Counting the success of Kabini & Temash

[lefty2]

I've being doing a casual survey of design wins for Kabini and Temash.
It seems like most of the design wins were for the A-series (I counted about 20). The E series and Temash, however got very few design wins (4 and 3, respectively). That is puzzling for me, because the previous Brazos E-series was hugely successful.

 

[mrmt]

I've being doing a casual survey of design wins for Kabini and Temash.
It seems like most of the design wins were for the A-series (I counted about 20). The E series and Temash, however got very few design wins (4 and 3, respectively). That is puzzling for me, because the previous Brazos E-series was hugely successful.

Probably ARM and Atom are squeezing Jaguar from the bottom and that market bracket is dominated by tablet, a form factor where Temash is not well suited to.

 

[SiliconWars]

Yes that'll be why AMD's CPU revenue was up 12% mostly on Jaguar sales last quarter, while Intel's was up only 1.4% (and their tablet segment was down 3.7%). I think it's quite obvious who is being squeezed by ARM.

 

[mrmt]

Yes that'll be why AMD's CPU revenue was up 12% mostly on Jaguar sales last quarter, while Intel's was up only 1.4% (and their tablet segment was down 3.7%). I think it's quite obvious who is being squeezed by ARM.

Well, their gross margins didn't move after releasing Jaguar in a *very* mature node, and the 12% increase comes after a >35% drop since 2011, so I don't think things are good here, just within forecast.

As I said before, Jaguar is having a much harder time competing against Intel current line up, and we have yet to see whatever impacts caused by Silvermont and Celeron/Pentium Haswell. We already know that Jaguar is toast on the tablet market, let's see how it goes against Intel after the launches in Q3/Q4.

 

[SiliconWars]

Jaguar isn't having a hard time competing with Intel's current line-up, in fact AMD was up ~3% in mobile last quarter.

Now sure ASP will be down but that was what happened with Brazos as well. As AMD ramps Jaguar I would expect them to gain even more share in the second half of the year.

In terms of tablets, AMD isn't even trying to compete yet. That's why they are using a salvage die of a laptop chip fabbed on a high performance process. They will probably shift a few 100K units all year, which is fine because as the process is so mature they are probably deliberately castrating a few Kabini's just to make up the Temash's anyway.

 

[Exophase]

Maybe AMD isn't trying to compete yet in tablets but if you ask me that's where they should be trying hardest because they're never going to capture the same kind of sales they did with Brazos by targeting the same markets. The netbook market shrunk tremendously and HTPCs are being squeezed out by smart TVs, consoles, and HDMI dongles (now Google's). Thin laptops are also getting hit by Intel selling off some Celerons to some OEMs at aggressive prices and Haswell bringing huge idle efficiency improvements.

It's not surprising that they'd see some growth anyway when releasing a new product given that the one it's replacing is positively ancient now and dead in the water. Doesn't mean that they're selling as many as they hoped, I really don't see how they could be with the number of real devices Jaguar is showing up in so far :/

 

[mrmt]

Jaguar isn't having a hard time competing with Intel's current line-up, in fact AMD was up ~3% in mobile last quarter.

AMD two-year refresh that accounts for 40% of AMD mobile shipments only gets 3% in added mobile market share and 0% increase in gross margins? What do you expect for the next two years in the mobile market?

Sure, Jaguar margins might be better than Brazos', but with that would imply in AMD losing margins in their big core line, and that line isn't exactly a star here. Maybe someone will leak the breakdown of AMD product mix and we'll have a better idea on the subject.

 

[SiliconWars]

Well it's the first quarter of sales remember. It was a while back but if I recall correctly Ontario sales went something like 4 million in Q1 2011, 6 million Q2 2011, 8 million Q3 2011. After that I'm unsure but it was probably around 8 million in Q4 as well (AMD said they sold ~30 million in their first year I think).

So clearly there is plenty of room for Jaguar to grow. The market isn't great but there's no good reason why they wouldn't have similar sales increases over the coming 2-3 quarters at least.

As for tablets, both Intel and Nvidia are losing a truckload of money there so why wouldn't AMD? In an ideal world they'd have enough money to pay for a pure tablet design, but they just don't have it. It's not an important market for AMD right now, they'll get into it with an ARM core though I'm sure.

 

[Exophase]

Well it's the first quarter of sales remember. It was a while back but if I recall correctly Ontario sales went something like 4 million in Q1 2011, 6 million Q2 2011, 8 million Q3 2011. After that I'm unsure but it was probably around 8 million in Q4 as well (AMD said they sold ~30 million in their first year I think).

So clearly there is plenty of room for Jaguar to grow. The market isn't great but there's no good reason why they wouldn't have similar sales at least.

As for tablets, both Intel and Nvidia are losing a truckload of money there so why wouldn't AMD? In an ideal world they'd have enough money to pay for a pure tablet design, but they just don't have it. It's not an important market for AMD right now, they'll get into it with an ARM core though I'm sure.

Of course there's a good reason why they wouldn't have similar sales with Jaguar if they're selling it in the same stuff. You don't really think people are buying those things like they were in 2009 do you?

Interested in how Intel is losing a ton of money in tablets specifically. Their tablet chip (Clover Trail) is not an awful lot different from their prior phone chip (Medfield) or even their netbook chip (Cedar Trail), so it wouldn't seem like they're spending a bunch of money just on tablets. If they're losing it's probably across the board w/Atom, which is not good news for AMD because it competes in all the same places.

I'm not talking about doing anything extra special for AMD - the CPU and GPU designs are already paid for, will make enough money from consoles and so forth. Maybe even then they could have lost money on a tablet-oriented chip. But the way I see it this is the only area where they really have a potential advantage. They're the only ones who can offer an x86 tablet with competent GPU performance right now; the gap between Clover Trail's GPU and Temash's is immense. But yes, we don't know if enough buyers are even interested in that. All we know is that buyers lost a lot of interest in the stuff they bought Brazos for.

I agree, it sucks that AMD doesn't have money to make more chips. That goes for more than just the missing tablet chip. There's also the missing native 4/6 core FX chips and missing native 2 core Trinity/Richland APUs. Having one chip per product line that is meant to last 1 year if you're lucky (FX), 2 if you're not (Brazos and Trinity/Richland) is in stark contrast to the multitude Intel is pumping out. But yeah, what can you do if you don't have the cash?

 

[AtenRa]

AMD Kabini plays alone in the Laptop/desktop entry level segment from 15W up to the 25W, Intel doesnt have anything in that price/performance and 22nm ATOM will still have way lower iGPU performance and it should cost higher (bigger die on a higher cost 22nm process)

I dont have a lot of information as of now about Tablets, so ill wait and see a couple of months.

 

[mrmt]

I agree, it sucks that AMD doesn't have money to make more chips. That goes for more than just the missing tablet chip. There's also the missing native 4/6 core FX chips and missing native 2 core Trinity/Richland APUs. Having one chip per product line that is meant to last 1 year if you're lucky (FX), 2 if you're not (Brazos and Trinity/Richland) is in stark contrast to the multitude Intel is pumping out. But yeah, what can you do if you don't have the cash?

I don't think it's a matter of cash but a matter of ROI. AMD volumes are too low to pay back a specific SKU for all these market segments. They have to do with a single die.

 

[SiliconWars]

Of course there's a good reason why they wouldn't have similar sales with Jaguar if they're selling it in the same stuff. You don't really think people are buying those things like they were in 2009 do you?

No but they'll be getting more for Jaguar so the revenue totals should be fairly similar to what Brazos was bringing in.

Interested in how Intel is losing a ton of money in tablets specifically. Their tablet chip (Clover Trail) is not an awful lot different from their prior phone chip (Medfield) or even their netbook chip (Cedar Trail), so it wouldn't seem like they're spending a bunch of money just on tablets. If they're losing it's probably across the board w/Atom, which is not good news for AMD because it competes in all the same places.

Yeah well that's the point isn't it. They have a phone and tablet chip combined, should AMD be competing in phones as well, losing cash like the other two? Or should they just spend money they don't have on a pure tablet design? There isn't that much money in tablets unless you are Apple.

AMD is doing the right thing, while boring from a technology point of view you just have to look at the damage Tegra is doing to Nvidia to see that AMD would be sunk with a similar strategy.

I agree, it sucks that AMD doesn't have money to make more chips. That goes for more than just the missing tablet chip. There's also the missing native 4/6 core FX chips and missing native 2 core Trinity/Richland APUs. Having one chip per product line that is meant to last 1 year if you're lucky (FX), 2 if you're not (Brazos and Trinity/Richland) is in stark contrast to the multitude Intel is pumping out. But yeah, what can you do if you don't have the cash?

That's basically it. They have to pick the markets that they know will bring in revenue for them, there isn't any money for risks or entering the cut-throat markets like phones and tablets. They'd just lose money same as Intel and Nvidia are. ARM cores will help them here, but that's probably 2 years in the future at the earliest.

 

[SiliconWars]

I don't think it's a matter of cash but a matter of ROI. AMD volumes are too low to pay back a specific SKU for all these market segments. They have to do with a single die.

To be a winner in tablets you need to be in phones (and probably a winner in phones). The last thing AMD needs right now is fighting it out here, let the others do that and they can pick up some pieces later.

What AMD should be doing is licensing their GPU like Nvidia is doing, however.

 

[Exophase]

AMD Kabini plays alone in the Laptop/desktop entry level segment from 15W up to the 25W, Intel doesnt have anything in that price/performance and 22nm ATOM will still have way lower iGPU performance and it should cost higher (bigger die on a higher cost 22nm process)

You say that but then one just has to look at the Temash Acer Aspire V5 with the 17W Celeron one that costs $50 less (last I saw)..

Then you have mobile Richland which has 17W, 19W, and 25W variants. I don't know how much they cost compared to Kabini but if they performance is worse why would AMD even be bothering?

 

[monstercameron]

To be a winner in tablets you need to be in phones (and probably a winner in phones). The last thing AMD needs right now is fighting it out here, let the others do that and they can pick up some pieces later.

What AMD should be doing is licensing their GPU like Nvidia is doing, however.

Or just extend their custom apu department to offer arm/mobile offerings instead of licensing.

 

[SiliconWars]

Then you have mobile Richland which has 17W, 19W, and 25W variants. I don't know how much they cost compared to Kabini but if they performance is worse why would AMD even be bothering?

The 19W and 25W quad Richlands will be a bit better than Kabini, and a lot better graphically I'd imagine with 3x the SP's (even though they aren't GCN).

 

[Exophase]

Or just extend their custom apu department to offer arm/mobile offerings instead of licensing.

If they can't afford to do a tablet-oriented Jaguar SoC then they won't be able to afford an ARM based one either. It's the physical design process that costs them, hence why they can't make different versions of the same chip to knock off cores...

Actually this raises a big question in my mind, if a tablet SoC is a poor investment for them then is a big multicore Cortex-A57 micro-server only chip going to be a good investment? And they're even coming out with two chips here, an 8 core one and later a 16 core one. I don't think they're going to be tiny and I don't think they're going to be much cheaper to make than Temash/Kabini. Maybe if they're licensing hard macros.

 

[rainy]

Then you have mobile Richland which has 17W, 19W, and 25W variants. I don't know how much they cost compared to Kabini but if they performance is worse why would AMD even be bothering?

I would tend to agree with you about 17/19W Richlands but A10-5745M is not that case.
On CPU side is faster than A6-5200 and it's GPU is a different class.
128 SP vs 384 SP (600 Mhz vs 533/626 Mhz) plus dual channel IMC means huge advantage for Richland.

http://www.cpubenchmark.net/cpu.php?cpu=AMD+A6-5200+APU&id=1975
http://www.cpubenchmark.net/cpu.php?cpu=AMD+A10-5745M+APU&id=1962
http://www.notebookcheck.net/AMD-Radeon-HD-8400.93720.0.html
http://www.notebookcheck.net/AMD-Radeon-HD-8610G.93719.0.html

 

[lefty2]

...22nm ATOM will still have way lower iGPU performance and it should cost higher (bigger die on a higher cost 22nm process)

Is the die size of Silvermont even known at this stage? I would have thought that it's smaller than Kabini.

 

[monstercameron]

I would tend to agree with you about 17/19W Richlands but A10-5745M is not that case.
On CPU side is faster than A6-5200 and it's GPU is a different class.
128 SP vs 384 SP (600 Mhz vs 533/626 Mhz) plus dual channel IMC means huge advantage for Richland.

http://www.cpubenchmark.net/cpu.php?cpu=AMD+A6-5200+APU&id=1975
http://www.cpubenchmark.net/cpu.php?cpu=AMD+A10-5745M+APU&id=1962
http://www.notebookcheck.net/AMD-Radeon-HD-8400.93720.0.html
http://www.notebookcheck.net/AMD-Radeon-HD-8610G.93719.0.html

True but at the cost of a blown tdp budget. Notebook check has the 17w a6-4455m going over 40w in a Samsung notebook. Pile driver apus will exceed their tdp while jaguar will be relatively close.

 

[Exophase]

Is the die size of Silvermont even known at this stage? I would have thought that it's smaller than Kabini.

Atom SoCs have never been 110mm^2, I doubt Intel is going to start now.

 

[Sweepr]

110mm² is a little bit smaller than a 2C/4T 4MB L2 (GT2 16 EUs IGP) Ivy Bridge, no way Silvermont Atom is bigger.

 

[VirtualLarry]

As for tablets, both Intel and Nvidia are losing a truckload of money there so why wouldn't AMD? In an ideal world they'd have enough money to pay for a pure tablet design, but they just don't have it. It's not an important market for AMD right now, they'll get into it with an ARM core though I'm sure.

AMD ARM core, with Radeon graphics? Could be a hot seller. I personally prefer my devices x86, but I wouldn't mind a cell phone with Radeon inside.

 

[wlee15]

If they can't afford to do a tablet-oriented Jaguar SoC then they won't be able to afford an ARM based one either. It's the physical design process that costs them, hence why they can't make different versions of the same chip to knock off cores...

Actually this raises a big question in my mind, if a tablet SoC is a poor investment for them then is a big multicore Cortex-A57 micro-server only chip going to be a good investment? And they're even coming out with two chips here, an 8 core one and later a 16 core one. I don't think they're going to be tiny and I don't think they're going to be much cheaper to make than Temash/Kabini. Maybe if they're licensing hard macros.

To me it's all about Seamicro. The ARM A57 chips fills and expands in the classic microserver and large memory roles that are currently held by the Atom chips and the Piledriver Opeterons respectively.

 

[Exophase]

To me it's all about Seamicro. The ARM A57 chips fills and expands in the classic microserver and large memory roles that are currently held by the Atom chips and the Piledriver Opeterons respectively.

But that doesn't really answer the question: will it gross enough more than a tablet oriented SoC to justify the creation of the former over the latter?

In the tablet space AMD has unique value with their GPUs. In the micro-server space they lack that. They'll be competing with others like Calxeda who will also put Cortex-A57s on a server SoC, and worse, Applied Micro who is making their own 64-bit ARM specifically intended for servers. AMD/SeaMicro may have unique value in their interconnects and IP but it remains to be seen if they really offer much more than what the other server-oriented firms have..