New AMD low power X86 core, enter the Jaguar

[Sxotty]

Car PC?

Yeah I was hoping for a lot more choices, but this is ok at least. I had wanted the 5000 not the 5200 so I could save 10watts.

 

[Alexko]

But what's a car PC? Apart from a computer in a car. I mean how do you use it?

Depending on what you're trying to build exactly, this might be a good fit: http://techreport.com/news/25702/zotac-intros-amd-kabini-powered-mini-pc (featuring the 5000 model). Otherwise underclocking and undervolting the 5200 should work too.

 

[Sxotty]

But what's a car PC? Apart from a computer in a car. I mean how do you use it?

Depending on what you're trying to build exactly, this might be a good fit: http://techreport.com/news/25702/zotac-intros-amd-kabini-powered-mini-pc (featuring the 5000 model). Otherwise underclocking and undervolting the 5200 should work too.

How you use it is limited by your imagination and time you have on your hands I guess. Anyway that is not a bad idea. If it is a standard motherboard in the zotac I might be very tempted to get one as it would probably be the cheapest. I would have to either be able to attach a normal power supply to use the automotive supply I have, or make a new circuit that turns the computer on and off with the car (by temporarily connecting the circuit to the power button when the car is turned on or off).

 

[AnarchX]

"Diffused in Germany"

http://www.computerbase.de/bildstrecke/56283/2/

Printing error or Kabini was moved to Global Foundries?

 

[Thorburn]

"Diffused in Germany"

http://www.computerbase.de/bildstrecke/56283/2/

Printing error or Kabini was moved to Global Foundries?

Diffused != Fabbed?

 

[AnarchX]

Diffused = fabbed.

The big AMD APUs have also this printing.
A bit strange is also that this Kabini chip is packaged/made in Taiwan (= TSMC!?), while the big APUs and other CPUs at AMD are packaged in Malaysia.

 

[3dilettante]

There could be a different entity responsible for this product's package assembly besides TSMC in Taiwan. Since other Jaguar chips say they are diffused and made in Taiwan, it might make sense to use existing package assembly right now.

 

[Exophase]

Diffused = fabbed.

The big AMD APUs have also this printing.
A bit strange is also that this Kabini chip is packaged/made in Taiwan (= TSMC!?), while the big APUs and other CPUs at AMD are packaged in Malaysia.

Diffusion would appear to be a separate step from fabrication.

http://www.siliconfareast.com/diffusion.htm
http://www.dowaytech.com/en/d/file/products/eCompass/2014-01-08/Silicon_Wafer_Processing.pdf

 

[3dilettante]

Diffusion in this case is a phase of the lithography process that is also being used as a general label for the fab's part of the process of creating a wholly functional product.

 

[Exophase]

Diffusion in this case is a phase of the lithography process that is also being used as a general label for the fab's part of the process of creating a wholly functional product.

That's pretty confusing - do you have a reference on this?

 

[3dilettante]

Referencing a discussion in this forum on the same topic would include posts by me.

Diffused in this case is shorthand for the fabbing of the wafer.

IBM discusses some of the requirements for the diffused and made in labels, and it distinguishes between diffusion and the bonding and packaging steps that involve taking a die and affixing it to something that makes it usable.

http://www-03.ibm.com/procurement/proweb.nsf/contentdocsbytitle/United+States~Country+of+origin+marking+instructions?OpenDocument&Parent=Global+Procurement

A Google search brought up this (odd thing about this link, it might try to print something):
http://rulings.cbp.gov/detail.asp?ru=w968421&ac=pr

This clearly differentiates between the diffusion step and cutting things into separate die and testing/mounting them.

There may be subtleties to customs enforcement or taxation reasons for doing this. Until the chip is fully packaged, it's not really a functional product.

 

[liolio]

http://www.realworldtech.com/jaguar/

 

[Deleted member 13524]

https://ad.moontoast.com/7e9f5f0e-c...mpid=social_20140423_22532994&optimize=direct

Beema SoC announced within 3 days with the new Puma cores.

 

[kalelovil]

http://techreport.com/review/26377/a-first-look-at-amd-mullins-mobile-apu

So, what's changed since the previous generation? From a CPU and GPU architecture standpoint, not much. The diagram above refers to "Puma+" CPU cores, but Mullins and Beema are actually based on the same Jaguar CPU microarchitecture as their predecessors. No architectural changes have been made to boost instructions per clock, we're told. AMD has, however, taken a number of steps to raise clock speeds when appropriate and to boost power efficiency. Some of those steps extend to the integrated Graphics Core Next GPU. In all, the transistor count has risen from 914 million to 930 million transistors due to "low level changes." AMD has also made improvements to the memory and display interfaces, and it's enabled an ARM Cortex-A5 core that was present but disabled in Temash and Kabini.

 

[Wynix]

http://techreport.com/review/26377/a-first-look-at-amd-mullins-mobile-apu

I think this is worth quoting:

Thanks to a mix of circuit power optimizations and process scaling improvements, Mullins and Beema reduce power leakage substantially compared to the previous generation. AMD claims to have achieved a 19% leakage reduction across the CPU cores and a 38% leakage reduction in the integrated GPU. Less leakage means less energy wasted as heat, which in turn means AMD was able to raise clock speeds within the same thermal envelopes. The fastest 15W chip from last year's Kabini lineup ran at 1.55GHz with a 500MHz GPU speed, but the fastest 15W Beema chip runs at 2GHz, can hit 2.4GHz with Turbo Core, and clocks its GPU at 800MHz. That's quite impressive, considering both processors are manufactured on the same 28-nm process.

That is impressive.

There's a lot of useful info in that article.

 

[kalelovil]

I think this is worth quoting:
That is impressive.

There's a lot of useful info in that article.

It probably is impressive, but I'd wait until AMD allow real-world power consumption figures to be produced before coming to any firm conclusions.

Remember that the Kaveri 7600T in 45W mode looked incredibly impressive next to its 45W Richland predecessor, but this was diminished somewhat by the 45W Kaveri consuming a significant amount more power than the 45W Richland.

 

[lanek]

Thanks for the link... The results shown on the Mullin seems good..

 

[Alexko]

It probably is impressive, but I'd wait until AMD allow real-world power consumption figures to be produced before coming to any firm conclusions.

Remember that the Kaveri 7600T in 45W mode looked incredibly impressive next to its 45W Richland predecessor, but this was diminished somewhat by the 45W Kaveri consuming a significant amount more power than the 45W Richland.

True, but so long as the TDP is respected, that's OK. Chances are that Temash/Kabini SKUs were often operating well below their TDPs, leaving performance on the table. Now Beema/Mullins should be able to operate very close (or temporarily above) their TDPs. To be sure, effective performance/W is unlikely to have increased as much as AMD claims, but so long as the performance is there and the TDP is respected, it's not an issue. AMD also claims that getting things done quicker and going to sleep/idle saves energy (and therefore battery life).

That said, it doesn't change the fact that there does not seem to be a whole lot of interest in x86 tablets. I also wonder why AMD hasn't released a 25W variant of Beema. Perhaps they believe this segment would be better served by the mobile Kaveri, but since the latter has yet to show up…

It's also worth noting that this new SoC is made by GloFo, which means their process must be quite good.

Edit, after reading The Tech Report's article (http://techreport.com/review/26377/a-first-look-at-amd-mullins-mobile-apu/ [benchmarks on pages 3 and 4]): looks pretty good.

Our time with the Discovery tablet has answered a lot of our questions—but it's also left us with many more.

We learned that, for a tablet processor, Mullins is quite fast. It performs comparably in many tests to last year's A4-5000, a 15W part that landed in mainstream notebooks, and its graphics performance outpaces that of the competition from Intel and Nvidia. (To be fair, the Atom Z3740 we tested isn't the fastest Bay Trail incarnation—but it does have the same GPU speeds as the flagship Z3770.

 

[sebbbi]

Intel has reduced their TDP impressively by improving the clock management and monitoring (huge improvements from Nehalem->Sandy->Ivy->Haswell). They can shut down / restore the chip parts in a very fine grained manner (and very quickly). AMD did something similar for R9 290X, and it seems that they now are integrating similar systems to their new APUs. Excellent news indeed.

 

[Lightman]

@Alexko

AMD will be releasing 25W Beema, but it will be only to select OEM who requested it.
It makes sense to reuse already developed 25W laptop platform and ask AMD to deliver better performance at the same time.

These special 25W(?) chips will be called A8-xxxx
From TechReport

Happily, Beema may have an easier time working its way into the marketplace. Lenovo has already announced several laptops based on it: the Flex 2 and the B and G series. Those systems are coming in June, and they'll feature a special, A8-branded version of Beema that AMD offers "for select opportunities." (That model isn't part of the standard Beema lineup.)

There is also that possibility that I'm reading too much into this sentence