Tony, Re: "Check the Vcc requirements to get an 1800+ up to 1.9 and 2.0 GHz: 1.95 and 2.15 volts respectively. Shouldn't the suckers punch through or something like that with these voltages on a 0.13 device?"
Well, according to the article, only 1.9GHz was able to run all the benchmarks reliably, but I doubt you'll see AMD punch up the voltage that high in the future. And that's the thing with overclocking - increasing the voltage gives a little more headroom, but unless Intel or AMD is able to make all chips that way, there is no correlation between the overclocking results, and actual headroom that can be achieved.
One thing that bothers me, though, about the voltage levels is leakage, which on .13u devices gets much larger. I imagine those overclocked chips dissipate quite a lot of watts. I am frankly surprised that the chip can actually handle those voltages at all - it does speak favorably of AMD's robustness in their chip.
But in terms of frequency, though, I think that AMD has run out of gas. The article confirmed that an 1800+ runs at 1.53GHz, just like the Palomino part. And so far, only the 1800+ part has undergone any tests. What about the 2000+ or 2200+ Thoroughbreds, which are supposed to launch at the same time? Those should at least have been available at the same time, but apparently even those are MIA to the overclocking community.
Granted, newer steppings are sure to allow AMD some faster speeds, once they optimize speed paths and fine tune their process. But, how much work will have to go into it, just to be competitive with Intel? With all of AMD's resources being devoted to Hammer, will they be able to allocate enough budget to continue optimizing the TBred core? They have 2600+ (2.06GHz) set for the end of the year, but if all they can manage now is 1.53GHz, it's doubtful that they can wring another 35% frequency in just 9 months with limited budget. Looks like Hammer is going to be all or nothing for them.
After thought: another thought I had was in mobile. If AMD's desktop chips are at 1.65V already, and they are barely able to reach the speeds of Palomino, how far will AMD be able to get with mobile chips at 1.4V and below? I wonder if those roadmaps will have to be adjusted, too. With Banias coming available at the beginning of next year, perhaps Intel will be able to hold their mobile market pretty steady.
wbmw |
