Yes and no. I wouldn't really be surprised if the die shrink has done nearly nothing for AMD's IPC: cache size could have made a difference and I wouldn't be surprised if AMD has done some minor things like adjusting the number of entries in TLBs. So, unless AMD has been hiding something, I'm expecting the main significance of these changes to be an (eventual) improvement in clock speed and an improvement in AMD's costings.
Now, given that this is a new process, I would expect it to take AMD some time to get yields up at the higher clock speeds, but that there will be an eventual improvement in the availability of higher clock speed parts and maybe some decrease in prices on the middle and lower speed parts.
OTOH, do I believe that this memo is genuine? Not really. Nothing in it is different from what an outside oberver would conclude from knowing the background. It doesn't add any new insight.
is pretty illiterate, and you would expect that an educated person wouldn't make that kind of mistake (but you'd often be wrong). It also seems that the author doesn't accept AMD's recent marketing thrust; given that this kind of thing can be career death, it is at least a slightly surprising thing to go into print with.
the overclocking potential of the CPU’s is once again dwarfed by the overclocking potential of the Q6600
Again a surprising statement; not a surprising "fact", but surprising to look at in that way. Given that one way of looking at overclocking potential, is as the amount of margin that shipped parts have inherent in them. From that point of view, and given the context, it is massively unsurprising that AMD is having to ship parts with less unutilised margin than Intel. In fact, assuming that AMD is being succesful at this (ie, not getting excessive rates of returns), you could say that this is a sign of success; in spite of having a less good hand to play, AMD is succeeding at playing it as well as could be expected (not that this is what overclockers want to hear).
the memo kept mentioning how the CPU’s will get a performance boost when used in conjunction with the new sockets
OK, some good news, at last, I suppose...But oddly, percentages aren't mentioned. I would have expected anyone making this kind of case to be saying "and, of course, we have to remember that the existing clock speed parts overtake (eg) Q6600 when the (eg) 10% boost from the new platform is available" (except that I expect the speed boost from the platform to be lower than that)
which lead me to believe that all these internal tests were done using already existing sockets, which means that even AMD does not have the new AM3 or G34 socket ready
Or that the initial CPUs are packaged in the existing packaging. Given that its a bit of a pain to get a new part into a new package if you don't already have the target package in production on another producty, this would be unsurprising. But it could indicate that AMD have had to devote all resources to getting Deneb out in time to rain a little on Intel's parade and haven't yet been able to put enough R & D resource onto the new platform, which be a little worrying, but hardly surprising.
Facebook
Twitter
Instagram