AMD Releases New Athlon FX CPU SKUs, Wraith Coolers Join The Party

[alextheblue]

as well as the Excavator architecture

Note: the A10-7860K does not have an Excavator core. It wish it did, but it's still Steamroller-based.

This isn't Carrizo for the desktop. They killed that, and it's still dead.

They were referring to the 845, which is Excavator. The improved GPU was referring to the 7860K. He could have made it more clear which comment referred to which chip, but it's pretty obvious the 845 is the only Excavator based chip if you read the article.

Anyway they really need to release an 865K or something based on Excavator. I know it wasn't designed to scale up as much but it would be a nice sendaway present for FM2+.

 

[jimmysmitty]

No, that's also incorrect. the 7850K was succeeded by the 7870K.

So that changes which CPU is the flagship?

It does. The 7870K is faster, and therefore succeeds it. Sort of like the i7-4770K being replaced by the Haswell Refresh i7-4790K.

It replaced it as the top end but did not succeed it. Succeed would mean that the former is no longer sold but it is still sold. Same with the 4770K, it is still sold with the 4790 becoming the top end.

Either way there is nothing wrong with pointing out inaccuracies in an article, in fact we encourage it as TH wants the best quality articles. But putting it a certain way is also desirable. You could have pointed it out instead of stating the article is incorrect. The article is correct, there is just a misprint or inaccuracy in one point of it.

 

[bit_user]

as well as the Excavator architecture

Note: the A10-7860K does not have an Excavator core. It wish it did, but it's still Steamroller-based.

This isn't Carrizo for the desktop. They killed that, and it's still dead.

They were referring to the 845, which is Excavator. The improved GPU was referring to the 7860K. He could have made it more clear which comment referred to which chip, but it's pretty obvious the 845 is the only Excavator based chip if you read the article.

Dude, why the downvote? Even after reading the article, I really wasn't sure about the A10-7860K, and those power numbers hinted at a substantial change of some sort. So, I hunted down information from external sources, just to be sure. Then, I thought I'd share that on here, in case anyone else was wondering.

Furthermore, two of the three changes @turkey3_scratch mentioned were in the A10-7860K, so it seemed plausible @turkey3_scratch was under the wrong impression.

IMO, downvotes are for things that are factually wrong, or people just misbehaving.

 

[alextheblue]

This isn't Carrizo for the desktop. They killed that, and it's still dead.

IMO, downvotes are for things that are factually wrong, or people just misbehaving.

Since you insist... you said that they killed Carrizo on desktop. What the heck is the 845 then? It's Carrizo. Now go downvote yourself.

 

[bikerepairman1]

The DTP between AMD and Intel is measured differently. For intel it's the average power consumption and for AMD it is the real max, and consumes mostly lower.
(at least that is how it always was, simply said).

 

[bit_user]

This isn't Carrizo for the desktop. They killed that, and it's still dead.

IMO, downvotes are for things that are factually wrong, or people just misbehaving.

Since you insist... you said that they killed Carrizo on desktop. What the heck is the 845 then? It's Carrizo. Now go downvote yourself.

Nonsense. Carrizo is an APU, while the Athlon X4 845 is not.

The latest roadmap has a new APU, due this year (Bristol Ridge). Since it won't be Zen-based, I wonder if they will use the Excavator core. If you have information on this, please share.

 

[jimmysmitty]

The DTP between AMD and Intel is measured differently. For intel it's the average power consumption and for AMD it is the real max, and consumes mostly lower.
(at least that is how it always was, simply said).

That is actually backwards. Intel always uses a TDP based estimate, meaning the maximum power it should draw under normal conditions. AMD moved to an ACP based estimation which gives the average power used based on work loads.

Either way the best estimation of power draw is with a third party.

 

[turkey3_scratch]

as well as the Excavator architecture

Note: the A10-7860K does not have an Excavator core. It wish it did, but it's still Steamroller-based.

This isn't Carrizo for the desktop. They killed that, and it's still dead.

They were referring to the 845, which is Excavator. The improved GPU was referring to the 7860K. He could have made it more clear which comment referred to which chip, but it's pretty obvious the 845 is the only Excavator based chip if you read the article.

Dude, why the downvote? Even after reading the article, I really wasn't sure about the A10-7860K, and those power numbers hinted at a substantial change of some sort. So, I hunted down information from external sources, just to be sure. Then, I thought I'd share that on here, in case anyone else was wondering.

Furthermore, two of the three changes @turkey3_scratch mentioned were in the A10-7860K, so it seemed plausible @turkey3_scratch was under the wrong impression.

IMO, downvotes are for things that are factually wrong, or people just misbehaving.

Yeah if AMD needs to change anything it's the namingt of their processors. With Intel, it's simple to know what CPU is based off what architecture. Very simple numbering system with a Core I3, I5, or I7 (and Pentium or Celeron). With AMD< you have Athlons out the wazoo, you got FX out the wazoo, the whole naming scheme is a mess, you got different architecture-based CPUs being refreshed simultaneously. You got Athlon A6, A8, A10, FX 6xxx, 4xxx, 8xxx. I tell you, it drives me nuts alone when companies can't create a normal naming system like Intel. Xeons are sort of different, but the consumer chips are nicely named.

 

[alextheblue]

Since you insist... you said that they killed Carrizo on desktop. What the heck is the 845 then? It's Carrizo. Now go downvote yourself. [/quotemsg]Nonsense. Carrizo is an APU, while the Athlon X4 845 is not.

The latest roadmap has a new APU, due this year (Bristol Ridge). Since it won't be Zen-based, I wonder if they will use the Excavator core. If you have information on this, please share.

With all due respect, you're wrong again. In terms of silicon the 845 is an APU with the GPU disabled. They call it a CPU but it's still the same silicon. A Llano Athlon is still a Llano, same for all APU-based "Athlon" CPUs. Trinity Athlons are Trinity, Kaveri Athlons are Kaveri, and the Carrizo Athlon X4 845 is a Carrizo.

You asked me where you were wrong and there you are.

Bristol Ridge is very strongly rumored to be Excavator-based as a sort of stand-in for Zen to help get AM4 established. Plus investing in a new FM2+ build is getting less appealing. People are more likely to buy AM4 platforms with today's chips, knowing there's an upgrade path. OEMs are similarly more likely to want to purchase and build designs around AM4, knowing they won't get stuck with boards nobody wants once Zen comes around.

 

[turkey3_scratch]

The term "APU" is just silly, throwing in "accelerated" doesn't change the fact it's just a CPU with integrated graphics. Intel doesn't call all their processors "APU"s, it's just marketing.

 

[bit_user]

With Intel, it's simple to know what CPU is based off what architecture. Very simple numbering system with a Core I3, I5, or I7 (and Pentium or Celeron).

Actually, the generation is the first digit after "i3", "i5", or "i7". So, iN-2xxx is Sandybridge, iN-3xxx is Ivybridge, iN-4xxx is Haswell, iN-5xxx is Broadwell, and iN-6xxx is Skylake.

I agree that AMD would do well to reduce confusion about their products. And not just their CPUs & APUs, but also their GPUs. I never liked the way they would rebadge and include previous gen GPUs, when they introduce a new model line. It would be nice if a certain model line of Radeons would all have the same level of API support, for instance. Wishful thinking, I know.

 

[bit_user]

With all due respect, you're wrong again. In terms of silicon the 845 is an APU with the GPU disabled. They call it a CPU but it's still the same silicon. A Llano Athlon is still a Llano, same for all APU-based "Athlon" CPUs. Trinity Athlons are Trinity, Kaveri Athlons are Kaveri, and the Carrizo Athlon X4 845 is a Carrizo.

You asked me where you were wrong and there you are.

Do you have a source on that? Or is it just speculation? Please provide a reliable source, or else we can only assume the latter.

I can speculate, too. It sounds really odd that they wouldn't sell an APU version, if they went to all the trouble of building a die with a full GPU (which costs money, not just due to the time & effort to design & test it, but also since it means fewer dies per wafer). But the point is that Carrizo is a codename for an APU product, which features a GPU, has a different socket, and was to succeed Kaveri in various market segments.

Kaveri and Carrizo are AMD code names. If you're so intent on proving me wrong, then you really need to show some evidence that AMD refers to the 845 as Carrizo. It doesn't matter what you call them in your own mind. You can't lawyer your way out of this.

Honestly, it saddens me that you seem to have nothing better to do with your time than to tear people down and try to win petty arguments on these forums. This exchange is long past the point of benefiting anyone.

 

[bit_user]

Bristol Ridge is very strongly rumored to be Excavator-based as a sort of stand-in for Zen to help get AM4 established. Plus investing in a new FM2+ build is getting less appealing. People are more likely to buy AM4 platforms with today's chips, knowing there's an upgrade path. OEMs are similarly more likely to want to purchase and build designs around AM4, knowing they won't get stuck with boards nobody wants once Zen comes around.

I had also read the Excavator rumor. I didn't know timing was such that it would hit before Zen, though.

Thanks, that was helpful.

 

[turkey3_scratch]

With Intel, it's simple to know what CPU is based off what architecture. Very simple numbering system with a Core I3, I5, or I7 (and Pentium or Celeron).

Actually, the generation is the first digit after "i3", "i5", or "i7". So, iN-2xxx is Sandybridge, iN-3xxx is Ivybridge, iN-4xxx is Haswell, iN-5xxx is Broadwell, and iN-6xxx is Skylake.

I know the first number indicates generation, I never said it didn't

 

[bit_user]

I know the first number indicates generation, I never said it didn't

Fair enough. I was trying to tie together that first sentence with the second. Didn't mean to put words in your mouth, though.

 

[IInuyasha74]

With all due respect, you're wrong again. In terms of silicon the 845 is an APU with the GPU disabled. They call it a CPU but it's still the same silicon. A Llano Athlon is still a Llano, same for all APU-based "Athlon" CPUs. Trinity Athlons are Trinity, Kaveri Athlons are Kaveri, and the Carrizo Athlon X4 845 is a Carrizo.

You asked me where you were wrong and there you are.

Do you have a source on that? Or is it just speculation? Please provide a reliable source, or else we can only assume the latter.

I can speculate, too. It sounds really odd that they wouldn't sell an APU version, if they went to all the trouble of building a die with a full GPU (which costs money, not just due to the time & effort to design & test it, but also since it means fewer dies per wafer). But the point is that Carrizo is a codename for an APU product, which features a GPU, has a different socket, and was to succeed Kaveri in various market segments.

Kaveri and Carrizo are AMD code names. If you're so intent on proving me wrong, then you really need to show some evidence that AMD refers to the 845 as Carrizo. It doesn't matter what you call them in your own mind. You can't lawyer your way out of this.

Honestly, it saddens me that you seem to have nothing better to do with your time than to tear people down and try to win petty arguments on these forums. This exchange is long past the point of benefiting anyone.

That is correct we cannot say it is a Carrizo part for sure, or that it is an APU. One of AMD's current goals is to make their hardware components more modular, so that they can quickly and easily produce customized pieces of silicon for other companies. They have been working to do this for a number of years, and it wouldn't be hard for them at this point to create an Excavator CPU without a GPU for the FM2+ socket, and it would make sense to do so if they wanted to offer the chip at a lower price as it would be cheaper to produce. They could have a GPU inside of the CPU as well, and just have it deactivated, but they wouldn't have any reason to do that and then turn around and not release a CPU with the GPU enabled.

 

[alextheblue]

Honestly, it saddens me that you seem to have nothing better to do with your time than to tear people down and try to win petty arguments on these forums. This exchange is long past the point of benefiting anyone.

It saddens me that you demand that I only downvote you when you're wrong (and not just when I disagree with something in your comment), and when I prove you wrong TWICE you change gears and start calling me petty (in your own fashion) for taking the time to do exactly what you told me to do.

First you insist no Carrizo on desktop (you're wrong) and then you insist that 845 isn't Carrizo (wrong). Then you say the burden of proof lies on me, attack me, and run away all in one swoop.

http://anandtech.com/show/10009/amd-launches-excavator-on-desktop-the-65w-athlon-x4-845-for-70

Read the last sentence. The update where AMD confirmed to them that it was Carrizo. I don't expect an apology, but next time keep in mind that I generally don't make such firm statements without confidence in the information, and people don't all vote by your rules, Herr bit_user.

 

[alextheblue]

That is correct we cannot say it is a Carrizo part for sure, or that it is an APU. One of AMD's current goals is to make their hardware components more modular, so that they can quickly and easily produce customized pieces of silicon for other companies. They have been working to do this for a number of years, and it wouldn't be hard for them at this point to create an Excavator CPU without a GPU for the FM2+ socket, and it would make sense to do so if they wanted to offer the chip at a lower price as it would be cheaper to produce. They could have a GPU inside of the CPU as well, and just have it deactivated, but they wouldn't have any reason to do that and then turn around and not release a CPU with the GPU enabled.

Actually we can say that it's Carrizo. See the above link. Also modular design or not, making a new chip is still quite involved. Ask one of the TH editors. It would cost a considerable amount of money, time, and effort - clearly above what they're willing to invest at this point. I am not even sure they would recoup the costs to develop a new Excavator-based chip with no GPU silicon. I mean it would only see mild success at best, and even then most likely only on the desktop. I will also say that the custom chips they produced for Sony and MS were not something they slapped together in a couple months.

Look, think about it: If it's so cheap and easy to spit out a new chip, why are ALL the FM-based Athlons (from FM1 on up to FM2+) actually APUs with a disabled GPU? It's a major investment, a bad use of engineering manpower, and they don't sell enough Athlons to justify it. But that's not the only reason. They save the "bad" silicon by binning/repurposing. If the GPU is busted or not up to snuff, disable it and sell an Athlon. I don't know if they even originally intended to release Carrizo on desktop, but perhaps they got enough chips with issues in the GPU to warrant it. If it sees mild success perhaps we'll see an FM2+ Carrizo APU. I personally think we're more likely to see a fully-enabled Carrizo APU or an enhanced version of it on AM4, but we can't be certain.

Oh, and I'm seeing rumors that they're also releasing an 835 Athlon that is similarly a Carrizo sans functioning GPU. If it's priced right that could also be an interesting cheap upgrade option for existing FM2+ systems that are equipped with something outdated or are thermally limited (SFF perhaps).

 

[bit_user]

It saddens me that you demand that I only downvote you when you're wrong (and not just when I disagree with something in your comment)

Hitting me back with my own words undermines your sincerity. It's my right to take issue with your downvote, especially for a post consisting of factual information relevant to the thread.

As for the proper use of the voting system, my statement was meant as more of a thesis. But I think we actually agree on one point, here. I've been giving some thought to the purpose of the voting system. Since it has the effect of burying or highlighting posts, it would also seem appropriate (and consistent with common practice) to vote up or down posts advancing a strong opinion with which one agrees or disagrees (I regularly use up-votes for that, but I'm much more restrained in my use of down-votes). I also feel it'd be appropriate for someone to down-vote highly off-topic comments, including those of our exchange.

and when I prove you wrong TWICE you change gears and start calling me petty (in your own fashion)

If you read what I actually said, I never called you petty. This entire flame amounts to you trying to defend your downvote through technicalities and rhetoric. It's entirely about ego, and any benefit to the topic of the thread is almost incidental. Now, try to convince any reasonable bystander that's not a petty argument...

First you insist no Carrizo on desktop (you're wrong)

Nope. You haven't shown that, at all. The desktop Carrizo, like Kaveri, Kabini, Richland, etc. was an APU. That's still dead, as far as we know.

and then you insist that 845 isn't Carrizo (wrong).

AMD uses the same die, for different products, all the time. But they give it a different code name, because it's a different product, with different features, addressing a different market. And what this whole line of argumentation is really about is you trying to nail me on a technicality, which I feel is hardly constructive. I call it a technicality, because, in your response to my complaint you didn't even mention anything about Carrizo - that's just a rationale you constructed, after the fact.

And when one immediately precedes a statement with "I can speculate, too.", it seems a bit rich to say that "I insisted".

Then you say the burden of proof lies on me, attack me, and run away all in one swoop.

I also didn't attack you. Examples of that would be namecalling or trying to undermine your credibility. I have only sought to dismantle your argument against me, which I can only do because it's so weak.

You're the one making the accusations, here. It's only proper that the burden of proof lies with the accuser. And what's this about running away? I never said such a thing.

Herr bit_user.

If defending oneself against unwarranted downvotes and attacks earns one that distinction, so be it. It does no service to the strength of your argument, however. Usually, one resorts to such tactics out of desperation, implying a recognized weakness in one's own position.

 

[bit_user]

I am not even sure they would recoup the costs to develop a new Excavator-based chip with no GPU silicon.

Nobody is saying they should remove a GPU they're not using - the point is that if it had a GPU, then why didn't they release an APU with it enabled?

I'm only pointing out the rationale behind our skepticism. Maybe there's a good reason for it, like that it has a critical bug that can't be fixed in the metal layer, and they're just going to fab it at 14/16 nm, instead of doing a respin w/ a fix at 28 nm. Even that's a bit of a stretch, since the chip that makes it to market is almost never the first spin.

 

[alextheblue]

Fine, I'll set aside your attacks and deflections, because they're just too clever and nobody would notice. You refuse to acknowledge you were wrong. I get that. You didn't even read the last part in the article I posted. It IS Carrizo. That is fact. That makes you wrong twice. I would never have even brought any of this up if you didn't insist my downvotes were unwarranted.

Processor dies aren't perfect. They have defects. We have been over this before. When they have an APU with a defective GPU, they disable the GPU and sell an Athlon. Does that not sufficiently explain why they would release one with the GPU turned off?? You aren't going to sell many (zero) GPU-free Athlons to laptop OEMs, so they make them desktop chips.

I've also heard a rumor that FM2+ doesn't support the voltage planes necessary for Carrizo's GPU portion. If that's true that means no Carrizo fully enabled on FM2+, but we still get Carrizo Athlons... like the 845.

So in summary, 845 is Carrizo, 845 is on desktop, therefore Carrizo is on desktop.

 

[bit_user]

Fine, I'll set aside your attacks and deflections,

That's awfully big of you.

You didn't even read the last part in the article I posted. It IS Carrizo.

Of course I did. They said is was the same die as Carrizo. I said I doubted that (though I did not state that as fact), but there it is.

However, the desktop Carrizo line that they cancelled, last year, was to have a functioning GPU. This does not. Whether or not it's based on the same die, I still can't go out and buy a Carrizo APU with a functioning GPU (which I was hoping to do, last year, as I wanted to experiment with HSA). That was my whole point, if you go back to my very first post. You're just trying to "win" on technicalities (though I'd say we've both lost, at this point).

I would never have even brought any of this up if you didn't insist my downvotes were unwarranted.

Which is was. My I post contained relevant and factual information. You said nothing about Carrizo, in your initial rationale - that was only something you came up with, after the fact. Given that you've never acknowledge this, I can only assume you know I'm right and are hoping I'll just drop that inconvenient fact (a fact which really has a lot more to do with this thread than whatever die the Athlon X4 845 uses).

Processor dies aren't perfect. They have defects. We have been over this before.

Yes, that's why they have engineering samples. It's pretty uncommon (or at least it used to be), for a rev A. chip even to make it to market.

Anyway, this is a tangent to an irrelevant point. In case you need me to say it clearly: you were correct on them using the same die. I was skeptical of that, but I didn't state it as fact that they didn't (in case you didn't bother to read my two previous posts).

 

[alextheblue]

The processor die is quite effectively... the processor. Look at a die wafer sometime. Saying they use the Carrizo die is the same as saying it is Carrizo. Which is it. Whether or not the GPU is functional doesn't change the chip's designation to not-Carrizo. Trust me. Or look up a list of AMD APUs on Wikipedia. All the disabled-GPU FM Athlons are the same as their APU brethren... with the GPU disabled.

Now for the important part which you still do not fully grasp:

[

Processor dies aren't perfect. They have defects. We have been over this before.

Yes, that's why they have engineering samples. It's pretty uncommon (or at least it used to be), for a rev A. chip even to make it to market.

That is NOT what I mean. For example, all Kaveri chips use the same die. They all have the same units. When you produce a wafer of processor dies, there are FLAWS in the wafer. It is completely normal for chips to be produced with imperfections. Look up things like processor yield and binning. They test and sort dies. Really good ones become high-end chips. Ones with minor faults that don't clock as good or have other issues get marked for use in mid-grade SKUs. Ones with more problems? Well... depending on where the problems ARE, you can shut off large sections of the chip and figure out what to do with them, or throw them away.

AMD can't afford to throw away many chips. In the case of their APUs, the GPU takes up a huge portion of the die. If there's relatively minor issues, you can take a 512 shader part and shut off large chunks, producing a 384 or 256 shader part. Or even, depending on where the flaws lie or how greatly they affect operation, turn off the GPU entirely and sell an Athlon. In some cases you have faults in the CPU side and have to disable a module (resulting in a dual core) or toss the CPU.

That's just a super simplistic overview. They've been doing this sort of thing for AGES with chips, CPUs, GPUs, APUs. FX-4xxx chips use the same 8-core die as the FX-8xxx chips. There's other factors too, like built-in redundancy, and market demand. That's why (with certain designs) you used to be able to "unlock" cores (typically through the BIOS). The extra cores were there all along, but sometimes they were disabled due to flaws, and other times due to market demand they let good chips get marked down. So there was a sort of core-unlocking gamble for those willing to chance it. If you unlocked one with flawed component, it could be unstable, for example.

 

[bit_user]

...

Instead of spouting a bunch of generalities that we all know, why not address the point of your downvote. It was never about Carrizo. You downvoted me because I posted something not stated in the article that you took as given. Since you've never acknowledged this, I can only assume you know I'm right and are hoping I'll just drop that inconvenient fact (a fact which really has a lot more to do with this exchange than whatever die the Athlon X4 845 uses).

And you accuse me of deflections...

 

[alextheblue]

They're not generalities if they're information you didn't know. You didn't understand why they were disabling the GPU (hence why you were skeptical). You brought engineering samples up in a discussion about chip defects. You thought a Carrizo die could be something other than Carrizo. But now you know.

As for the other bickering - not really, I just didn't take the bait at the moment since I was flabbergasted that you didn't understand why they fuse off sections of their APUs. If you really want to know, the reason I didn't bring up Carrizo initially was that I didn't want to have to argue about it with you specifically. It wasn't until you said to me "IMO, downvotes are for things that are factually wrong, or people just misbehaving" that I said fine, why not have you argue with me about this being Carrizo or not for days on end.

THAT'S why I didn't say anything to start with. Like you care. I should have just told you I downrated you because you didn't say the word "Shrubbery" anywhere in your post. In fact, from now on, that's my reason for downvoting someone. Because they didn't adequately use the word "Shrubbery".