    If my post is not against the rules and you have nothing to say other than be critical of how I post then I suggest you learn how to ignore things.

[mr moose]

5 hours ago, Brooksie359 said:

I dont know how much you know about the US court system but from what I know "expert testimonies" is fancy talk for someone paid to say things that support one sides argument. It is very routine for one side ot pay for an expert to come and testify on a topic and they just so conveniently end up testifying in favor of the one that paid them to come in. Saying the judge was well informed by biased experts with ulterior motives is just not true. 

Yes they are paid to say things, but they will not lie under oath and are still their in a professional capacity.  I did say this:

17 hours ago, mr moose said:

So when the courts find in favor of consumers on tech related stuff we all congratulate them, but when they find something like this all of a sudden they are ignorant?

 

I was under the impression the reason they had expert testimony was to inform the judge of the technicalities from a knowledgeable standpoint, thus the judge doesn't have to be an expert, they just have to know if the expert is being reasonable given the counter arguments etc.

 

 

 

Both sides present an expert witness, their job is to explain to the judge in laymen's terms the technical aspect of the case.  The Judge only has to balance the explanations and opinions of the witnesses against each other and not his own knowledge.

 

If you hire an expert witness and they are good enough at articulating exactly which aspect you want promoted then that is not a failing of the court.  That is a failing of your case.

[Lathlaer]

24 minutes ago, mr moose said:

If you hire an expert witness and they are good enough at articulating exactly which aspect you want promoted then that is not a failing of the court.  That is a failing of your case.

Judging by the discussion in this topic, clearly both sides should've hired LTT forum members as their experts  

[porina]

4 hours ago, 79wjd said:

The law isn't quite that black and white. There are many ways things can be interpreted and the outcome can often vary quite drastically between different judges.

Agreed that things are never straight forward, but it is the general principle.

2 hours ago, leadeater said:

A big reason why the performance wasn't very good was AMD chose to share the wrong thing, everything was shifting to FP and vectorized code which meant a CMT design with a 1 FP and 2 INT was at a severe disadvantage. If you were doing any kind of heavy INT work the performance was far closer to 2 tradition cores.

I don't recall, were they actually good at non-FP code?

 

I never owned one, but did ask owners to run benchmarks for me, and they pretty much sucked as much as expected if you try to get them to do anything FP intensive. Only recently did we get per-core equality (or even slightly exceeding) FP rate compared to Intel consumer cores. Zen(+) was still effectively half rate, with the AMD story being that for consumer uses, FP wasn't important. I guess you could partially argue that having more cores for your money meant that overall the deficit wasn't as big. Still, they have decided to finally catch up with Zen 2. Cutting FP performance gives big savings on both die area and peak power consumption, so it always felt more like justifying a cost cutting measure.

[leadeater]

13 minutes ago, porina said:

I don't recall, were they actually good at non-FP code?

Well not nearly as bad as FP code in comparison to number of units but still behind Intel's much faster INT units.

[Curufinwe_wins]

See my comments here. But the core definition doesn't actually matter compared to the patently false and intentionally misleading marketing AMD used with the device.

 

 

 

And then this beauty.  AMD claiming for years after the release of bulldozer to have the first and ONLY native 8 core desktop processor 'for unmatched multitasking performance'

 

[Jito463]

19 hours ago, TechyBen said:

Ah. That may change things then. If the OS cannot really make much use of the secondary cores with that shared cache, then they would perform (functionally) much closer to a "hyper thread", using spare resources, not an entire CPU resource chain. Hmmmm.

It's similar to SMT/HT, but with actual physical cores instead of running a second thread on the same core.  The theory was that one core of the cluster could be processing data, while the other core could be receiving new data to process.  It just turned out that the scheduling didn't work like AMD hoped it would, so the performance was lacking.  Here's a picture showing the inner workings of the Bulldozer architecture (compared to the improved Steamroller architecture).

18 hours ago, Tedny said:

but those cores still a cores, so how AMD lost this, I don't know 

They didn't technically lose.  They settled out of court so they could get rid of the lawsuit, and not risk the judge siding against them in an even bigger judgment (or even just shelling out more money for lawyers over a longer period, only to end up at the same conclusion).

15 hours ago, LAwLz said:

I see this as a victory for consumers, because it hopefully scares other companies off from trying to create deceptive marketing material in the future. It's a shame that this happened to AMD now that things are finally going well for them, but I think it's good that the law stepped in and protected consumers.

 

I'm just surprised that it happened with the FX cores rather than the ******** marketing AMD tried to pull with "compute cores". They started combining the CPU and GPU cores in some processors. So you could buy a "6 compute core CPU" which turned out to be a dual core with 4 GPU cores.

Exactly what did "the law" protect people from?  People claim AMD used deceptive marketing, but I can't recall any from that time period and no one has been able to actually provide any specific examples.  Reminds me of people on this forum (and elsewhere) hyping up rumors about AMD's new video cards, and then when they don't live up to the hype, others use it to attack AMD, despite the fact they never spread the rumors to begin with.

 

Though I do agree with you about the APU "compute cores".  While it's technically true, it was irritating trying to decipher how many CPU cores vs GPU cores an APU had.

15 hours ago, LAwLz said:

It's not just cache that was shared between the "cores" though.

It was the fetch, the decode, the FP scheduler, the i-cache, and L2 cache that were shared between each integer core inside each module.

 

And it's not up to AMD to decide what is and isn't part of a core, or "complement system".

Just like it wasn't up to Nvidia if the 4GB of VRAM in the 970 was actually 4GB or misleading marketing.

Why isn't it up to AMD to decide?  It's not like there's a set definition of what specifically is included in a CPU core, apart from some very basic elements.  It literally had two physical cores to process data, that they shared components didn't make either of them less of a core.

14 hours ago, TechyBen said:

(same as a Pentium 4 2 core (yes they made those!) at 3GHz is not quicker than the latest i3 with 2 cores running at 2.9Ghz...).

Funny thing about those early Pentium dual cores, they literally just tacked (glued?) them onto the same substrate, but didn't design any way for them to directly communicate, requiring a rerouting out to RAM in order for each core to share data with the other.  My word, those things were awful.  Hot, slow and power hungry.  If people think the Bulldozer FX chips are bad, they should try using one of those.

12 hours ago, Mira Yurizaki said:

Do you actually have proof of this?

Each side of a case will bring their own hired expert witnesses to testify.  It just comes down to which witness made the better presentation to the judge and/or jury.

3 hours ago, Curufinwe_wins said:

AMD claiming for years after the release of bulldozer to have the first and ONLY native 8 core desktop processor 'for unmatched multitasking performance'

And in certain tasks, they did have superior multicore performance.  Not all tasks benefited from the Bulldozer design, but certain ones did.  Ultimately, AMD bet on FP performance being less important, and they were wrong.  Does that justify a lawsuit?  I suppose the judge thought it did.  Personally, I'm not convinced.

 

In any event, now that AMD has settled out of court, this should be the end of the FX saga.