AMD Carrizo Pre-release thread

[Phynaz]

Becoming?

 

[ShintaiDK]

If people are looking to buy an AMD "APU" desktop chip as an upgrade from RIchland or Kaveri, the only thing worth mentioning is Zen parts that *should* arrive next year (if AMD doesn't screw up those also).

Zen have been missing a lot lately in updates. So the chip may not even exist anymore. However even if it does, its unlikely to be more than something between cat cores and big cores. Its quite certain it wont beat 7850K. Also K12 is told to be 28nm from AMD themselves. If Zen still exist, it would be the same.

With Carrizo pulled from the desktop. It seems AMD completely dropped the desktop segment and tries to focus while it can on mobile where its easier to get away slower CPUs.

 

[Abwx]

Zen have been missing a lot lately in updates. So the chip may not even exist anymore. However even if it does, its unlikely to be more than something between cat cores and big cores. Its quite certain it wont beat 7850K. Also K12 is told to be 28nm from AMD themselves. If Zen still exist, it would be the same.

With Carrizo pulled from the desktop. It seems AMD completely dropped the desktop segment and tries to focus while it can on mobile where its easier to get away slower CPUs.

So much fud in this post, i bolded one of the most innaccurate statement..

 

[Phynaz]

So much fud in this post, i bolded one of the most innaccurate statement..

You would be better off not playing that game, lest somebody quote your posts from this very topic.

 

[ShintaiDK]

So much fud in this post, i bolded one of the most innaccurate statement..

Your slide is 8 months old. And 28nm came from Kumar quite recently.

So are you saying Kumar lied? Or just that you know better than him?

http://seekingalpha.com/article/273...nductors-software-and-supply-chain-conference

 

[hunkeelin]

Honestly, just give up with AMD. Not a single AMD cpu as of now can rival the 5 years old 980x. They are just so miserable.....

 

[piesquared]

Your slide is 8 months old. And 28nm came from Kumar quite recently.

So are you saying Kumar lied? Or just that you know better than him?

http://seekingalpha.com/article/273...nductors-software-and-supply-chain-conference

Umm no, the liars here are you and that dip sh0t mrmt. The thread he or she started was erroneous and neither mrmt or the so called mods fixed the title. But you already know this, you're just a low life shill spreading FUD and anti-AMD propaganda. You kids belong to the same cult?


What, you think mod callouts are ok ?
enjoy your vacation.
Markfw900

 

[Phynaz]

Umm no, the liars here are you and that dip sh0t mrmt. The thread he or she started was erroneous and neither mrmt or the so called mods fixed the title. But you already know this, you're just a low life shill spreading FUD and anti-AMD propaganda. You kids belong to the same cult?

Quoting for mods.

 

[cbn]

It makes sense since Carrizo runs at much lower frequencies than Kaveri so any desktop part would be inferior to 7850K (even tho there is certain IPC jump, it's not enough to compensate the frequency deficit vs Kaveri).

I thought both Kaveri and Carrizo were on 28nm SHP?

So unless there is something about the Excavator uarch that lends its self to lower clocks shouldn't they both have the same max frequency potential?

 

[geoxile]

Ashraf Eassa (tech writer for Motley Fool) said he contacted AMD about the statement by Kumar and they told him K12 would be on finfet.

http://semiaccurate.com/forums/showthread.php?t=8379

It was quickly posted on this forum too, in a fairly active thread
http://forums.anandtech.com/showthread.php?t=2411975&page=5

Also, complaining about the users on anand won't get anything done. You should've known this site's rep coming in.

 

[Abwx]

You would be better off not playing that game, lest somebody quote your posts from this very topic.

You can report all the way, there s about nothing that is right in said post, if you cant point something accurate i m all ears, but from the tone of your post you have no argument other than eventualy trying to silence me through deffamatory claims, and dont forget to point what was on point in the post you re so eagerly unable to support with real arguments...

Your slide is 8 months old. And 28nm came from Kumar quite recently.

So are you saying Kumar lied? Or just that you know better than him?

http://seekingalpha.com/article/273...nductors-software-and-supply-chain-conference

You know that Intel17 contacted AMD on the subject and he said that Kumar wasnt talking of Zen or K12, so why are you insisting on spreading innaccuracies that were debunked by a member of this forum who can hardly be branded an AMD supporter..?.

Perhaps that you want me to point other innaccuracies in your previous post? let s start for a few of thoses wrong infos, not to say deliberate fabrications :

Zen have been missing a lot lately in updates. So the chip may not even exist anymore.

I could answer that if my aunt has some....she would be called mly uncle, we have discussed this already and i pointed that AMD didnt release any roadmap for some products like the R9s, and announced Carrizo only six months before release, it s obvious that they are not communicating as much as they used to be.

However even if it does, its unlikely to be more than something between cat cores and big cores. Its quite certain it wont beat 7850K. Also K12 is told to be 28nm from AMD themselves. If Zen still exist, it would be the same.

As said Intel17 got confirmation that it wont be 28nm, but let s admit that this is the case, so Zen/K12 would be, according to you, less powerfull than Kaveri but more powerfull than Beema, are you serious.??.

Why would they release a chip that is not as good as Kaveri and be below Carrizo level.?.

These are wishfull thoughts of a rare wildness, surely nice for entertainment purposes but certainly not worthy of even a third rate tech site.

With Carrizo pulled from the desktop. It seems AMD completely dropped the desktop segment and tries to focus while it can on mobile where its easier to get away slower CPUs.

Or they have another chip for DT, why wouldnt it be the case, it s surely much more probable than your quite ridiculous assumptions that Zen/K12 will be less performing than Carrizo or even Kaveri, that is, they ll re release Richland.?.

 

[cbn]

With Carrizo pulled from the desktop. It seems AMD completely dropped the desktop segment and tries to focus while it can on mobile where its easier to get away slower CPUs.

Actually, I tend to think of mobile (in some ways) being harder for AMD because in that segment Intel's node advantage is more useful for power savings and integration.

On desktop, if it weren't for cost even the older nodes would be fine. (In fact, Intel appears to believe this too and I think this is one reason we see mobile on 14nm first with desktop not adopting the new node for many months afterward. LGA 2011 style socket is even slower to adopt new nodes).

 

[piesquared]

Quoting for mods.

Here quote this too, "you fcking little pssy".


Well this got you a vacation
Markfw900

 

[Abwx]

Quoting for mods.

You should seriously ask yourself about the value you are bringing in this forum, so far if i check your post history i can clearly see that you write few relevant posts, most are single line posts that have no argument at all, only posted to create some tension within a preferably AMD related thread, you managed to be on point recently and i aknowledged it as well as the fact that it didnt often occur, now can you stay on a civilized and purely technical debate or should i put you on ignore list, make your choice with your next post.


Insulting other members is not allowed.
Markfw900

 

[positivedoppler]

Your slide is 8 months old. And 28nm came from Kumar quite recently.

So are you saying Kumar lied? Or just that you know better than him?

http://seekingalpha.com/article/273...nductors-software-and-supply-chain-conference

Shintai, why are you trying to troll a response from a statement you know is false interpretation of the facts?

This subject was discuss a month ago.
http://forums.anandtech.com/showthread.php?t=2411975&page=5

You continue to follow that thread after
NTMBK already point out the 28 nm rumor to be false and I'm sure that did not escape your notice. Yet you don't miss a chance to draw an AMDtard into your vortex.

Ashraf emailed AMD. Apparently Devinder mis-spoke, and K12 is on a FinFET process and not 28nm: http://semiaccurate.com/forums/showthread.php?t=8379

(Sorry to steal your thunder Ashraf )

 

[erunion]

Delete please. Accidentally posted.

 

[Markfw]

You guys keep this up, and we will have a record for the most infractions in a thread

 

[maarten12100]

Honestly, just give up with AMD. Not a single AMD cpu as of now can rival the 5 years old 980x. They are just so miserable.....

Not for mobile. Efficiency is great and only getting better and better. And I still have hope that the new Zen core design fixes the cpu deficit.

You guys keep this up, and we will have a record for the most infractions in a thread

So let's not keep this up all of a sudden 2 guys flying to each others throat over data that can only be qualified as a rumour.

Anyhow back on topic I'm really disappointed that AMD stated on their Newsroom page that they would show Carrizo stuff then only showed a wimpy piece of plastic that they allegedly put Carrizo into. I don't care give me performance and power numbers plz.

 

[mrmt]

Anyhow back on topic I'm really disappointed that AMD stated on their Newsroom page that they would show Carrizo stuff then only showed a wimpy piece of plastic that they allegedly put Carrizo into. I don't care give me performance and power numbers plz.

I don't think anyone should expect Carrizo to be anything better than a power optimized, low TDP Kaveri.

This is a very interesting twist, given that two or three quarters ago AMD was claiming that their most promising growth opportunities were in the desktop arena, the very market bracket that will be left in the cold until, and if, they can deploy a competitive chip with Zen.

 

[maarten12100]

I don't think anyone should expect Carrizo to be anything better than a power optimized, low TDP Kaveri.

This is a very interesting twist, given that two or three quarters ago AMD was claiming that their most promising growth opportunities were in the desktop arena, the very market bracket that will be left in the cold until, and if, they can deploy a competitive chip with Zen.

They made it a true SoC since the chipset is now moved onto the die. The last time they did this with Kabini -> Beema efficieny doubled.

"In terms of power draw, AMD expects to match Broadwell-U's "full dynamic range," with TDPs ranging up to 35W. AMD didn't share a lower limit, but it did note that Carrizo won't quite manage to slip into the 5-6W TDPs required for fanless tablets."
Despite they saying this I have no doubt you can easily make it fanless actually a little extra engineering and you can make high tdp chips fanless. Beema total platform power is 4W the Beema SoC would be about using half that idle. I don't expect huge perf gains over Kaveri which I have a notebook of here I just expect huge gains in efficiency.

But I can't make a decisions without numbers.

 

[mrmt]

They made it a true SoC since the chipset is now moved onto the die. The last time they did this with Kabini -> Beema efficieny doubled.

Source?

"In terms of power draw, AMD expects to match Broadwell-U's "full dynamic range," with TDPs ranging up to 35W. AMD didn't share a lower limit, but it did note that Carrizo won't quite manage to slip into the 5-6W TDPs required for fanless tablets."

I really expect them to be another case of AMD over-promising and under-deliverying, but even if they deliver what they are promising, they will still have a huge cost handicap because of the huge die. The more I read about Carrizo, the more I ask: Why did they bother?

 

[maarten12100]

Source?

Carrizo will be AMD’s first APU to implement a full Fusion Control Hub (FCH) — what we used to call a “south bridge” — on-die, but it’s doing so at some cost to overall features. The FCH will be a stripped-down variant, with two SATA 6G ports, four USB 3.0 ports, and four USB 2.0 ports. This will only apply to mobile variants — drop a Carrizo APU into the FM2+ socket, and the onboard FCH will disable itself while the motherboard FCH takes over.
Up until now, all AMD’s southbridges have been built on 65nm. Bringing the FCH on-die means that Carrizo will save a few precious watts for mobile form factors in that fashion, while slashing the total feature size cuts die size and power consumption without boosting the APU’s physical area much.

I really expect them to be another case of AMD over-promising and under-deliverying, but even if they deliver what they are promising, they will still have a huge cost handicap because of the huge die. The more I read about Carrizo, the more I ask: Why did they bother?

AMD marketing track record is indeed not that great. Nevertheless they from time to time put out amazing products and since they claimed "biggest leap from en energy perspective" I think Carrizo will be one of those amazing products. The only problem is that it will arrive a bit late.

As for the cost 28nm node is dirt cheap since we are 3 years down the line also the APU remained the same size ~245mm^2. Despite the huge ramp in density it should be a cheap to produce die.

 

[mrmt]

AMD marketing track record is indeed not that great. Nevertheless they from time to time put out amazing products and since they claimed "biggest leap from en energy perspective" I think Carrizo will be one of those amazing products. The only problem is that it will arrive a bit late.

As for the cost 28nm node is dirt cheap since we are 3 years down the line also the APU remained the same size ~245mm^2. Despite the huge ramp in density it should be a cheap to produce die.

First, I asked for a source for Beema's double efficiency quote. Except the misleading perf/TDP, I didn't see anything about it anywhere.

Second, the same 245mm^2 die puts AMD squarely in the same situation that has been making them bleed share for Intel in the APU segment, which is selling cheaper chips which are twice the die size compared to Intel's . Selling a SoC won't matter here, because Broadwell-U is also a SoC, so the net effect in terms of cost handicap is 0.

 

[ShintaiDK]

They made it a true SoC since the chipset is now moved onto the die. The last time they did this with Kabini -> Beema efficieny doubled.

Kabini already got the chipset integrated.

 

[maarten12100]

First, I asked for a source for Beema's double efficiency quote. Except the misleading perf/TDP, I didn't see anything about it anywhere.

Second, the same 245mm^2 die puts AMD squarely in the same situation that has been making them bleed share for Intel in the APU segment, which is selling cheaper chips which are twice the die size compared to Intel's . Selling a SoC won't matter here, because Broadwell-U is also a SoC, so the net effect in terms of cost handicap is 0.

Ah sorry you also quoted that first part so I thought you wanted a source for that.

Anyhow:
Kabini a4-5000:
http://www.notebookcheck.net/Review-HP-Pavilion-15-n050sg-Notebook.106490.0.html
Beema A8-6410:
http://www.notebookcheck.net/HP-Pavilion-17z-Notebook-Review.124669.0.html

The 15 inch Kabini is much slower and consumes more power than the 17 inch Beema notebook.

Kabini already got the chipset integrated.

Didn't know that guess it's just the superior GF process that made Beema so much better over Kabini then. Either way SoC will shave of a few watts which is big for a mobile chip. The race is no longer about highest performance the race is all about who can offer mediocre performance at the lowest power consumption.