AMD Bulldozer 'Core' Lawsuit: AMD Settlement

That's nothing, think AMD made the right decision. Fighting this in California would've been much more expensive.
 
Well, Bulldozer is effectively an 8 core CPU, but justice is not about the truth but money. AMD never pretended it was the best 8 core ever. I suppose some lawyers made plenty of money out of nothing worth it.
 
I bought all three Bulldozers from Newegg which is in California. BS lawsuit won’t bother.
 
Leave it to California...
Cheaper to pay than fight, our court systems at work, lol.
So... What is the definition of a core then? Must have an ALU and FPU? What about sse/mmx/avx... If they share an avx unit can they not be called individual cores? Just because it was slower than a 4 core doesn't mean it's not 8 cores, it just means the product wasn't great. You can get a 4 core ARM CPU that gets beat by a single core CPU.. it didn't make it not 4 cores, just means you didn't research and buy the right part.
 
damn. I bought a few of these back in the day, but I'm not from California nor did I buy them from AMD site.
 
there's probably some language somewhere on some marketing material that led to this ....not the argument of whether or not 8 core was 8 core because it did or did not share a component of what makes up a processor's execution pipeline.

Considering that FPU's were considered completely separate processors up until the 486, I'm totally on board with the argument that fpu's can be removed from the definition of a processor core.

Now can we finally get a class action lawsuit against disk storage manufacturers for redefining words to make their capacities inflated? A terabyte is not one thousand gigabytes and a gigabyte is not one thousand megabytes. A megabyte is not one thousand kilobytes, and a kilobyte is not 1000 bytes. Bytes are measured in base 2 because they measure bits which are binary. It's not hard, every new metric is 2^10 of the last (1024). This is far more intentionally misleading and far more impactful false advertising than what AMD could be considered to have done.
 
Darth Ender said:
there's probably some language somewhere on some marketing material that led to this ....not the argument of whether or not 8 core was 8 core because it did or did not share a component of what makes up a processor's execution pipeline.

Considering that FPU's were considered completely separate processors up until the 486, I'm totally on board with the argument that fpu's can be removed from the definition of a processor core.

Now can we finally get a class action lawsuit against disk storage manufacturers for redefining words to make their capacities inflated? A terabyte is not one thousand gigabytes and a gigabyte is not one thousand megabytes. A megabyte is not one thousand kilobytes, and a kilobyte is not 1000 bytes. Bytes are measured in base 2 because they measure bits which are binary. It's not hard, every new metric is 2^10 of the last (1024). This is far more intentionally misleading and far more impactful false advertising than what AMD could be considered to have done.
Click to expand...

AMDs marketing was pretty clear how the CPU was layed out, and it's not like they didn't have benchmarks to look at before buying. The claim in the lawsuit was based on AMD calling them cores, since they shared an fpu (which could do 2 simultaneous 128 bit calcs at the same time) and some of the decoding hardware (which IS a more valid argument, but still not really). I'm not sure, but the fact that it only covers 8 core (not 6 that are the same layout) seems very odd.
 
maybe has to do with the fact that the 8 core was the "world's first" marketing wise ...perhaps 6 wasn't.
 
Now can we finally get a class action lawsuit against disk storage manufacturers
Click to expand...

and then gas fuel, right....I mean 9/10? Come on!!!

Honestly, though, they were upfront with the processor, and I feel the AMD's argument is legit. But masses of asses will win, eh? I had no issues with understanding the µ architecture of this processor. If this is the case then there should be a mass of lawsuits with so many other issues, especially with Intel. Anyone liked their loop stream detector disabled with Skylake? How about security issues (this one is still in the process)? Shit man, I can possibly sue for everything in my system!
 
Last edited by a moderator:
Darth Ender said:
Now can we finally get a class action lawsuit against disk storage manufacturers for redefining words to make their capacities inflated? A terabyte is not one thousand gigabytes and a gigabyte is not one thousand megabytes. A megabyte is not one thousand kilobytes, and a kilobyte is not 1000 bytes. Bytes are measured in base 2 because they measure bits which are binary. It's not hard, every new metric is 2^10 of the last (1024). This is far more intentionally misleading and far more impactful false advertising than what AMD could be considered to have done.
Click to expand...
Umm I thought this has to do with bits vr bytes and how the OS see's space or something?
 
dany man said:
Umm I thought this has to do with bits vr bytes and how the OS see's space or something?
Click to expand...


we have a word for the decimal representation of bits ... it's called kilobits, megabits, gigabits, terabits, etc. This is often used to measure transferrate / bandwidth. Not size since bits are stored in bytes and bytes are base 2.

It's a difference in storage vs transfer. Storage has always been represented as bytes and bytes have always been represented as forms of base 2 numbers because of the physical way it's stored. software represents it correctly, only storage manufacturers use their own stupid definition.

Storage manufacturers had their marketing teams decide that numbers that dont end in zero's are too complicated for the public and they could say they were XYZ Megabytes when in fact they were less if they just redefined what a megabyte was. So they did.

This is fully and intentionally misleading because a consumer who is told by software, correctly, that they need say 1GB for a certain application and that consumer goes out to buy a 1GB drive will think the drive is large enough but in fact, they're buying a 0.93GB drive.

Later on new terms were created for the traditional <term>byte names to represent the correct way to use bytes since nobody could get storage companies to correct their definition. This is called the kibibyte, mebibyte, gibibyte, tebibyte. But it takes a long time for these things to get adopted by software and society as a whole.

We technically should never refer to basically anything as megabyte or terabyte since it's only a term used by hdd manufacturers to sell drives. Everywhere else uses the measurement that would refer to the new ebi and ibi terms.
 
Darth Ender said:
maybe has to do with the fact that the 8 core was the "world's first" marketing wise ...perhaps 6 wasn't.
Click to expand...
Doesn't matter if they claimed it was first or last... if they say it wasn't 8 cores... then it wasn't 6 cores either, whether it was first or not. The thing is, 'core' is not a standard term, but can mean slightly different things. HT (hyper threading) shares an ALU, so this isn't just an HT vs. core... it does a good amount more than just hold a thread as it can process operations simultaneously with the other cores. It's shown that each core (this is in synthetics) past 4 adds about 80% (average some apps better/worse) performance (so not 100% extra core performance, but much more than a HT). They are just upset that the CPU sucks. They didn't do their due diligence when they purchased as non of this information was kept a secret. Like I said, just because something has more cores, doesn't automatically make it better. There are plenty of systems with high core counts that aren't fast. If I bought a 24-core ARM chip (https://www.microcontrollertips.com...egrates-24-arm-cortextm-a53-cores-single-die/) and expected it to outrun a 16 core Ryzen, I'd be sorely disapointed. That doesn't mean it doesn't have cores, it just means I suck at understanding what I'm buying and should probably have asked someone who had a clue. Can you imagine someone suing GM because their engine cuts from 8 - 4 cylinder mode during cruise because it's not ALWAYS an 8 cylinder? Or sue ford because all cylinders used to share a carburetor therefore they weren't independent and where fed from a common source? See has silly this sounds. I'm sure AMD didn't want this to drag out and try to explain this stuff to a Jury of who knows how much tech they would understand. Cheaper to just pay and be done.
If it's not a core, and it's not HT... then should they have made up another name for it instead of saying it was a core? I don't know if that would have led to less confusion or more.
 
AMD Bulldozer's new feature: Hyperspreading! Full core throttle speed for you dear consumer! With hyperspreading you can work ten fold faster utilizing spread core technologies! Very fast!
 
Ready4Dis said:
Doesn't matter if they claimed it was first or last... if they say it wasn't 8 cores... then it wasn't 6 cores either, whether it was first or not. The thing is, 'core' is not a standard term, but can mean slightly different things. HT (hyper threading) shares an ALU, so this isn't just an HT vs. core... it does a good amount more than just hold a thread as it can process operations simultaneously with the other cores. It's shown that each core (this is in synthetics) past 4 adds about 80% (average some apps better/worse) performance (so not 100% extra core performance, but much more than a HT). They are just upset that the CPU sucks. They didn't do their due diligence when they purchased as non of this information was kept a secret. Like I said, just because something has more cores, doesn't automatically make it better. There are plenty of systems with high core counts that aren't fast. If I bought a 24-core ARM chip (https://www.microcontrollertips.com...egrates-24-arm-cortextm-a53-cores-single-die/) and expected it to outrun a 16 core Ryzen, I'd be sorely disapointed. That doesn't mean it doesn't have cores, it just means I suck at understanding what I'm buying and should probably have asked someone who had a clue. Can you imagine someone suing GM because their engine cuts from 8 - 4 cylinder mode during cruise because it's not ALWAYS an 8 cylinder? Or sue ford because all cylinders used to share a carburetor therefore they weren't independent and where fed from a common source? See has silly this sounds. I'm sure AMD didn't want this to drag out and try to explain this stuff to a Jury of who knows how much tech they would understand. Cheaper to just pay and be done.
If it's not a core, and it's not HT... then should they have made up another name for it instead of saying it was a core? I don't know if that would have led to less confusion or more.
Click to expand...
I can't imagine any user or group of being so disgruntled over this as to dedicate so many man hours for its cause. One would have to admit that they are stupid for doing no research on the matter when there were bountiful reviews from day one.
This is lawyers being lawyers. They will make a lot of money and aren't really that concerned how many get their $35.
 
funkydmunky said:
I can't imagine any user or group of being so disgruntled over this as to dedicate so many man hours for its cause. One would have to admit that they are stupid for doing no research on the matter when there were bountiful reviews from day one.
This is lawyers being lawyers. They will make a lot of money and aren't really that concerned how many get their $35.
Click to expand...
Exactly, who feels like they were robbed $35? There is a reason an 8 core top end bulldozer was selling for less than 1/2 the price of Intel CPUs.... Who expected it to run faster? It's $/perf was in line, it was just a dud of a CPU/architecture so they sold it cheaper.
 
Ready4Dis said:
Exactly, who feels like they were robbed $35? There is a reason an 8 core top end bulldozer was selling for less than 1/2 the price of Intel CPUs.... Who expected it to run faster? It's $/perf was in line, it was just a dud of a CPU/architecture so they sold it cheaper.
Click to expand...

The lawsuit seems completely stupid to me. Legal masturbation.

I've been involved in IT and Systems Engineering for 30 years. Core counts have always been based on the integer core. The majority of the instructions your computer does are integer calculations.

This is why they used to sell systems with an optional discrete FPU.
 
Spirit_Retro said:
The lawsuit seems completely stupid to me. Legal masturbation.

I've been involved in IT and Systems Engineering for 30 years. Core counts have always been based on the integer core. The majority of the instructions your computer does are integer calculations.

This is why they used to sell systems with an optional discrete FPU.
Click to expand...
I he thing is, even the "shared" CPU could do two instructions at the same time... It was a double wide FPU, so as long as the app was using single precision (most apps), the cores could both perform FPU ops simultaneously as well.. IF but the cores where using double precision at the same time, then it would have to share. Im not sure the scheduler would be smart enough to keep threads that used double precision separated.
 
A scheduler wouldn't middle man opcodes being run by a process. That would be incredibly slow. There would need to be some kind of 'exclusive core' flag the process would need to set that the scheduler would have to be aware of. I don't think such a thing exists on Windows or Linux.
 
Darth Ender said:
A scheduler wouldn't middle man opcodes being run by a process. That would be incredibly slow. There would need to be some kind of 'exclusive core' flag the process would need to set that the scheduler would have to be aware of. I don't think such a thing exists on Windows or Linux.
Click to expand...
I mean just like modern computers have a flag to say if it uses the FPU so it knows whether it needs to store/load the fpu registers on task switch, if it had one for double precision tasks it could keep them from running on the same shared FPU (keep them separate) so you could have up to 4 double precision tasks running that aren't sharing or waiting. If they are on the same shared group, they would have to wait for one another. It can run 8 things (including single precision floating point) at the same time... This I would consider 8 cores with some shared resources. I didn't mean for th scheduler to middle man anything, just be aware it's using it so it can intelligently run threads/tasks... Kind of like the latest updates (I think in 1903) they Microsoft finally made it smart enough to not switch ccx on AMD cpus. (Linux had this solved during gen 1 zen).

edit: Some spelling fixes, didn't realize how much my phone tries to auto correct.
 
Last edited:
I must say, I did not see this coming. Disagree with the decision. Bulldozer had 8 cores - 8 crappy cores, yes, that shared an FPU... but still 8 cores.

Oh well, gotta be wrong some of the time, I suppose.
 
It doesn't mean you're wrong. It just means that wrong decisions are made in the justice system.
 
It wasn't a decision in the justice system, it was an agreement made to not have to bring it to the justice system (at least that's the impression I got).
 
Can't wait to get my 27 cents!

nc_oc=AQnNL3xdp73BNRkGs237ICwaLDAI7GV2666xnecrdgf1DpHdBjKg_8YqSWCwVHHM4Bg&_nc_ht=scontent-ort2-2.jpg
 
I should probably nuke the thread I just opened. I was looking for this one...

https://www.amdcpusettlement.com/

AMD already lost the money, now it's time for you to jump on the bandwagon and claim your pound of flesh. They are refunding up to 300 dollars per processor you purchased.
 
I only ever got Bulldozer's cousins (the APUs), so no monies fir me.
 
Meanwhile the attorney will get a nice 33% of that settlement and in the end each person that bought a chip, if they claim it, or even know they can get a whopping $35.00
 
Legendary Gamer said:
I should probably nuke the thread I just opened. I was looking for this one...

https://www.amdcpusettlement.com/

AMD already lost the money, now it's time for you to jump on the bandwagon and claim your pound of flesh. They are refunding up to 300 dollars per processor you purchased.
Click to expand...

Agreed.

It was a stupid frivolous lawsuit. A CPU core is defined as the integer core, and AMD's FX line had up to 8 of them, and this is a damned fact.

They were absolutely terrible cores, but they were real cores.

All that said, failing to file a claim if you are eligible will not give AMD their money back, so file away.
 
pendragon1 said:
i knew what i was getting when i bought them.
Click to expand...

That's not really the point, which is the settlement has already been finalized. If you don't get your money, the lawyers will get it instead. It won't be refunded to AMD if that's what yer worried about.
 
thesmokingman said:
That's not really the point, which is the settlement has already been finalized. If you don't get your money, the lawyers will get it instead. It won't be refunded to AMD if that's what yer worried about.
Click to expand...
usa only though, isnt it?
 
Zarathustra[H] said:
Agreed.

It was a stupid frivolous lawsuit. A CPU core is defined as the integer core, and AMD's FX line had up to 8 of them, and this is a damned fact.

They were absolutely terrible cores, but they were real cores.

All that said, failing to file a claim if you are eligible will not give AMD their money back, so file away.
Click to expand...
I knew what I was getting into when I bought mine. It was a great processor for none of the things I used it for. I turned it into a file server eventually and it worked admirably in that role until I donated it to the Boys and Girls Clubs. It ended up making some kid really damn happy.

I too thought this lawsuit was utter horseshit.
 
Legendary Gamer said:
I knew what I was getting into when I bought mine. It was a great processor for none of the things I used it for. I turned it into a file server eventually and it worked admirably in that role until I donated it to the Boys and Girls Clubs. It ended up making some kid really damn happy.

I too thought this lawsuit was utter horseshit.
Click to expand...

I'm curious. What are the advantages of re-purposing an x86 desktop as a file server instead of Just using a NAS, Rasberry Pi, or plugging an ext storage drive into the router?
 
penn919 said:
I'm curious. What are the advantages of re-purposing an x86 desktop as a file server instead of Just using a NAS, Rasberry Pi, or plugging an ext storage drive into the router?
Click to expand...

Capacity? Performance? Flexibility to set things up the way you want it?

I tend to always try to do things with a real server abd avoid appliances as much as possible, including even more Enterprise oriented brands like QNAP.

Whenever I can use server hardware, I do. Even my router is running on PC hardware with using pfSense.

Also, IMHO, it sounds like a really bad idea to out your storage on the same device that is your WAN bridge and firewall.

My storage server has 2x 8c/16t xeons 256GB of RAM, 12x 10TB hard drives, 8x SSD's in various caching etc. purposes, and dual 10gig Ethernet adapters.

I can probably get about 2GB/s reads off of the hard drive storage array, and it supports a lot of stuff I do in my home.

Can't do that with a raspberry pi. :p
 
Zarathustra[H] said:
Capacity? Performance? Flexibility to set things up the way you want it?

I tend to always try to do things with a real server abd avoid appliances as much as possible, including even more Enterprise oriented brands like QNAP.

Whenever I can use server hardware, I do. Even my router is running on PC hardware with using pfSense.

Also, IMHO, it sounds like a really bad idea to out your storage on the same device that is your WAN bridge and firewall.

My storage server has 2x 8c/16t xeons 256GB of RAM, 12x 10TB hard drives, 8x SSD's in various caching etc. purposes, and dual 10gig Ethernet adapters.

I can probably get about 2GB/s reads off of the hard drive storage array, and it supports a lot of stuff I do in my home.

Can't do that with a raspberry pi. :p
Click to expand...
Basically this. X86 offers plenty of options for different usecases, in efficient sff designs all the way up to full server hardware (which is still efficient if you configure it properly and fully utilize it).

Nothing wrong with rpi or other arm solutions, they're fine too. But you probably wouldn't want tarabytes of raid storage on a rpi nas with just 1gbps serving multiple (3+) clients concurrently, for example. NAS boxes are expensive, so if you have a spare system with lots of sata ports and decent network hardware (a couple gbit ports, 10gbit, or room to add a 10gbit or dual 1gbit nic), why not use it?
 
You must log in or register to reply here.
Back
Top