What Makes It Different? How About Turbo CORE?
Architecturally, Phenom II X6 is a fairly unremarkable evolution from its quad-core predecessor. Total L1 and L2 cache increases corresponding to the two additional cores (3MB total L2 now), but shared L3 is expected to persist at 6MB. AMD is planning Black Edition models of the X6, and will step out of the gate with greater than 3 GHz default clock rates (3.2 GHz is what we're expecting based on early e-tailer indications). Overclocking headroom on the six-core part is expected to be “in line with what enthusiasts are already seeing from the X4,” which is probably better news for the liquid nitrogen crowd than the folks who overclock on air.
Undoubtedly the highest-profile change being made to Phenom II X6 is the addition of Turbo CORE technology. The feature sounds a lot like Intel’s Turbo Boost technology, and in many ways, it is. Based on CPU utilization, Turbo CORE is able to spin down three cores and spin up three cores by as much as 500 MHz when you’re running workloads that would otherwise leave a trio of cores idle.
The processor doesn’t shut down the three unused cores, though; it throttles them down to 800 MHz and drops their voltage, in turn increasing voltage to the hardware being asked to run faster. One side of the equation—dropping voltage and frequency of unused cores—is what Cool’n’Quiet has been doing for years. The other side is a lot like Cool’n’Quiet in reverse. I’d dub it Hot’n’Loud, but because the other half of the CPU spins down at the same time, AMD keeps the processor within its TDP, so you’re not noticing a difference, and it neither increases thermal output nor the cooling needed to keep up.
In comparison, the Power Control Unit responsible for managing Intel’s Turbo Boost technology samples temperature, current, power consumption, and operating system P-states. Using this information, it’s able to almost completely shut down unused cores and accelerate others. AMD’s solution is less elegant, but still interesting in practice, using the target TDP to set universal conditions. When load is at a certain level, dropping the voltage/frequency of three cores and increasing the voltage/frequency of three cores keeps us under maximum TDP. Now, Turbo CORE is dynamic in that the three “boosted” cores will jump around, depending on where the operating system schedules threads.
As far as we’ve been able to determine, Turbo CORE isn’t as granular as Turbo Boost. It operates based on operating system P-states (performance-states), so when three or more cores are at low utilization levels and the active cores are in P0 (for more on P-states, check this out), the CPU capitalizes on TDP budget to increase performance by a flat 400 or 500 MHz. Now, this might mean that a Phenom II X6 will consume more power at idle and over the course of time than an X4 with its cores throttled down to 800 MHz. This remains to be seen in our benchmarks. You can be sure that efficiency is one of the criteria we’ll be eyeing.
With that said, AMD claims that its goals with Phenom II X6 were faster performance in highly threaded apps and more speed in single-threaded titles. I’d expect Turbo CORE to help AMD in both regards.
One last little bit. AMD won't say whether it will or will not launch a quad-core processor based on its Thuban design with Turbo CORE functionality, but we do hear that this is in the works.