AMD CPUs, SoC Rumors and Speculations Temp. thread 2

[jimmysmitty]

Yeah, it was a bad marketing choice however you slice it.

'Intel quad core faster than amd 8 core' has been the message ever since it released, when really the module concept is loosely comparable to an Intel core with ht (certainly similar in terms of size, number of transistors and so on).

Minus the fact that in most consumer cases Intels quad core i5s were faster than AMDs "8" cores and in gaming the i3 which is 2c/4t is equal to and sometimes faster.

That is why I always tried to get people to understand the module approach instead of just live by the marketing.

That is one areas where AMD has never done will, marketing. Intel has always done well and never seems to mis-market their products. AMD really needs good marketing. I mean look at Apple. Their products are now just over priced PCs and most of what used to be better to do on a Mac is now just as doable on a PC for less but their marketing department always seems to be able to convince people they are "better".

 

[Guest]

http://www.anandtech.com/show/9764/globalfoundries-and-amd-announce-first-14nm-finfet-sample-production-success
SO which chip do you think this is.

 

[juanrga]

The lawsuit is pure nonsense. AMD Bulldozer family uses conjoined cores, whose concept was introduced in the literature years before by the academic people advancing computer science. The paper is question in "Conjoined-core Chip Multiprocessing" by Kumar et al and published in IEEE Micro on year 2004. It is ironic that two of the authors of the paper are from the Department of Computer Science and Engineering University of California and this nonsensical lawsuit is filled on California Court.

 

[juanrga]

http://www.anandtech.com/show/9764/globalfoundries-and-amd-announce-first-14nm-finfet-sample-production-success
SO which chip do you think this is.

GPU.

 

[cdrkf]

Yeah, it was a bad marketing choice however you slice it.

'Intel quad core faster than amd 8 core' has been the message ever since it released, when really the module concept is loosely comparable to an Intel core with ht (certainly similar in terms of size, number of transistors and so on).

Minus the fact that in most consumer cases Intels quad core i5s were faster than AMDs "8" cores and in gaming the i3 which is 2c/4t is equal to and sometimes faster.

That is why I always tried to get people to understand the module approach instead of just live by the marketing.

That is one areas where AMD has never done will, marketing. Intel has always done well and never seems to mis-market their products. AMD really needs good marketing. I mean look at Apple. Their products are now just over priced PCs and most of what used to be better to do on a Mac is now just as doable on a PC for less but their marketing department always seems to be able to convince people they are "better".

Well you get that with a few companies. Case in point I've found there's a general consensus that nVidia are somehow better just because. I mean yeah from time to time they are ahead, however when ATI / AMD have had the lead, nVidia are still often held up as 'better' somehow. I mean HD5000 was first DX11 gpu, and more efficient than the DX10 GTX 200 cards. HD6000 range was more efficient than GT400 range (however efficiency wasn't important apparently). HD7000 demolished GTX 500, message was "wait for Kepler". Which was faster for all of a few months until AMD did driver updates.

Now we have Maxwell, which is more efficient, even though performance wise we're near as dammit parity. Now all of a sudden efficiency is very important. The conclusion each generation by gamers / gaming media almost always sides with NV even when the hard facts don't align with that conclusion.

My conclusion is nVidia are waaay better at marketing than AMD 😛 I mean don't get me wrong, I've owned plenty of NV cards over the years (as well as ATi, AMD and even 3DFX back in the day haha), and I always buy based on the best card I can get for my budget. There are so many though that won't touch an AMD branded card because 'drivers' or other such nonesense because they are convinced NV are superior, it's the old' feels better' or 'more premium' or 'the thing you can't put your finger on'. Similar kinda sentiments people use to describe Apple gear.

 

[salgado18]

Well you get that with a few companies. Case in point I've found there's a general consensus that nVidia are somehow better just because. I mean yeah from time to time they are ahead, however when ATI / AMD have had the lead, nVidia are still often held up as 'better' somehow. I mean HD5000 was first DX11 gpu, and more efficient than the DX10 GTX 200 cards. HD6000 range was more efficient than GT400 range (however efficiency wasn't important apparently). HD7000 demolished GTX 500, message was "wait for Kepler". Which was faster for all of a few months until AMD did driver updates.

Now we have Maxwell, which is more efficient, even though performance wise we're near as dammit parity. Now all of a sudden efficiency is very important. The conclusion each generation by gamers / gaming media almost always sides with NV even when the hard facts don't align with that conclusion.

My conclusion is nVidia are waaay better at marketing than AMD 😛 I mean don't get me wrong, I've owned plenty of NV cards over the years (as well as ATi, AMD and even 3DFX back in the day haha), and I always buy based on the best card I can get for my budget. There are so many though that won't touch an AMD branded card because 'drivers' or other such nonesense because they are convinced NV are superior, it's the old' feels better' or 'more premium' or 'the thing you can't put your finger on'. Similar kinda sentiments people use to describe Apple gear.

You have just summarized what I've been thinking for a very long time. AMD needs to overhaul not only their products, but their marketing department. Otherwise, even great CPUs and GPUs will have a hard time recovering the company.

About the unnamed chip, probably GPU. From what I remember, they want to launch a new GPU arch in the first half of 2016, and Zen is expected on the second half, so first comes first.

 

[juanrga]

Zen is expected in 2017.

 

[cdrkf]

Juan, on what do you base that? All comments from AMD indicate they expect the first Zen cpus to ship in 2016 (albeit the last quarter)... I know the Zen apu aren't until 2017 but I've not seen anything on the cpu side indicating a delay as of yet...?

 

[8350rocks]

Oh dear...

http://www.guru3d.com/news-story/amd-faces-lawsuit-over-core-count-on-bulldozer.html

I said all along whoever decided to market 'dozer as '8' core should have been sacked....

From my perspective, AMD is right. As far as I'm concerned, a "core" is merely a unique register context. By that definition, an i3 is a Quad core though, but until someone defines what makes a unique Core, it's the best definition I've got to work with.

Except that AMD also refereed to this as what it is, CMT. It is an advanced version of SMT.

While I think the lawsuit is BS, like all lawsuits which come mostly out of the crazies of California, I do think that they should have never marketed it as a 8 core CPU. They should have just marketed it as a 4 core with CMT or their own name for it much like Intel advertises their CPUs as 4 cores with SMT/HT giving it 8 threads.

That is my personal opinion. I highly doubt AMD will spend the money it would take to dismiss this case based on petty stuff. More than likely they will settle if it goes to court so save the lawyer fees.

Attorney fees? 2 guys from the legal dept could likely get that dismissed...

This is the equivalent of suing Kellogg because the cereal box you bought is substantially larger than what it contained. Even though you got what was said on the box, and what was advertised, you feel slighted because you expected more.

That is all that is...nothing more.

 

[8350rocks]

http://www.anandtech.com/show/9764/globalfoundries-and-amd-announce-first-14nm-finfet-sample-production-success
SO which chip do you think this is.

GPU.

Next GPU is coming out of TSMC on 16nm FF...that is Zen @ GF

 

[jimmysmitty]

Yeah, it was a bad marketing choice however you slice it.

'Intel quad core faster than amd 8 core' has been the message ever since it released, when really the module concept is loosely comparable to an Intel core with ht (certainly similar in terms of size, number of transistors and so on).

Minus the fact that in most consumer cases Intels quad core i5s were faster than AMDs "8" cores and in gaming the i3 which is 2c/4t is equal to and sometimes faster.

That is why I always tried to get people to understand the module approach instead of just live by the marketing.

That is one areas where AMD has never done will, marketing. Intel has always done well and never seems to mis-market their products. AMD really needs good marketing. I mean look at Apple. Their products are now just over priced PCs and most of what used to be better to do on a Mac is now just as doable on a PC for less but their marketing department always seems to be able to convince people they are "better".

Well you get that with a few companies. Case in point I've found there's a general consensus that nVidia are somehow better just because. I mean yeah from time to time they are ahead, however when ATI / AMD have had the lead, nVidia are still often held up as 'better' somehow. I mean HD5000 was first DX11 gpu, and more efficient than the DX10 GTX 200 cards. HD6000 range was more efficient than GT400 range (however efficiency wasn't important apparently). HD7000 demolished GTX 500, message was "wait for Kepler". Which was faster for all of a few months until AMD did driver updates.

Now we have Maxwell, which is more efficient, even though performance wise we're near as dammit parity. Now all of a sudden efficiency is very important. The conclusion each generation by gamers / gaming media almost always sides with NV even when the hard facts don't align with that conclusion.

My conclusion is nVidia are waaay better at marketing than AMD 😛 I mean don't get me wrong, I've owned plenty of NV cards over the years (as well as ATi, AMD and even 3DFX back in the day haha), and I always buy based on the best card I can get for my budget. There are so many though that won't touch an AMD branded card because 'drivers' or other such nonesense because they are convinced NV are superior, it's the old' feels better' or 'more premium' or 'the thing you can't put your finger on'. Similar kinda sentiments people use to describe Apple gear.

I have seen the same thing with AMD. When Athlon 64 was king it was all about the performance per watt, because AMD was more expensive. Once Core 2 came out those same people started beating the price per dollar. Once we went to multiple cores IPC no longer was relevant. Each massive fan side does this.

Personally I go with what is best. I have bought ATI/AMD for years but finally moved back to nVidia and I don't regret it. I never regret my purchasing decision. Well I do wish I held out a bit longer on my Q6600 and go a 45nm part instead (Q9450) because man they OCed so much better.

8350rocks,

The box says (8-core CPU). What you get is not quiet a 8-core CPU.

I agree the lawsuit is stupid but they have a somewhat valid point and in the end it will still cost AMD money they do not have.

 

[gamerk316]

Oddly, the lawsuit hasn't touched their stock price. Still hanging around $2.20, or about 25% more then it was when I purchased it. 😛

 

[cdrkf]

Yeah, it was a bad marketing choice however you slice it.

'Intel quad core faster than amd 8 core' has been the message ever since it released, when really the module concept is loosely comparable to an Intel core with ht (certainly similar in terms of size, number of transistors and so on).

Minus the fact that in most consumer cases Intels quad core i5s were faster than AMDs "8" cores and in gaming the i3 which is 2c/4t is equal to and sometimes faster.

That is why I always tried to get people to understand the module approach instead of just live by the marketing.

That is one areas where AMD has never done will, marketing. Intel has always done well and never seems to mis-market their products. AMD really needs good marketing. I mean look at Apple. Their products are now just over priced PCs and most of what used to be better to do on a Mac is now just as doable on a PC for less but their marketing department always seems to be able to convince people they are "better".

Well you get that with a few companies. Case in point I've found there's a general consensus that nVidia are somehow better just because. I mean yeah from time to time they are ahead, however when ATI / AMD have had the lead, nVidia are still often held up as 'better' somehow. I mean HD5000 was first DX11 gpu, and more efficient than the DX10 GTX 200 cards. HD6000 range was more efficient than GT400 range (however efficiency wasn't important apparently). HD7000 demolished GTX 500, message was "wait for Kepler". Which was faster for all of a few months until AMD did driver updates.

Now we have Maxwell, which is more efficient, even though performance wise we're near as dammit parity. Now all of a sudden efficiency is very important. The conclusion each generation by gamers / gaming media almost always sides with NV even when the hard facts don't align with that conclusion.

My conclusion is nVidia are waaay better at marketing than AMD 😛 I mean don't get me wrong, I've owned plenty of NV cards over the years (as well as ATi, AMD and even 3DFX back in the day haha), and I always buy based on the best card I can get for my budget. There are so many though that won't touch an AMD branded card because 'drivers' or other such nonesense because they are convinced NV are superior, it's the old' feels better' or 'more premium' or 'the thing you can't put your finger on'. Similar kinda sentiments people use to describe Apple gear.

I have seen the same thing with AMD. When Athlon 64 was king it was all about the performance per watt, because AMD was more expensive. Once Core 2 came out those same people started beating the price per dollar. Once we went to multiple cores IPC no longer was relevant. Each massive fan side does this.

Personally I go with what is best. I have bought ATI/AMD for years but finally moved back to nVidia and I don't regret it. I never regret my purchasing decision. Well I do wish I held out a bit longer on my Q6600 and go a 45nm part instead (Q9450) because man they OCed so much better.

8350rocks,

The box says (8-core CPU). What you get is not quiet a 8-core CPU.

I agree the lawsuit is stupid but they have a somewhat valid point and in the end it will still cost AMD money they do not have.

Yeah it's true it does go both ways I guess. My point is though, *even when AMD had the best kit around* back with the A64, most people still only wanted Intel, because it's Intel and Intel is better. This is what I'm talking about.

With the A64, AMD had Intel in whatever metric you care to use (and whilst their FX cpu's were grossly overpriced, they had plenty of kit in the more sensible price ranges that was actually highly competitive, people seem to overlook the fact that with the release of the FX AMD did in fact have a full range of cpu's covering all price points). They should have picked up far more market share than they did, but Intel had the mindshare. I mean at that point in time AMD had repeatedly matched or bested Intel for years (since the very first Athalon), it wasn't like the A64 was the only successful processor they'd had a pretty strong run up to then. They just never got that fact over to consumers- you had to be interested in PC's to even know their name- yet everyone knew "Intel Inside".

 

[jimmysmitty]

You are mixing it up with marketing. The reason Intel had that was due to name recognition. Intel marketed the Intel Inside while AMD did not market as much. To add more to it, before the Athlon AMD was just a Intel clone with nothing special. Sure K6-II was there but was never a real threat while Athlon did pose a threat, still had its issues.

What you are talking about are the more fanboy sides of people but in some ways they can be right. If AMD releases Zen and Intel has a new chip right around the corner it would be wise to wait for the Intel chip to launch before buying. Hell before Fury hit people said to wait before jumping on a Maxwell GPU. it is all about timing.

Either way AMD has never marketed well. After they bought ATI ATIs marketing went to crap when before ATI had marketed pretty well and there were vastly more games with the ATI logo on them. Hell Half Life 2 was one of them and was given to people for free with 9800XTs (I bought one just for that game performance wise). AMDs marketing is either too weak or they mess up, as seen with Bulldozer. They need a better marketing team for sure.

 

[8350rocks]

I agree the lawsuit is stupid but they have a somewhat valid point and in the end it will still cost AMD money they do not have.

Hmm...

So, I guess the actual definition of a computer core is inaccurate then, according to several sources this is a computer core:

Definition of: processor core

The processing part of a CPU chip minus the cache. It is made up of the control unit and the arithmetic logic unit ( ALU ). See control unit and ALU.

So...the only way that a single AMD core does not fit the pre-conception of what is actually a CPU core would be to assume that a non-essential component that used to be a co-processor and was moved on die after sufficient process/arch improvements is now an essential component. Though, the irony there is, you do not need a FPU to actually run code...sure...it is more efficient at some things, but you can run code just fine without one.

In essence....the 8 core AMD processors are 8 core processors by the very definition accepted by the world.

Suppose on Skylake Intel elected to move the Southbridge back off the die for whatever reason, and their uarch that way ended up being basically the same performance as haswell/broadwell (since there was zero improvement in the move to broadwell either...)

Would you expect someone to suddenly sue Intel claiming "this is not really a core, because the preconceived notion of what a core is in my head, while vastly different from what the entire world accepts as the definition of a cpu core, must be the best definition because I said so, and it goes against what Intel claims..."

 

[jimmysmitty]

I agree the lawsuit is stupid but they have a somewhat valid point and in the end it will still cost AMD money they do not have.

Hmm...

So, I guess the actual definition of a computer core is inaccurate then, according to several sources this is a computer core:

Definition of: processor core

The processing part of a CPU chip minus the cache. It is made up of the control unit and the arithmetic logic unit ( ALU ). See control unit and ALU.

So...the only way that a single AMD core does not fit the pre-conception of what is actually a CPU core would be to assume that a non-essential component that used to be a co-processor and was moved on die after sufficient process/arch improvements is now an essential component. Though, the irony there is, you do not need a FPU to actually run code...sure...it is more efficient at some things, but you can run code just fine without one.

In essence....the 8 core AMD processors are 8 core processors by the very definition accepted by the world.

Suppose on Skylake Intel elected to move the Southbridge back off the die for whatever reason, and their uarch that way ended up being basically the same performance as haswell/broadwell (since there was zero improvement in the move to broadwell either...)

Would you expect someone to suddenly sue Intel claiming "this is not really a core, because the preconceived notion of what a core is in my head, while vastly different from what the entire world accepts as the definition of a cpu core, must be the best definition because I said so, and it goes against what Intel claims..."

I am not going to get into another "what defines a core" argument. All I can say is this: AMD also called it CMT, which is a more advanced version of SMT where two cores share certain resources that are normally part of a single core. That alone means that even AMD knows it is not a full 8 core CPU in the respect that if you compare it to a 8 core Xeon it is missing parts.

Southbridge is a bad example since that is not part of the core of a CPU it is along side it.

Again I am agreeing with you in that the actual lawsuit is BS, much like the majority of them these days. I am just saying that AMD did market it as a 8 core CPU when it was a 4 module 8 thread CPU using a technology called SMT.

There is also a reason AMD is abandoning it, CMT just is not worth it.

 

[Vogner16]

Oh dear...

http://www.guru3d.com/news-story/amd-faces-lawsuit-over-core-count-on-bulldozer.html

I said all along whoever decided to market 'dozer as '8' core should have been sacked....

This depends on what the judge decides as what a core really is. From what I understand we have decided that one needs an integer and fp unit to be a core making FX a dual thread module BUT if you look into the past CPU's did not have FPU's. so historicaly speaking the FX line is mearly a 8 core with 4 extra FPU's.

This sounds like another money grab from some intel/nvidia fanboi who wants to watch the world burn.
Even if I had bought a 970 I wouldn't have filed a lawsuit on them or claimed on it. I don't like hurting business in the tech world and think that competition will iron out all the issues that pop up. NVidia knows that that .5 gb business was crap and I would be VERY supprised if they did it again. amd knows their 8 core nonsense was crap and as far as we know are changing their situation with zen labeling a quad core a quad core with 8 threads.

 

[jimmysmitty]

Oh dear...

http://www.guru3d.com/news-story/amd-faces-lawsuit-over-core-count-on-bulldozer.html

I said all along whoever decided to market 'dozer as '8' core should have been sacked....

This depends on what the judge decides as what a core really is. From what I understand we have decided that one needs an integer and fp unit to be a core making FX a dual thread module BUT if you look into the past CPU's did not have FPU's. so historically speaking the FX line is merely a 8 core with 4 extra FPU's.

This sounds like another money grab from some intel/nvidia fanboi who wants to watch the world burn.
Even if I had bought a 970 I wouldn't have filed a lawsuit on them or claimed on it. I don't like hurting business in the tech world and think that competition will iron out all the issues that pop up. NVidia knows that that .5 gb business was crap and I would be VERY supprised if they did it again. amd knows their 8 core nonsense was crap and as far as we know are changing their situation with zen labeling a quad core a quad core with 8 threads.

Yes and no. Problem is that now most every CPU, and for quite a while, has included the FPU in the core of the CPU itself. While we could go all the way back to the original 8086 and say that it doesn't even need the math co-processor to be considered a CPU core, would you ever buy a modern CPU without one?

That is why this is so tricky and considering how sue happy people are over the dumbest of things (oh no, I spelt hot coffee on my lap and its your fault) any company would try to avoid any grey areas that could lead to any liability at all.

Zen is changing because they are moving back to a design much like K10 and using a very similar SMT implementation that Intel uses. It would be stupid to try to call a 4 core 8 thread Zen chip a 8 core CPU. If I was running AMD and anyone even mentioned that I would fire them.

 

[juanrga]

Juan, on what do you base that? All comments from AMD indicate they expect the first Zen cpus to ship in 2016 (albeit the last quarter)... I know the Zen apu aren't until 2017 but I've not seen anything on the cpu side indicating a delay as of yet...?

In the last official statement after Keller's departure, AMD didn't say shipping or release in 2016. They said "sampling" in 2016. Moreover, recall the interview to the Zen team. The article said first Zen products expected for 2017.

 

[juanrga]

Oh dear...

http://www.guru3d.com/news-story/amd-faces-lawsuit-over-core-count-on-bulldozer.html

I said all along whoever decided to market 'dozer as '8' core should have been sacked....

This depends on what the judge decides as what a core really is. From what I understand we have decided that one needs an integer and fp unit to be a core making FX a dual thread module BUT if you look into the past CPU's did not have FPU's. so historicaly speaking the FX line is mearly a 8 core with 4 extra FPU's.

This sounds like another money grab from some intel/nvidia fanboi who wants to watch the world burn.
Even if I had bought a 970 I wouldn't have filed a lawsuit on them or claimed on it. I don't like hurting business in the tech world and think that competition will iron out all the issues that pop up. NVidia knows that that .5 gb business was crap and I would be VERY supprised if they did it again. amd knows their 8 core nonsense was crap and as far as we know are changing their situation with zen labeling a quad core a quad core with 8 threads.

No. The judge doesn't define what is a core... the judge doesn't define what is a planet, what is an electron, or what is a bacteria. Scientists and engineers in a given field define the terminology and concepts used in the field.

No, we didn't decide that "one needs an integer and fp unit to be a core". Not only you can design a core without FPU, but as shown above, the concept of cores sharing resources with adjacent cores was introduced in the year 2004 in a peer-reviewed specialized journal.

AMD Zen is not a different design because the terminology is wrong, but because AMD couldn't get the CMT concept to work.

AMD Zen will be labeled like 8C/16T, because it is a CMP+SMT design, not because AMD is changing the terminology.

 

[vmN]

I cant see this lawsuit gaining anything.

Are there any real clear definition of what a 'core' is?
Looking at bulldozers design, I do personally think 8c is more appropriate than 4c. However 4m/8t is the most descriptive, but the 'module' term arent as recognizable like the 'core' term.

I don't see it sharing the FPU would be the biggest problem, but rather it sharing certain frontend components.
But is conjoined twins one or two persons?

 

[Vogner16]

I cant see this lawsuit gaining anything.

Are there any real clear definition of what a 'core' is?
Looking at bulldozers design, I do personally think 8c is more appropriate than 4c. However 4m/8t is the most descriptive, but the 'module' term arent as recognizable like the 'core' term.

I don't see it sharing the FPU would be the biggest problem, but rather it sharing certain frontend components.
But is conjoined twins one or two persons?

@juan the judge will decide for that case. to make the rule for all rules he will not. this is why the whole argument is invalid. to call 1 mb L1 cashe 2 mb L1 cashe would be reason for false advertising. I honestly think that a "core" should be based on the original intel 4004 processor and simply call all other things attached to it what they are. FPU,IPU decode, L1 etc... the indivual core is just the exact pipeline.

Your arguments regarding would you buy a cpu without a FPU are valid however irrelevant regarding the definition of "core". I could argue that amd's cpu design is really a 32 core cpu because of 4 pipelines per L1 cashe if you use the base definition of a 1970's "core". Its like saying that cpu's today are just cores. no they are a UNIT. they have all kinds of things aside from "cores" attached to them from graphics memory controllers to cashes.

The entire thing is marketing and software. How do developers use each module/ core. do they write 8 integer heavy processes or 4 floating point heavy lines... does amd want to say we assume the first situation saying we can handle 8 integer processing lines at a time there for 8 cores?

marketing decides the "core" count. real world performance REGARDLESS of what marketing says is all that matters.

we can say that amd's FX chip for $200 does 7000 points on 3dmark 11 and Intel 3770K does 9000 points for $330. that's all that matters. end of discussion. as jingles would say why you haff to be mad?

This lawsuit? exactly as jimmy said, sue happy people with spilt coffee. It is based in the intent to grab money with no real background and "shouldn't" win to a rational judge.

if amd marketing claimed FX had a core for each transistor in the cpu then we might have some claims to make BUT they did not.

 

[ex_bubblehead]

Let's take the "what is/is not a core" discussion to another thread, it's off topic here.

 

[Guest]

From AMD's own site, first Zen cores will be available in 2016 -
https://community.amd.com/community/amd-corporate/blog/2015/05/12/amd-financial-analyst-day-2015

 

[Vogner16]

I watched this live. We all know how roadmaps change. as we know working samples have already been tapped out I would assume they are still on track.

https://community.amd.com/servlet/JiveServlet/downloadImage/38-1076-1955/AMD+FAD+2015_Zen.png

We have all seen the slide

http://wccftech.com/amd-14nm-cpu-apu-gpu-zen-arctic-islands-globalfoundries/

@juan if you can verify with your amazing ability to sniff out fake info to how real this is.

IF the wccftech article is correct they are on track for 2016 release. imagine amd and intel back on the same process node for the first time since 2010 when amd released llano as intel is scheduled to release 10nm by 2017.

EDIT: http://wccftech.com/amd-zen-k12-taped/

This link too