• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

AMD Launches Three Kaveri APU SKUs in February 2014 – Feature Set For A10 and A8 APUs Detailed

Davedree said:
Or the 28nm process is a pooper and you'll hit a brick wall at 4.1 ghz ;)
Ooh sorry i'm so negative with Amd these days, must be too many years of disapointments.
Click to expand...

The GF 28NM process is bulk as opposed to the 32NM SOI process they are using now. Density will of course go up,but OTH I am getting the impression clockspeeds have not. AMD really cannot do much ATM about it as they are too small a company to own their own fab and develop process nodes which keep up with Intel,and even though GF has the second best large volume node for big core CPUs it is quite behind what Intel has. IBM themselves are using a 22NM SOI process for their POWER chips,but I get the impression it is quite expensive and small volume.

Edit!!

You can see the 28NM node probably has issues as the AMD performance estimates were higher when leaked months ago. The node has had big delays and GF only started producing small ARM based SOCs on it last year. I suspect,like with Llano,Kaveri has missed its clockspeed targets. That is three out of the last four process nodes GF has had problems with it seems.
 
Last edited:
CAT-THE-FIFTH said:
The GF 28NM process is bulk as opposed to the 32NM SOI process they are using now. Density will of course go up,but OTH I am getting the impression clockspeeds have not. AMD really cannot do much ATM about it as they are too small a company to own their own fab and develop process nodes which keep up with Intel,and even though GF has the second best large volume node for big core CPUs it is quite behind what Intel has. IBM themselves are using a 22NM SOI process for their POWER chips,but I get the impression it is quite expensive and small volume.

Edit!!

You can see the 28NM node probably has issues as the AMD performance estimates were higher when leaked months ago. The node has had big delays and GF only started producing small ARM based SOCs on it last year. I suspect,like with Llano,Kaveri has missed its clockspeed targets. That is three out of the last four process nodes GF has had problems with it seems.
Click to expand...

GF have been a never end source of AMD delays and problems, for years now, there must be better foundries out there that AMD could use.

In anycase i don't think overclocking these CPU's can be any worse than Haswell.

We will find out soon enough.
 
CAT-THE-FIFTH said:
The GF 28NM process is bulk as opposed to the 32NM SOI process they are using now. Density will of course go up,but OTH I am getting the impression clockspeeds have not. AMD really cannot do much ATM about it as they are too small a company to own their own fab and develop process nodes which keep up with Intel,and even though GF has the second best large volume node for big core CPUs it is quite behind what Intel has. IBM themselves are using a 22NM SOI process for their POWER chips,but I get the impression it is quite expensive and small volume.

Edit!!

You can see the 28NM node probably has issues as the AMD performance estimates were higher when leaked months ago. The node has had big delays and GF only started producing small ARM based SOCs on it last year. I suspect,like with Llano,Kaveri has missed its clockspeed targets. That is three out of the last four process nodes GF has had problems with it seems.
Click to expand...


Like you say Amd need their own foundry again, but thats never going to happen. Gf have hampered the progress of Amd recently. Krishna and Witchita should have been on 28nm-SHP, Richland should have been on 28nm but at the time gf had issues with their 28nm. It seems they were stuck in between tsmc which provided 28nm but at low clock speeds hence Temash and kabini were low clocked and entered into the markets accordingly, or to respin on the long in the tooth 32nm. It is my belief and opinion that as they were stuck on the 32nm, Richland could have offered a gcn derived gpu if Amd had been on a 28nm process. Thus Kaveri would have then been a stronger performing product than it is now,, But then again they haven't addressed the memory bandwidth issues even now.
Still I will be buying Kaveri to have a play with in itx size.

Ibm 22nm has two problems low yield products and a high price. But as you know it would be useless for Amd that targets the oem sector and needs a high yield low cost production.

I'm not hugely clued up on the fab work, but how did Amd manage to transpose lithiography from the 28 tsmc process of their gcn pitcairn into the gf 28nm process.

Edit: I wonder if the clocks are reduced due to instability/brick wall or to reduce the power consumption of the soc. These are things that can be assessed when they make it to shelves I suppose. Also a concern is that they havent hinted at the mobile kaveri platform at all, unless they are being very quiet, i'm worried that maybe the tdp/prformance won't scale well in a notebook format, compared to richland maybe?
 
Last edited:
CAT-THE-FIFTH said:
The GF 28NM process is bulk as opposed to the 32NM SOI process they are using now. Density will of course go up,but OTH I am getting the impression clockspeeds have not. AMD really cannot do much ATM about it as they are too small a company to own their own fab and develop process nodes which keep up with Intel,and even though GF has the second best large volume node for big core CPUs it is quite behind what Intel has. IBM themselves are using a 22NM SOI process for their POWER chips,but I get the impression it is quite expensive and small volume.

Edit!!

You can see the 28NM node probably has issues as the AMD performance estimates were higher when leaked months ago. The node has had big delays and GF only started producing small ARM based SOCs on it last year. I suspect,like with Llano,Kaveri has missed its clockspeed targets. That is three out of the last four process nodes GF has had problems with it seems.
Click to expand...

I hear IBM only need 1 solid sale (read mainframe) to make a whole wafer profitable. OTOH even Intel will begin to feel the pinch, with a somewhat floundering market and increasing difficulty to fab at smaller nodes. If, and it's a big if, AMD can hold on til 2017-2020, I reckon we could see process parity. Hopefully X86 is long dead in the mainstream by then, and we can see what both camps can produce without all this IP license bull
 
Your original link was the game chart and I read it as you saying it was mantle performance :p

I'ma take the charts with a grain of salt, because Mantle's doing over 300% there as it's FPS when it's just the Kaveri.
 
Last edited:
You must log in or register to reply here.
Back
Top Bottom