Judge Allows Lawsuit Over AMD's FX Processors to Continue

[Curufinwe_wins]

BTW this lawsuit is a tiny shell of what it should be. Affecting only California consumers and those that visited AMD's website before purchase.

[Dabombinable]

1 hour ago, RejZoR said:

AMD's 8 cores were actual real 8 cores. There was no hyperthreading involved. It's just that they had separate INT units, but shared FP unit. And that's seen by some as cheating...

Note how in nearly all games the FX series performed in a similar manner to early quad cores, and dual core i3?

[Quadriplegic]

58 minutes ago, Subway said:

FX 8350 is still pretty good in 2019 consider it is 8 years old processor! 

Yet obviously worse than Sandy Bridge CPUs, which are even older.

[Dylanc1500]

42 minutes ago, Taf the Ghost said:

FX 6300, 6350, 8300, 8320, 8350 and I think there's an 8370 in there.

Don't forget 4300, 4320, 4350, 8320E, 8370E and 9370. Can't leave out the legendary 9590 since it is part of that family too lol.

[yian88]

Not trying to defend, but AMD did not cheat on core count and hyperthreading is a completely different thing.

AMD had shared resources for 2 cores making 1 module, hyperthreading is software split up of a single core resources, hyperthreads are not physical.

Yes you do have less performance per core in 1 module because of shared resources in many scenarios that became the bottleneck but you do have 8 physical cores in FX 8000 series etc.

AMD didnt lie they had their designs public and i do remember people writing articles about the shared resources reducing performance.

I wonder who sued them i doubt regular users did, is it a "rival" or someone who has beef with AMD behind this.

You could sue nvidia for gtx 970 with 3.5gb fast GDDR5 and 512mb of slow vram, but they advertised 4GB vram gpu which the GPU does have, hence you would lose in a sue case.

 

[Tristerin]

 

2 hours ago, 2Buck said:

As much as the FX CPUs sucked (I type this from a FX 8320 @ 4.5ghz), this is bullshit. There are 8 physical cores in the CPU, you can't change or deny that fact. Yes, they share resources, but there are, in fact, 8 physical cores. What's next, are we gonna claim the 2990WX only has 16 cores because of the way half the cores access RAM?

Agreed - and tbh while they suck...as more games take advantage of higher cores my FX is still competitive in 2019.  Builds me PLENTY of frame time for what I want to see (60fps locked @ 1600x900.

 

I have yet to see a game shut down a core while playing because it was sharing resources.  Not my older titles nor any of the new titles. 

[Brooksie359]

2 hours ago, dalekphalm said:

 

 

Plus as computing evolves, what's actually inside a CPU core in, say, 10 years from now, might not even have any of the components inside one now (obviously this is a bit of a stretch, but you get the idea).

 

If someone wants to sue AMD for a shit processor that failed to deliver promised performance? All the power to them. But suing them over "Your 8 cores are different than Intel's cores!" is just stupid to me.

Next thing you know people are going to be suing Qualcomm because their cores suck for Windows. 

[thorhammerz]

2 hours ago, dalekphalm said:

So you're implying the lawsuit exists because there's no definition? Source for that claim?

"No definition"? The implication is whether or not the definition was ambiguous (or not!), and how said definition was marketed and perceived. Good try attempting to twist words though.

Quote

We explored the Bulldozer concept when we reviewed the FX-8150 back in 2011. AMD had previously taken a bit of a holier-than-thou approach to counting cores because it believed Intel's Hyper-Threading, which offers two logic processors for each physical core, was disingenuous. But then AMD developed the Bulldozer module to share "instruction fetch and decode stages, floating-point units, and the L2 cache" between two cores and changed its tune.

This is where the lawsuit comes in. Some people who bought the FX-8120, FX-8150, FX-8320, FX8350, FX-8370, FX-9370, and FX-9590 processors believe AMD misled them with its claims about how many cores the processors had. The company marketed its FX processors as the world's first eight-core CPUs, but because of the shared resources between cores in Bulldozer modules, the lawsuit alleges that this claim was deceptive about real-world performance.

We pointed out nearly a decade ago that AMD had suddenly changed its definition of what a "core" is for these processors:

"To best accommodate its Bulldozer module, the company is saying that anything with its own integer execution pipelines qualifies as a core (no surprise there, right?), if only because most processor workloads emphasize integer math. I don’t personally have any problem with that definition, 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."

AMD's goal for Bulldozer modules was to offer many of the benefits of having a bunch of cores without actually having to, you know, use completely separate cores. The company said at the time that a Bulldozer module could average 80% the performance of two complete cores. But the lawsuit's plaintiffs weren't looking for eight-core processors that performed like six-core processors; they wanted the performance of eight individual cores.
 

Relevant bits are quoted for your convenience, from Tom's Hardware.

 

The lawsuit is ultimately about how AMD marketed their products, and whether said products in the eyes of the plaintiffs matched up to said marketing.

[JoeCoke]

2 minutes ago, Tristerin said:

 

Agreed - and tbh while they suck...as more games take advantage of higher cores my FX is still competitive in 2019.  Builds me PLENTY of frame time for what I want to see (60fps locked @ 1600x900.

 

I have yet to see a game shut down a core while playing because it was sharing resources.  Not my older titles nor any of the new titles. 

Y'know, I take it back, even though their IPC is really bad (compared to Sandy bridge, and even phenom ii), saying they suck is a little harsh. All those AMD fanboys from back then turned out to be right, the CPUs haven't aged too bad now that games can take advantage of all the threads.

 

My FX 8320 has done me very well. I edit 1080p 60 FPS video in premiere, stream/record at 720p 60 FPS in OBS, hosted game servers while playing games (before I built dedicated game servers), and plenty of other things. This CPU has always kept up.

 

Oh, and as they get cheaper on eBay, it's gonna be cool seeing them take over socket 775's role as the "extreme budget" CPUs. Just now on eBay I found two 8320s under 60$, maybe soon they'll be 40$ and under. Hell, there's one for 40$ on there that has couple of bent pins. Before you know it, you'll be able to buy a 6300 for 20$.

 

The FX CPUs started out really bumpy, and they were far behind the competition in single threaded tasks, but overall they've turned out... Alright. Way to go Faildozer.

[porina]

22 minutes ago, yian88 said:

You could sue nvidia for gtx 970 with 3.5gb fast GDDR5 and 512mb of slow vram, but they advertised 4GB vram gpu which the GPU does have, hence you would lose in a sue case.

From memory, people did, and people won. At least, it was for some part of the US.

 

1 hour ago, Curufinwe_wins said:

Again. Intel lost (well agreed to settle once it was evident they were going to lose) a class action suit over similar marketing with regards to the Pentium 4 lineup amidst the 'Gigahertz War'.

I don't recall that one, maybe a bit before my time. To my understanding, settling may be preferable to winning, if winning is going to cost you a load of time and resources you want to spend on other things. It's less likely in the UK, as the system is generally that loser has to pay winners costs, so you're going to have to be pretty confident to take anyone to court. Maybe I'm wrong, I don't believe that is the case in US, so even if you win a defending case, you're out of pocket the costs racked up to do so. Settling may be cheaper and you can move on faster.

[Sharkyx1]

This lawsuit is entirely frivolous and has no more merit than the Samsung curved corner law suit and the Sony 1080p Killzone lawsuit

[Tristerin]

6 minutes ago, 2Buck said:

Y'know, I take it back, even though their IPC is really bad (compared to Sandy bridge, and even phenom ii), saying they suck is a little harsh. All those AMD fanboys from back then turned out to be right, the CPUs haven't aged too bad now that games can take advantage of all the threads.

 

My FX 8320 has done me very well. I edit 1080p 60 FPS video in premiere, stream/record at 720p 60 FPS in OBS, hosted game servers while playing games (before I built dedicated game servers), and plenty of other things. This CPU has always kept up.

 

Oh, and as they get cheaper on eBay, it's gonna be cool seeing them take over socket 775's role as the "extreme budget" CPUs. Just now on eBay I found two 8320s under 60$, maybe soon they'll be 40$ and under. Hell, there's one for 40$ on there that has couple of bent pins. Before you know it, you'll be able to buy a 6300 for 20$.

 

The FX CPUs started out really bumpy, and they were far behind the competition in single threaded tasks, but overall they've turned out... Alright. Way to go Faildozer.

Agreed - I think those that knock this CPU either don't have one and are usually just regurgitating what they have read.  My only knock is the TDP.  Its usefulness in 2019 is fairly competitive.

 

I actually have the i5 3470 (4c/4t), and the FX 8350 (direct competitors timeframe wise).  I prefer gaming on the FX 8350.  It loads games faster (important in FPS competitive play) plain and simple.  Frames are nearly identical (I slapped one of my HD 7850's in it and checked it out) but loading times are not.  Both as similar environments as I could make them for the testing purposes (why I game on the FX and not the i5 and its in my living room - its also WAY lower TDP and its on more lol)

[thorhammerz]

13 minutes ago, porina said:

From memory, people did, and people won. At least, it was for some part of the US.

Yep, $30 per card (article from anandtech).

 

• Overprice your pieces of junk off all you want to (read: current state of RTX pricing)
• Force your consumers to pay extra to stay in your ecosystem (read: G-Sync for the past 5 years).
• Implement aggressive planned obsolence as business needs dictates (Apple and Nvidia are big on this).
• But if (emphasis on if) you're ever caught lying about whatever product you're selling (however good or bad it is), you're in line for some court action .

[mr moose]

It'll be interesting to see which way this goes.  I agree their marketing was crap and on that score they'll likely lose, but I also think the loss would be unfair like I thought the 970 loss was unfair.

 

 

[Sakkura]

1 hour ago, yian88 said:

You could sue nvidia for gtx 970 with 3.5gb fast GDDR5 and 512mb of slow vram, but they advertised 4GB vram gpu which the GPU does have, hence you would lose in a sue case.

More importantly, the GTX 970 had fewer enabled ROPs than they originally claimed. That was indisputably false, while the memory thing was debatable to some extent, just like the Bulldozer core thing is.

[Brooksie359]

1 hour ago, mr moose said:

It'll be interesting to see which way this goes.  I agree their marketing was crap and on that score they'll likely lose, but I also think the loss would be unfair like I thought the 970 loss was unfair.

 

 

To be fair there is a standard for ddr5 and the last .5gb wasn't really meeting that standard. 

[Arika]

I think i"m going to go with the guy who has 30 years of experience/education in the legal field

 

Awards

• 2013-2014: Criminal Defense: White Collar, Best Lawyers in America
• 2013-2014: Future Star, Benchmark Litigation
• 2008-2013: Northern California Super Lawyers
• 1998: San Francisco NAACP 1998 Thomas I. Atkins Civil Rights Award
• 1994: United States Law Week award for outstanding service and unfailing commitment to the Stanford Law Review
• 1993-1994: Article editor, Stanford Law Review
• 1992-1994: Member, Stanford Law Review
• 1992: Hilmer Oehlmann Jr. Prize for outstanding work in first-year Legal Research and Writing, Stanford Law School
• 1991: Thatcher Memorial Prize for encouragement of extemporaneous debate among undergraduates, Yale University
• 1991: Master's Cup for service to Timothy Dwight College, Yale University

 

You don't need to have a technological background to understand what is being claimed on both sides.

[yolosnail]

Suing AMD for this is like buying a double cheeseburger then suing the company because you got one cheeseburger with two burgers in rather getting two separate cheeseburgers!

[mr moose]

20 minutes ago, Brooksie359 said:

To be fair there is a standard for ddr5 and the last .5gb wasn't really meeting that standard. 

I would argue in the same context that in the customers eye there is also a standard for what constitutes a processor core.  Technically the 970 did have the full 4gb off ddr5, it's just the last .5 shared L2 cache.  The same thing will be argued here, the FX has all 8 cores, they just don't operate at the performance you would expect from a core because each core didn't have it's own FPU or L2cache, just like the last .5Gb didn't have it's own L2 and shared the L2 with it's neighboring MC.  It is almost identical in technical marketing aspects.

 

But more importantly, what I am getting at, is we can't really just claim on a technicality they are right, because others will claim on a technicality they are wrong and what it really is about is product representation to the consumer. 

[79wjd]

30 minutes ago, mr moose said:

I would argue in the same context that in the customers eye there is also a standard for what constitutes a processor core.  Technically the 970 did have the full 4gb off ddr5, it's just the last .5 shared L2 cache.  The same thing will be argued here, the FX has all 8 cores, they just don't operate at the performance you would expect from a core because each core didn't have it's own FPU or L2cache, just like the last .5Gb didn't have it's own L2 and shared the L2 with it's neighboring MC.  It is almost identical in technical marketing aspects.

 

But more importantly, what I am getting at, is we can't really just claim on a technicality they are right, because others will claim on a technicality they are wrong and what it really is about is product representation to the consumer. 

As an aside, if you take the weak core approach, then Intel is also guilty of it with their quad core ultra cheap Pentium/Celerons that performs nowhere near the traditional Pentium/Celeron lineups.

[Curufinwe_wins]

31 minutes ago, mr moose said:

I would argue in the same context that in the customers eye there is also a standard for what constitutes a processor core.  Technically the 970 did have the full 4gb off ddr5, it's just the last .5 shared L2 cache.  The same thing will be argued here, the FX has all 8 cores, they just don't operate at the performance you would expect from a core because each core didn't have it's own FPU or L2cache, just like the last .5Gb didn't have it's own L2 and shared the L2 with it's neighboring MC.  It is almost identical in technical marketing aspects.

 

But more importantly, what I am getting at, is we can't really just claim on a technicality they are right, because others will claim on a technicality they are wrong and what it really is about is product representation to the consumer. 

Also... 4GB of VRAM was not THE selling point of the product. Like the 8 cores at a reasonable price was THE selling point AMD spread everywhere for this product.

 

Nvidia advertised directly performance gains over previous products with the 970 (real performance gains that actually existed and were not faked), AMD basically did its best to ignore the performance of bulldozer in its marketing for the cpu (probably because the performance was downright bad).

 

So yes, if that lawsuit resulted in settlement, I very much expect this one to as well.

[mr moose]

Just now, Curufinwe_wins said:

Also... 4GB of VRAM was not THE selling point of the product. Like the 8 cores at a reasonable price was THE selling point AMD spread everywhere for this product.

 

Nvidia advertised directly performance gains over previous products with the 970 (real performance gains that actually existed and were not faked), AMD basically did it's best to ignore the performance of bulldozer in its marketing for the cpu (probably because the performance was downright bad).

 

So yes, if that lawsuit resulted in settlement, I very much expect this one to as well.

This is actually very true and will likely have a bigger impact on the outcome than people think.

[The Benjamins]

3 hours ago, yian88 said:

Not trying to defend, but AMD did not cheat on core count and hyperthreading is a completely different thing.

AMD had shared resources for 2 cores making 1 module, hyperthreading is software split up of a single core resources, hyperthreads are not physical.

Yes you do have less performance per core in 1 module because of shared resources in many scenarios that became the bottleneck but you do have 8 physical cores in FX 8000 series etc.

AMD didnt lie they had their designs public and i do remember people writing articles about the shared resources reducing performance.

I wonder who sued them i doubt regular users did, is it a "rival" or someone who has beef with AMD behind this.

You could sue nvidia for gtx 970 with 3.5gb fast GDDR5 and 512mb of slow vram, but they advertised 4GB vram gpu which the GPU does have, hence you would lose in a sue case.

 

There was a 970 lawsuit and it was settled, each person that bought the 970 got around $30 if I remember correctly.

[Ja Krispie]

7 hours ago, syn2112 said:

good, AMD should feel bad for that abomination of a processor.

It really was. I bought the FX6350 thinking (OK cool. I have a 6 core CPU and I can pair it with my GPU and be good!)

 

Little did I know at the time the CPU was garbage and I quickly upgraded to an i5. Been happy with that i5 until Ryzen released

[Jito463]

6 hours ago, Curufinwe_wins said:

For example, I had friends as recently as 2016 who still believed (directly as a result of AMD's marketing materials and claims) that the 8350 was a far better 8-core cpu than the 5820k 6-core.

I'm not aware of any time that AMD directly compared their FX-8350 to the i7-5820k.  Care to share?