so the enthusiast is still a business or rather a workstation cpu?
what clockspeed?
LGA for Enthusiast and BGA for Server.
Workstation, not on par with MCM Interlagos/Abu Dhabi but more on par with Valencia/Zurich/Seoul/Delhi and MCM Warsaw.so the enthusiast is still a business or rather a workstation cpu?
My guesses:what clockspeed?
http://www.anandtech.com/show/6751/amd-reiterates-2013-gpu-plans-sea-islands-beyondIt's more likely this was planned but was cancelled during one of the job cuts... hence the roadmaps showing nothing past Warsaw.
16c - 3.2 GHz Base / 3.4 GHz P1 / 3.7 GHz P0
It's dependent on the improvements.do you think this might work in place of a 8350 for a wide variety of tasks
I'm putting down as good enough probably if you disabled cores you can get to 5 GHz easy.and what about overclocking
I'm putting down as good enough probably if you disabled cores you can get to 5 GHz easy.
Can we expect 8-threaded 65W Carrizo in H1 2015?Basilisk is cancelled and was not an APU project but a Steamroller based server part/FX part. Basilisk equivalent on APU side is Kaveri, which already launched.
Gecko(8 module huge chip?) was to be a server part/FX part based on Excavator core but that one is ditched also. APU equivalent is already known to be Carrizo. Excavator is supposed to bring more threads per module capability, some form of CMT/SMT hybrid (although Bulldozer already has it since FP unit is SMT) - so 4 threads per dual "core" module. Oh and AVX2+BMI2 capability.
I honestly don't know. But what I wrote in previous post is coming from a good source. I suppose it's possible but 65W would imply a lot lower clock speed than what we have even with Kaveri, so any supposed gains would be probably partially negated by lower clocks. Power efficiency will probably go up, which is always a good thing.Can we expect 8-threaded 65W Carrizo in H1 2015?
It's a shame that Excavator based server/FX parts are cancelled. They would be a lot more competitive with intel's product line than PD/SR could ever be.
Excavator was going to be on 20nm GlobalFoundries and 20nm TSMC. Till someone on linkedin posted Carrizo was going to be on GF28A.Excavator would have to be competitive against Broadwell, and on top of that it would have to be competitive enough to the point of make up for the node difference (28nm vs 14nm).
I don't think the Excavator design could be that competitive, no matter how much efforts AMD has put on it. And given what we are seeing from Kabini parts, Globalfoundries 28nm seems very second class when compared to TSMC 28nm. That should also have added to AMD's pain.
Excavator was going to be on 20nm GlobalFoundries and 20nm TSMC. Till someone on linkedin posted Carrizo was going to be on GF28A.
I pretty much doubt that AMD would have designed their big core line for both TSMC and GLobalfoundries.
Excavator was inside their pipeline since 2011, and Globalfoundries roadmap at the time was very aggressive, plus Globalfoundries had an AMD-specific process in R&D pipeline until the end of 2012. That process is probably the 20nm the guys are referencing in their linkedin, which should have been based in IBM 22nm SOI process.
If anything, fielding this product on two different foundries would have complicated things immensely, because we would be talking about split in two an order volume that wasn't too big to start with, and AMD wouldn't get all the benefits with TSMC of whatever R&D they were paying Globalfoundries to make.
AMD has been pushing automated tools for CPUs more in recent years, which should in theory lower the costs of a port. Though I agree it's pretty unlikely for such a low volume part. They could justify porting Kabini because it piggybacked on the work done to port the console chips.
Kabini wasn't backported to Globalfoundries because of a sound business decision. The cat family was supposed to be with Globalfoundries since Krishna and Wichita, and it was only because of Globalfoundries inability to deliver a functional 28nm node that stuck with TSMC for the cat products.
After this turn of events, the cat family was supposed to stay with TSMC, but after the 2012 WSA amendment, they were forced to backport Kabini and Jaguar back to Globalfoundries.
Or alternatively, they chose to port Kabini and the console APUs to GlobalFoundries to free up more TSMC wafers for GPUs. We have just as much evidence to support either hypothesis.
Kabini wasn't backported to Globalfoundries because of a sound business decision. The cat family was supposed to be with Globalfoundries since Krishna and Wichita, and it was only because of Globalfoundries inability to deliver a functional 28nm node that stuck with TSMC for the cat products.
After this turn of events, the cat family was supposed to stay with TSMC, but after the 2012 WSA amendment, they were forced to backport Kabini and Jaguar back to Globalfoundries.
GPU is FCF breakeven and console apus has too small FCF. The cat core is where AMD is making money.
If AMD were to pick a product line to move to a bleeding edge node, why would they pick the worst of the three?
Gaaa. The only upside is delivery security and risk diversion here. And ofcource its of utmost importance for the consoles because of game sales.
Delivery security? Risk diversion? It's quite the opposite. AMD is tied to whatever glf is able to delivery. In fact, tge wsa is worsening the risks here.
It was already posted and is/was being discussed in the other Carrizo thread.Good for you...
20 clients of SK Hynix are/were/is interested in HBM.Approximately 100 people from 20 major clients and partners who are leading relevant industries participated in the symposium to show their attention to the Company’s HBM technology. With this symposium, the Company is looking forward to strengthening its cooperation with clients of various applications to expand the HBM ecosystem.