09-06-2018, 09:50 PM
https://www.extremetech.com/computing/27...ghz-on-air
Quote:Hitting 5.3GHz on air is a real achievement, but WCCFTech notes that other leaked chips have been hitting 5.5GHz with water cooling and a CPU voltage of 1.536 volts. I mention this not because 5.5GHz on water is a bad achievement — it absolutely isn’t — but 1.536v is an insane amount of voltage to pump through a CPU. By contrast, 5.3GHz on 1.25v with an entry-level cooler is a very good result. What this means, in aggregate, is that we don’t actually know much about how much headroom these cores will practically have. The 5.5GHz chip could be a bad overclocker, while the 1.25v CPU @ 5.3GHz might represent a semi-mythical “golden sample.” Said samples aren’t actually mythical but your chances of acquiring one are as bad as you think they are. The actual amount of improvement compared to a standard CPU or GPU always varies depending on the particulars of the manufacturing process and the characteristics of the product in question.
But the implication of these results, at least, is that Intel’s decision to move to solder could pay some modest dividends for enthusiasts and overclockers when these cores launch later this year. Overall performance on the 9700K is also said to be better than the 8700K, even at stock, which should help make them an overall upgrade — though the 8700K is an excellent chip in its own right and unlikely to drive much in the way of an upgrade cycle. We’d expect most buyers to come from customers using Skylake or older CPU cores.
Don’t expect solder to fundamentally change the physics of CPU scaling, however. Even if Intel’s manufacturing tweaks move the bar outwards by a few hundred MHz, just using solder between an IHS and die won’t magically make silicon scale more effectively above 5GHz. Enthusiasts may be able to scale their CPUs higher by 5-8 percent, but Intel won’t be riding this change to another GHz of clock — or even another 500MHz.