AMD in court ?

[Kenrou]

http://www.techpowerup.com/217327/amd-dragged-to-court-over-core-count-on-bulldozer.html

 

[EarthDog]

LOL, really? That's a shame... though I understand why people are doing it. I thought this whole module and Floating Point thing or w/e it was based off was just left to die (as it should...).

 

[Johan45]

All I can say Waaaa of the Day

What was more interesting was right under that. AMD has tested , I assume physical samples and it all seem on track. Went just as expected.

http://www.overclock3d.net/articles...ation_with_no_significant_bottlenecks_found/1

 

[Janus67]

All I can say Waaaa of the Day

What was more interesting was right under that. AMD has tested , I assume physical samples and it all seem on track. Went just as expected.

http://www.overclock3d.net/articles...ation_with_no_significant_bottlenecks_found/1

That's a pretty petty lawsuit, but if it is correct in that AMD isn't able to process 8 threads at once it isn't incorrect. Just stupid that a lawsuit is necessary to make marketing/naming schemes of anything accurate. Also strange that it is just coming out now since the CPUs have been out for years.

 

[Mandrake4565]

The guy is representing himself, he's just hoping AMD will pay him some cash and go away.

All I can say Waaaa of the Day

What was more interesting was right under that. AMD has tested , I assume physical samples and it all seem on track. Went just as expected.

http://www.overclock3d.net/articles...ation_with_no_significant_bottlenecks_found/1

Saw this as well, it sure will be nice to see a new chip from them. Here is hoping it's somewhat decent. It doesn't matter to me how good bad or indifferent it is, as long as I can Oc it.

 

[PetrolHead]

Threads are a misleading term here, I think. You can run thousands of threads on a single core. Just check your task manager, we're all doing it now. The key is parallel computation. AMD's octa-cores are not capable of running eight parallel tasks regardless of the task as a true octa-core should. Instead, in the case of floating point operations they can only do four parallel tasks. There are eight integer clusters, which should enable performing eight parallel integer operations, but to call the CPU an octa-core is not really true. Sure, it's closer to being an octa core than an Intel CPU with four cores and hyper threading (which just allows for a bit more efficient serial computing on a single core) is, but it's still not enough - and Intel hasn't been brave enough to be "optimistic" about the core count of its processors.

 

[Bluefalcon13]

I do find it slightly misleading, but at the same time, I know its functions and limitations. Does the definition of an x86 (and by extension x86_64) core actually include a dedicated FP unit? From what I remember, back in the day, FP units were a co-processor....

 

[PetrolHead]

Well, there was the match co-processor, but I'm not sure if that's exactly the same thing as you could buy a CPU without one (the 486SX/486SX2 CPUs). But I guess back then the whole microarchitecture was a bit simpler than nowadays. I remember a friend of mine having a Compaq desktop computer with a 66MHz 486SX2 CPU. At that time a 66MHz 486DX was considered a very capable processor, but the lack of a math co-processor made that 486SX2 very slow in certain games.

Btw, it should be said that using the term octa-core is not the only "crime" in CPU marketing these days. For example, this is how Intel advertises its 6th generation CPUs:

"Game Like a Pro

With 6th generation Intel® Core™ processor, play your favorite 3D games in higher resolution, better details, faster frame rates, giving you an amazing gaming experience on your PC.

Up to 30x better 3D graphics(4)

(4)Measured using 3DMark* 1.2.0 Cloud Gate for DX 10 graphics subtest - a 3D graphics benchmark that measures 3D gaming performance. Find out more at www.futuremark.com."

Up to 30x better!!!

 

[Bluefalcon13]

Well, there was the match co-processor, but I'm not sure if that's exactly the same thing as you could buy a CPU without one (the 486SX/486SX2 CPUs). But I guess back then the whole microarchitecture was a bit simpler than nowadays. I remember a friend of mine having a Compaq desktop computer with a 66MHz 486SX2 CPU. At that time a 66MHz 486DX was considered a very capable processor, but the lack of a math co-processor made that 486SX2 very slow in certain games.

Btw, it should be said that using the term octa-core is not the only "crime" in CPU marketing these days. For example, this is how Intel advertises its 6th generation CPUs:

"Game Like a Pro

With 6th generation Intel® Core™ processor, play your favorite 3D games in higher resolution, better details, faster frame rates, giving you an amazing gaming experience on your PC.

Up to 30x better 3D graphics(4)

(4)Measured using 3DMark* 1.2.0 Cloud Gate for DX 10 graphics subtest - a 3D graphics benchmark that measures 3D gaming performance. Find out more at www.futuremark.com."

Up to 30x better!!!

Yeah but the SX was still considered a processor or what we would call a core these days, just sans FPU.

As for the Intel marketing thing... That's in legal-ese. And 100% correct at that. Up to (but not always) 30x performance increase. What they don't say is that at x resolution your frame rate will be this. Which the lawsuit is essentially saying "I was promised an 8-core CPU, and in these conditions I don't have an 8 core CPU."

Ultimately this guy is using the court system to define what a CPU core is, while looking for money. While I personally feel that a core needs a better definition, I don't feel the court system is where you should get that definition.

 

[PetrolHead]

As for the Intel marketing thing... That's in legal-ese. And 100% correct at that. Up to (but not always) 30x performance increase.

Yes and no. The text your average Joe will read says "better 3D graphics" and that's where they will likely stop. Only if you read the fine print found at the bottom of the page will you find out that "better 3D graphics" does not mean actually better 3D graphics, but better 3D gaming performance - two very different things. After all, the benchmark they use has the same graphics regardless of what you run it on. They also forget to mention what they've used for comparison. A 5th generation Intel processor? An Intel Atom? Sure, they might get away with it by arguing that and increase in FPS can allow for eg. higher resolution in games and while the 30x is in this context nonsense, it does say "up to", but I still wouldn't call it an honest description of the processor's capabilities.

The point is, AMD is not the only one guilty of marketing spin.

Ultimately this guy is using the court system to define what a CPU core is, while looking for money. While I personally feel that a core needs a better definition, I don't feel the court system is where you should get that definition.

I agree.

 

[EarthDog]

I ear ya, but that is really being literal here with how you are taking the 'better 3d graphics' statement.

While I agree both do it, I don't see any issues with those particular examples. In fact, it's pretty clear to me that they didn't mean image quality, but performance. It's a shame it isn't more clear, but there is a footnote reference that shows where it came from. I'm good with that.

 

[Sologuy]

I think we will find is how the court defines "the core of an X86 cpu" - what exactly is this?? this will be the deciding factor imo..