Good, nice to see Intel still working on optimizing the 65 nm parts now that they also have 45 nm working, although I don't know if they are producing very much on it just yet. This new stepping might very well make the QX6800/Xeon X5365 produce 130 W of heat rather than 150 watts or so, which would allow it to work with most motherboards and coolers. I believe that Intel put a maximum energy delivery and TDP figure of 130-135 W on their latest motherboard spec (for the Pentium D 800s) so 150 W might cause some problems.
As of late I have been wondering why is seems the new batches of conroes seem to be poor overclockers. I know there are several different factors that can affect how well a CPU can overclock, if the problem lies within the weeks they were produced that may mean the silicon was not as pure as previous high clocking conroes... But if its the stepping then that would mean something within the process is now different.
Considering I have seen people on onverclocking forums/sites claiming their new e6420s/e6600s are struggling to hit 3.2 ghz in some cases, yet seem to run cooler for the most part, MY THEORY is the oxidation layer between the gate electrode and transitor channel has been made "thicker". I say this because reduced temps and power consumption correlate with a reduction in gate leakage.
Though this is purely conjecture, my understanding within the field of CPU technology is relatively limited. Clarifaction or further speculation would be weclome!
Anyone have any insight to share. I'm looking forward to see one of these getting tested and hopefully pushed to it's limit.
Also to answer your question, if my theory is correct, these CPUs will be poor overclockers due to a decreased capacitance.
It's hard to believe that thew Q6600 -> QX6700 has in increase of 25 watts yet the QX6700 -> QX6800 has an increase of 0 watts yet the clock speeds have the same jump. One would expect *more* than a 25-watt jump since the higher you go, the greater odds of a Vcore increase needed to reach the clock speed, and this increases head quadratically rather than just linearly with clock speed.
I guess it could be true through very picky binning, but yields will absolutely stink on that. Basically it's a pair of 65-watt X6800s versus the normal 75-watt ones put together on one die.
All internal steppings that are used to test out bug fixes, speed path issues and or power reduction.
This happens all of the time from Intel. Most of the time the production core that we see has no relationship to what the internal stepping is. This is Intel's way of improving the core for release to the paying public.
Okay. I am more used to seeing AMD's stepping methodology, where the stepping letter is incremented on each revision of the parent arch. The numerical portion of the stepping would have been "internal", so instead of the C, D, E, and F steppings disappearing, it would have been B4, B5, B6, and B7 steppings missing and then B8 being sold.
Oh, there were probably several of the C1, D1, D2 and so on steppings. This is how Intel compares with AMD's APM technology.
Remember that each mask layer now costs around 3 to 5 Million to make. A Intel Core processor uses up to 8 metal layers and I don't know how many just for the silicon logic. So if they can't reuse the earlier masks for the later revisions then Intel is sinking a lot of money just in Masks alone for all of those missing revisions.
Usually a number revision only changes a metal layer while the Alpha part is for logic revisions at least this is what I have gathered.
JumpinJack might have some more insight on this than me.
When a new chip is made it has the stepping and revision.
For the new Penryn processors from Intel they have a A revsion and 0 stepping.
Conroe launched on a B2 revision and stepping. At a guess there were 4 to 5 steppings and revisions prior to the launch of the Conroe at B2. The Kentsfield took one additional stepping to B3.
The new revisions and Steppings for Conroe that are supposed to have the 1333 FSB are supposed to launch on the G revision and I think the 2 stepping.
All this means is that Intel or AMD has improved some part of the processor or process to make the processor. This usually for fixing silicon bugs, improving binning/yields which include improving overal processor speeds, operating temperatures and such.