AMD CPU speculation... and expert conjecture

[jdwii]

That last "titbit" isn't exactly news.

Nor is the fact that AMD are going to release a new gen gpu at some point sooner or later. I mean, no s--t, Sherlock.

What would be news on that front would be whether they expect to be competitive on performance instead of just price. But he'd hardly admit it if he knew it wouldn't be, would he?

Good to see freesync being adopted though. Hope it's a reputable brand.

That's expected if its made on 20nm and if its based around the newer design or the design built in the 285 its going to kill the 980 its not like the 980 brings performance to a new level i mean its amazing but not in a performance killer way. Maybe Nvidia is waiting to produce their 980Ti on 20nm that would be cool but i'm not sure.

 

[blackkstar]

GlobalFoundries will Acquire IBM Chip Business
http://www.guru3d.com/news-story/globalfoundries-will-acquire-ibm-chip-business.html

This is huge. AMD has traditionally used IBM high performance nodes in their products or at the very least got a big helping hand from IBM, but 22nm was completely absent from AMD. And so were the high performance products like HEDT CPUs.

GloFo might end up offering different versions of nodes and we might, if we're really lucky, see HEDT AMD CPUs on FD-SOI FINFET (in before flamewar about SOI).

This is good news for AMD if you ask me. It opens up a lot of doors for them. Specially if they can get their GPUs on a superior process to Nvidias, they have potential to really put some hurt on them. AMD really needs that goign by last quarter's results.

 

[Embra]

Sadly GoFo has disappointed in the past, but I too hope this bodes well for AMD.

 

[etayorius]

AMD does not need GloFo, now with nVidia shitting all over Samsung they have a reason to support AMD.

 

[noob2222]

I wonder if that move has anything to do with AMD being able to get out from under gf's contract requirements, allowing them to move fully with Samsung while IBM goes to GF.

I don't see GF being big on soi, they trashed their 28nm attempts, only producing some test chips for st-ericsson.

 

[8350rocks]

Ahh, the biggest victory for GloFo is the IBM engineers that TSMC was trying so hard to steal from IBM.

As for competing in performance, colinp, you are quick to forget that it was the 290X that forced NVidia to come to the table with something more meaningful. I am unable to give estimates (NDA), but judging by numbers thrown at me, this will make a meaningful impact. Let us just say that there will probably not be an answer from maxwell that can compete, even a big die maxwell.

 

[esrever]

IBM payed glofo to take their silicon business. I wouldn't write it off as a win of any sort.

 

[szatkus]

GlobalFoundries will Acquire IBM Chip Business
http://www.guru3d.com/news-story/globalfoundries-will-acquire-ibm-chip-business.html

This is huge. AMD has traditionally used IBM high performance nodes in their products or at the very least got a big helping hand from IBM, but 22nm was completely absent from AMD. And so were the high performance products like HEDT CPUs.

It's not IBM's nodes.
http://www.commonplatform.com/

GloFo might end up offering different versions of nodes and we might, if we're really lucky, see HEDT AMD CPUs on FD-SOI FINFET (in before flamewar about SOI).

This is good news for AMD if you ask me. It opens up a lot of doors for them. Specially if they can get their GPUs on a superior process to Nvidias, they have potential to really put some hurt on them. AMD really needs that goign by last quarter's results.

It means nothing for AMD. At least for short term (e.g. 2-3 years).

 

[szatkus]

Ahh, the biggest victory for GloFo is the IBM engineers that TSMC was trying so hard to steal from IBM.

I'm not sure which engineers you mean. IBM kept their R&D staff, they only sold fabs.

As for competing in performance, colinp, you are quick to forget that it was the 290X that forced NVidia to come to the table with something more meaningful. I am unable to give estimates (NDA), but judging by numbers thrown at me, this will make a meaningful impact. Let us just say that there will probably not be an answer from maxwell that can compete, even a big die maxwell.

So 20nm. I don't see another explanation.

 

[8350rocks]

Ahh, the biggest victory for GloFo is the IBM engineers that TSMC was trying so hard to steal from IBM.

I'm not sure which engineers you mean. IBM kept their R&D staff, they only sold fabs.

As for competing in performance, colinp, you are quick to forget that it was the 290X that forced NVidia to come to the table with something more meaningful. I am unable to give estimates (NDA), but judging by numbers thrown at me, this will make a meaningful impact. Let us just say that there will probably not be an answer from maxwell that can compete, even a big die maxwell.

So 20nm. I don't see another explanation.

Where did you read that none of the staff came with the fab?

 

[szatkus]

Ahh, the biggest victory for GloFo is the IBM engineers that TSMC was trying so hard to steal from IBM.

I'm not sure which engineers you mean. IBM kept their R&D staff, they only sold fabs.

As for competing in performance, colinp, you are quick to forget that it was the 290X that forced NVidia to come to the table with something more meaningful. I am unable to give estimates (NDA), but judging by numbers thrown at me, this will make a meaningful impact. Let us just say that there will probably not be an answer from maxwell that can compete, even a big die maxwell.

So 20nm. I don't see another explanation.

Where did you read that none of the staff came with the fab?

R&D staff != fab staff.

 

[colinp]

8350rocks, the next time you speak to Huddy, instead of unquestionably swallowing everything he has to say like the obedient red team member you are, perhaps you could ask him why AMD has failed to make any laptop design wins for Kaveri or current gen, high end GPUs.

 

[8350rocks]

8350rocks, the next time you speak to Huddy, instead of unquestionably swallowing everything he has to say like the obedient red team member you are, perhaps you could ask him why AMD has failed to make any laptop design wins for Kaveri or current gen, high end GPUs.

AMD has design wins, ask the OEMs where they are.

That was actually something I have pointed out to multiple members of AMD upper management. They are working on getting more of these design wins out.

Also, for a high end Kaveri/GPU laptop win, look at the GX70 MSI.

 

[con635]

I am unable to give estimates (NDA), but judging by numbers thrown at me, this will make a meaningful impact. Let us just say that there will probably not be an answer from maxwell that can compete, even a big die maxwell.

I actually have a feeling thats going to happen.

 

[colinp]

You must have a different definition of "win" to me.

Right now, if I want to stay AMD when replacing my Llano laptop then my only choice would be a Richland one. Not a single Kaveri available.

 

[logainofhades]

You must have a different definition of "win" to me.

Right now, if I want to stay AMD when replacing my Llano laptop then my only choice would be a Richland one. Not a single Kaveri available.

Then you must not be looking very hard. I went to Newegg and found an A10 and two A8 kaveri based laptops in less than 5 mins.

 

[8350rocks]

MSI GX70 has kaveri available...not sure what you are smoking? Top mobile SKU at the moment is kaveri and MSI offers that SKU in their top end laptop.

Also, Asus has a nice laptop around $700-800 with 1080p screen and top mobile SKU.

EDIT: IBM engineers will be working with GloFo as part of the deal in a research partnership, GF get exclusivity for IBM silicon, and access to patents. Meaning FD-SOI nodes will be available.

 

[8350rocks]

You must have a different definition of "win" to me.

Right now, if I want to stay AMD when replacing my Llano laptop then my only choice would be a Richland one. Not a single Kaveri available.

Then you must not be looking very hard. I went to Newegg and found an A10 and two A8 kaveri based laptops in less than 5 mins.

There is some sanity in the world after all it seems.

 

[colinp]

None in the UK from what I've seen. They're all Trinitys, Richlands or Mullins.

 

[logainofhades]

None in the UK from what I've seen. They're all Trinitys, Richlands or Mullins.

Again, you are not trying very hard.

http://www.amazon.co.uk/s/ref=sr_nr_n_5?rh=n%3A340831031%2Cn%3A429886031%2Ck%3Aa10+7300&keywords=a10+7300&ie=UTF8&qid=1413902618&rnid=340832031

http://www.amazon.co.uk/s/ref=nb_sb_noss_2?url=node%3D429886031&field-keywords=A10+Pro-7350B&rh=n%3A340831031%2Cn%3A429886031%2Ck%3AA10+Pro-7350B

http://www.amazon.co.uk/s/ref=nb_sb_noss_2?url=node%3D429886031&field-keywords=++A8+Pro-7150B&rh=n%3A340831031%2Cn%3A429886031%2Ck%3A++A8+Pro-7150B

 

[colinp]

Strange, I DID search for laptops on Amazon using the "AMD" filter and it returned none of them. Do they really expect people to search by the full processor id?

Thanks, though.

 

[logainofhades]

Amazon's search kinda sucks, to be honest. Found them quicker on Ebuyer, just going through the search links. For whatever reason, finding hardware is a talent of mine, though.

http://www.ebuyer.com/store/Computer/cat/Laptops?a00035=AMD+A10+Series

 

[blackkstar]

GloFo screwed up 28nm SOI. IBM has (had?) 22nm SOI. GloFo basically gets to throw 28nm SOI in the trash and get 22nm SOI.

IBM was also working on 14nm FINFET on SOI. AMD's fab options for HEDT CPUs has been a complete mess the last few years.

 

[8350rocks]

POWER8 is exclusively on 22nm FDSOI.

 

[anxiousinfusion]

AMD's new pricing for Kaveri APUs puts 7850K at $143. Pretty much taking Richland's old price point. Is it still considered "too expensive to justify"?

Also is this price drop any indication of future product announcements? http://www.tomshardware.com/news/amd-apus-pricing-gaming-a-series,27933.html