AnandTech | AMD Announces Carrizo and Carrizo-L, Next Gen APUs for H1 2015
While not a surprise that Carrizo is coming, it's nonetheless good to hear that it is indeed on the way. What with the slow/non-existant launch of several Kaveri mobile processors, I think it's very appropriate to forget about Kaveri at this point, and start focusing on Carrizo instead!
For me, the most disappointing news is Carrizo will still be 28 nm, and the voltage adaptive operation is perhaps the most interesting, as if it does cut 10-20% of power use, that could go a long way towards making up for the same process and improving AMD's position power-wise. Of course, it will all be for naught, in terms of getting my money, if there aren't at least a few high-end models launched this time around, so I'm hoping this time the OEMs deliver on that.
(By the way, I propose that the current "Next-gen AMD CPU's" thread be focused on Zen, since that's been its current focus, and this be focused on Carrizo. That way we can more or less keep the discussion focused on one generation in each thread)
-
-
The voltage adaptation does sound quite interesting. I wonder how that works without impacting performance. On the other hand... still 28nm? Seriously? By H1 2015, the 22nm process will be 3 years old (first Ivies came out in April 2012). Even given the fact that Intel has gotten lazy, at the rate this is going, AMD will be behind by two nodes rather than one.
-
It is coming fast after Kaveri, not even laptops available yet with Kaveri (at least with the fastest one). Sad still no 45W version, which would reduce significantly the throttling issues under gaming. Hopefully this the last APU based on Buldozer cores :hi2:.!
-
-
Seems logical to me, though, considering my laptop. Its Core 2 Duo runs at 1.225V or so by default, IIRC. But I can safely undervolt it to 1.025V. I used to be able to get 1.0125, but it seems to be a little less tolerant as it gets older (maybe it just needs repasted?). At any rate, when it was in its prime, 1.0125V was safe, and 1.000V often was, but every so often it would crash - often enough that I wouldn't run it at 1.000. If it had had something like this, it could have downclocked when the 1.000V actually wound up being more like 0.994 or whatever was too low and made it reset, and it could've saved that extra 0.0125V, saving another 2% or so power. And that's 2% after my optimizations - it would've been a lot more significant relative to the factory voltage. So the rationale makes sense to me.
Forget AMD Kaveri, Carrizo On The Way
Discussion in 'Hardware Components and Aftermarket Upgrades' started by Apollo13, Nov 20, 2014.