• Guest, The rules for the P & N subforum have been updated to prohibit "ad hominem" or personal attacks against other posters. See the full details in the post "Politics and News Rules & Guidelines."

News [Register] AMD agrees to cough up $35-a-chip payout over eight-core Bulldozer advertising fiasco

Page 2 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

gdansk

Senior member
Feb 8, 2011
641
330
136
I never bought a Bulldozer style computer because I looked at benchmarks. But I'm curious how they reckon it's not 8 cores. There are entire architectures where it is common to not even have a wide FPU. And others where they don't have an FPU at all. And yet they are still clearly CPU cores. If anyone ever differs too much from Intel-style big cores with ultra-wide SIMD FPUs, are they vulnerable to lawsuit too?

I recall Sun shipped some similar CMT designs, the T2 at least. Are they (now Oracle) going to be sued next?
 

ondma

Platinum Member
Mar 18, 2018
2,166
815
106
Here in my country the "10 computes cores" was used to lie to buyers saying the PC had 10 cores. And they used "its written on the box" as a excuse. AMD knew this would happen when they used that denomination and is the reason of why is not used anymore.

For the FX thing, is not a quad core / octa core for EVERY TASK, it is for a specific kind of load only, this is why it cant be considered as one.
I never really had a problem with advertising the FX as eight cores. More shady, IMO, was advertising a quad core APU as having 10 or 12 compute cores. Technically it was correct I suppose, but not in the way people generally think of for cpus.

Edit: I think this settlement is pretty ridiculous.
 
  • Like
Reactions: scannall

ElFenix

Elite Member
Super Moderator
Mar 20, 2000
101,399
5,438
126
Let's be real, you just want $105 towards your next project, am I right? :p I wouldn't blame you, I got the $15 from Intel years ago. Anyone could submit those though. Here you have to have bought it from AMD or live in CA, interesting to see if that stands. And I did not lie on the Intel one, I did buy a P4, and an FX as well.
why wouldn't it stand?
 

ElFenix

Elite Member
Super Moderator
Mar 20, 2000
101,399
5,438
126
I've read speculation that it will be expanded to anyone who bought one, not just from AMD or someone that lives in CA. Kind of stupid to limit the criteria.
they'd have to renotice and recertify the class. probably not able to do that at this stage
 
Feb 4, 2009
29,935
10,478
136
I've read speculation that it will be expanded to anyone who bought one, not just from AMD or someone that lives in CA. Kind of stupid to limit the criteria.
So I might get money back for my utterly frustrating FX6300 purchase from 5 years ago?
 

NostaSeronx

Diamond Member
Sep 18, 2011
3,263
799
136
It's similar in that they share FPU. I looked it up and one single FPU was shared among 8 cores on T1, not T2 like I thought.
AMD's CMT involves the front-end, L2 interface, and the FPU. The T1 processor only shared the FPU execution units.

The closet Sun got to AMD's CMT is with rock;
-----


This design enables resource sharing among cores within a cluster, thus reducing the area requirements. All cores in a cluster share an instruction fetch unit (IFU) that includes the level-one (L1) instruction cache. We decided that all four cores in a cluster should share the IFU because it is relatively simple to fetch a large number of instructions per cycle. Thus, four cores can share one IFU in a round-robin fashion while maintaining full fetch bandwidth. Furthermore, the shared instruction cache enables constructive sharing of common code, as is encountered in shared libraries and operating system routines.

Each core cluster contains two L1 data caches (D cache) and two floating-point units (FGU), each of which is shared by a pair of cores, these structures are relatively large, so sharing them provides significant area savings.
----
 
Last edited:

gdansk

Senior member
Feb 8, 2011
641
330
136
AMD's CMT involves the front-end, L2 interface, and the FPU. The T1 processor only shared the FPU execution units.

The closet Sun got to AMD's CMT is with rock.
Like I said, it is similar but not identical. A much faster FPU being shared among two cores was enough to cause AMD to settle a $12 million lawsuit. T1's slower FPU was being shared among eight cores. And if you look at the T1 diagram you can see they share the L2 cache as well. Regardless, I didn't say it was identical, but some of Sun's CMT is similar to Bulldozer's CMT in that integer cores do not have a corresponding floating point unit. The Register said that was a crucial point:
The Register said:
In AMD’s mind, four modules times two CPU cores equals eight CPU cores. But to the angry consumers, who launched a class action lawsuit back in 2015, they are not real “cores” because they share resources, including frontend circuitry and – crucially in this case – a single floating point unit (FPU).
 
Last edited:

NostaSeronx

Diamond Member
Sep 18, 2011
3,263
799
136
A much faster FPU being shared among two cores was enough to cause AMD to settle a $12 million lawsuit.
There is slightly some more difference than that. T1's eight-cores wasn't optimized for the FPU. However, Bulldozer's FPU was optimized for dual-core workloads. It is just consumers being brain-dead as usual.
 

gdansk

Senior member
Feb 8, 2011
641
330
136
There is slightly some more difference than that. T1 eight-cores wasn't optimized for the FPU. However, Bulldozer's FPU was optimized for dual-core workloads. It is just consumers being brain-dead as usual.
Yes. I suppose this is the risk of selling to uninformed than selling to people who know what they actually need. People buying SPARC chips knew what they were getting. Hopefully this will incline AMD toward publishing more diverse benchmarks going forward so that consumer's cannot claim they were mislead.
 

NostaSeronx

Diamond Member
Sep 18, 2011
3,263
799
136
Hopefully this will incline AMD toward publishing more diverse benchmarks going forward so that consumer's cannot claim they were mislead.
That is a stupid consideration. There is no benchmarks that would prove that there aren't two cores.

AMD Zen1/Zen2 have an independent FPU pipeline. Intel Skylake/Sunnycove have an integrated FPU pipeline. Clearly not even AMD's Zen is a true core as the cores don't handle FPU instructions. AMD has only sold zero-core processors clearly. It doesn't include modern features like a fully integrated floating-point pipeline, not a core.

I bought a Cortex-A76 without a FPU, I have a zero-core processor. I went on SiFive, I didn't want the FPU on my processor, I am now stuck with zero-cores. All my modern Cortex M0, M0+, M1 products, no cores.
 
Last edited:

gdansk

Senior member
Feb 8, 2011
641
330
136
That is a stupid consideration.

AMD Zen1/Zen2 have an independent FPU pipeline. Intel Skylake/Sunnycove have an integrated FPU pipeline. Clearly not even AMD's Zen is a true core as the cores don't handle FPU instructions. AMD has only sold zero-core processors clearly. It doesn't include modern features like a fully integrated floating-point pipeline, not a core.

I bought a Cortex-A76 without a FPU, I have a zero-core processor. I went on SiFive, I didn't want the FPU on my processor, I am now stuck with zero-cores. All my modern Cortex M0, M0+, M1 products, no cores.
No idea what you're on about. There's no "stupidity" in AMD including SPECfp in their launch day materials. No surprise they have since Zen.

Apparently they rather settle than go to a judgement, which implies going forward it will in fact reduce their risk of future frivolous lawsuits. Most of the industry would agree that they are CPU cores and that a CPU core makes no claim about performance. But it doesn't matter when it comes to consumer lawsuits. AMD's marketing should be defensive going forward.
 

gdansk

Senior member
Feb 8, 2011
641
330
136
Benchmarks don't prove core counts.
No, it doesn't! But it does prove the performance so that any consumer claiming to be mislead has no basis whatsoever in court. I suppose they thought since it had twice as many cores it would be twice as fast, but with a geometric mean included in their marketing slides they can say it was clear from day 1 that it would not be the case.
 
Last edited:

ElFenix

Elite Member
Super Moderator
Mar 20, 2000
101,399
5,438
126
There is slightly some more difference than that. T1's eight-cores wasn't optimized for the FPU. However, Bulldozer's FPU was optimized for dual-core workloads. It is just consumers being brain-dead as usual.
you guys realize this guy was likely a straw plaintiff and the real impetus behind this is a law firm using california's truth in advertising laws to shakedown AMD, right? wouldn't be shocked if it's the same lawyers that sued the harddrive industry over gigabyte vs. gibibyte.
 
Feb 4, 2009
29,935
10,478
136
you guys realize this guy was likely a straw plaintiff and the real impetus behind this is a law firm using california's truth in advertising laws to shakedown AMD, right? wouldn't be shocked if it's the same lawyers that sued the harddrive industry over gigabyte vs. gibibyte.
Well of course, they are the ones approving 1/3rd the settlement money as payout money. How generous.
We already know who gets the other 2/3rds
 

piesquared

Golden Member
Oct 16, 2006
1,651
473
136
It say that the tech people have lost any form of power or credbility in their own domain, nowadays it is lawyers and other accountants who decide what is a core out of a uarch, no actual knowledge is required other than spin capabilities around vaguely technical IT press articles a la WCCFTech...
+1 for best comment in the thread.
 

Yeroon

Member
Mar 19, 2017
123
56
71
Well of course, they are the ones approving 1/3rd the settlement money as payout money. How generous.
We already know who gets the other 2/3rds
Maybe some of you guys should try and start a class action about being ripped off by class action lawyers.
Clearly this BD 8 core class action was a lawyer money grab, to the point where AMD just decided it was cheaper to settle than to keep paying the scumbags - I mean lawyers.
 
  • Like
Reactions: Thunder 57
Feb 4, 2009
29,935
10,478
136
Actually, the leeches lawyers are only taking 30% in this case.
Okay that better.
I’m still super pissed about the crappy FX6300 system I built. Admittedly I went cheap on most of the parts because it’s a Facebook machine. Damn thing had numerous failure, so many it ultimately cost me more than putting together a proper i3 system.
I’ll never again build a machine for anyone but myself. I broke the cardinal rule.
Would have been a tiny ray of sunshine if I go $30 toward a new amd chip but the legacy of FX6300 lives on.
 

Thunder 57

Golden Member
Aug 19, 2007
1,637
1,667
136
Okay that better.
I’m still super pissed about the crappy FX6300 system I built. Admittedly I went cheap on most of the parts because it’s a Facebook machine. Damn thing had numerous failure, so many it ultimately cost me more than putting together a proper i3 system.
I’ll never again build a machine for anyone but myself. I broke the cardinal rule.
Would have been a tiny ray of sunshine if I go $30 toward a new amd chip but the legacy of FX6300 lives on.
Yea, it's super crappy it's limited to the eight "core" models. And yes, cardinal rule indeed. Never do that unless you want to be free tech support for the life of that build. Last time I did that was 6-7 years ago for my brother, but that actually worked out very well. Never any problems thankfully. He's fairly computer competent though.
 

ASK THE COMMUNITY