EX core will surely be more complex, especially since AMD plans to give its FPU a native 256bit pipeline (2x of what SR has).
Resurrected from the closed thread:
But Kaveri is not a 2013 release, its a 2014 release.
AM3+ also hit the end of the line. Like in the server space there are no updates for it.
In related news, AMD takes another 500mio$ credit and layoffs may be coming:
http://vr-zone.com/articles/layoffs-coming-amd/63736.html
Also what product has been killed?
I'm not even sure we'll see Excavator at 20nm. In fact I wonder if the 14nm tapeout happening next quarter is Excavator at GF.
You serious? Excavator is most likely a 28nm part...no chance in hell that it's a FinFET.
I seriously doubt Excavator will be on 28nm. It'll probably wait for 20nm... however long it takes GloFo to get that working.
You serious? Excavator is most likely a 28nm part...no chance in hell that it's a FinFET.
I don't think that Excavator would be at 28nm - most probably 20nm and 1H 2015 as release date.
Edit - annoyingly the video has been removed, but that's what the board said.In other GF news, a Youtube video from DAC 2013 has some interesting stuff - http://www.youtube.com/watch?v=0cv8xxMxROM
Note the board around 1 minute 40 seconds. "High performance products likely to skip 20nm", which I would guess basically means AMD won't be releasing Excavator on 20nm
So what is the 14nm tapeout happening next quarter? This quarter is 20nm, surely must be graphics at TSMC. After that is 14nm, must be at GF. That only leaves Excavator and the small core APU's to choose between and so far AMD has never led a node with a small core and they've all been at TSMC.
Also, remember this? http://forums.anandtech.com/showpost.php?p=35220333&postcount=1
“We are typically at the leading edge across the technology nodes. We are fully top-top-bottom in 28nm now across all of our products, and we are transitioning to both 20nm and to FinFETs over the next couple of quarters in terms of designs. So we will continue to do that across our foundry partners. […] We will do 20nm first and then we will go to FinFETs,”
Don't get fooled into AMD's misleading marketing BS. Here is the quote from Lisa Su:
I don't see how AMD could be completing FinFET designs over the next quarter when Samsung and the rest of the FinFET players are just now talking about how their customers can now begin designing FinFET based parts. Expect AMD to launch FinFET chips in late 2015/early 2016.
20nm has to be TSMC graphics. She said the tapeouts were across their foundry partners, leaving the 14nm FinFET's at GF. The only question is what kind of chip is it, and the logic tells me it's Excavator.Lisa Su - Senior Vice President and General Manager, Global Business Units We will do 20 nanometer first, Hans and then we will go to FinFETs.
What you can't see is neither here nor there. Su flatly said (twice) that AMD was taping out 20nm this quarter and FinFET's next.
20nm has to be TSMC graphics. She said the tapeouts were across their foundry partners, leaving the 14nm FinFET's at GF. The only question is what kind of chip is it, and the logic tells me it's Excavator.
I don't think that Excavator would be at 28nm - most probably 20nm and 1H 2015 as release date.
IMHO Excavator at 28nm does not make sense, the 28nm node is quite old right now, it's going to be ridiculously outdated compared to Intel node by the time-frame that EX is supposed to be released. 28nm CPU in 2015? That does not sound remotely competitive, Intel will be at 14nm by then. 28nm design versus 14nm one will look silly. I'm not arguing that they won't release it on that node only that it won't make a competitive product but AMD has not been competitive with Intel for a long time(since late 2006) aside from some small cores products. Bobcat was way better then atom on netbooks.
That does not sound remotely competitive, Intel will be at 14nm by then. 28nm design versus 14nm one will look silly.
If AMD wants to keep up with Intels Core series, they'll have to increase performance. Frequency scaling seems to be dead, as Kaveris clocks are going down, not up. This leaves IPC increases, and this will require wider cores. And widening the core is a rather far-reaching change. You can't just slap 1-2 additional ALUs in there and be done; you need more scheduler dispatch ports, more scheduler entries, more register read ports, a more complex result forwarding network, a deeper reorder buffer, etc. And doing all this in one step is indeed rather complicated because it changes so many things in one step.Explain how EX is "drastically changed"? You do realize that Excavator is just a further evolution of Bulldozer, not some dramatically swapped-around uarch, right?
No, I don't have a source, because AMD is tight-lipped about the reasons for their inability to deliver their roadmap targets. However, I hope we can agree that AMD didn't delay and then downgrade performance targets of Kaveri voluntarily, but due to some unexpected technical problems. This pretty much leaves process/physical design problems, or problems related to architecture development.I'd also like a source for the claim of Kaveri being delayed due to process issues. Steamroller ver1, whatever it was, was canned at some point in 2012 and they decided to work on bdver3b, or Steamroller B.
Intel is immune to many of the problems that fabless companies have.Everyone is having increasing difficulties getting node shrinks to work - I see no reason why Intel is immune from this. It's just fundamentally a very, very difficult task. Look at how hard TSMC is finding the transition from 28nm to 20nm. Intel does have plenty of resources to throw at it, but they don't have some magical way of bending the laws of physics. The truth is that Moore's Law is running out. Intel might make it work, but I don't trust anyone's time tables or roadmaps on node shrinks any more. I'll believe the release date for 14nm (or any other new fab process from any company) when it happens.
>>AMD doesn't have Intel's R&D resources...the sooner that people rooting for AMD can reset expectations relative to this fact, the better.
I don't see what AMD's R&D has to do with it when they're totally reliant on GloFo and TSMC for their chips. EX APU @ 20nm in 1H 2015 is totally plausible, and I'm not sure why people are trying to pretend otherwise. The only thing that would prevent that would be GloFo yet again being incompetent.
BS. You do realize that as these nodes become more advanced, the R&D effort required to design an IC around them goes up, right?
Also consider that from an enthusiast perspective, there really hasn't been any substantial improvement since Sandy Bridge (32nm). Intel is prioritizing low power above almost everything else. But the real world still runs on workstations and servers, not tablets, and there are some high margins to be had there. If AMD can catch up to Sandy Bridge IPC with Excavator, as seems plausible, this would make them a strong contender in the high end again. They would have room to undercut Intel's Xeons on price while still making a healthy profit margin.