They're still bound to the same power limit, and 12LP offers 10% more perf/watt. soooo....
They're still bound to the same power limit, and 12LP offers 10% more perf/watt. soooo....As much as I would love to see that, even AMD's own figures for the performance improvement aren't quite near that (a 10% improvement was mentioned by multiple, reliable sources IIRC). They could be rather conservative and probably are, but I'm expecting a 15% improvement in general, significantly better in some cases but worse in others. Still, I hope that your prediction about the 12LP process they will be using are correct.
Don't necessarily agree about the max turbo, but yes, AMD has every reason to sandbag this one. Unlike previously, they now have a product to Osborne, which they obviously don't want to do until as late as possible.AMD is giving away the information which does not reveal PR's competitive positioning imho. The base clocks will probably be up 10-12% . So a 2800x with 4 - 4.1Ghz is expected. The killer is going to be the turbo clocks. Do not be surprised to see max turbo of 4.7 4.8 Ghz. So AMD are not technically wrong with what they are saying. btw there is a concept called sandbagging which AMD has already done with Zen.
Don't necessarily agree about the max turbo, but yes, AMD has every reason to sandbag this one. Unlike previously, they now have a product to Osborne, which they obviously don't want to do until as late as possible.
They're still bound to the same power limit, and 12LP offers 10% more perf/watt. soooo....
AMD is giving away the information which does not reveal PR's competitive positioning imho. The base clocks will probably be up 10-12% . So a 2800x with 4 - 4.1Ghz is expected. The killer is going to be the turbo clocks. Do not be surprised to see max turbo of 4.7 4.8 Ghz. So AMD are not technically wrong with what they are saying. btw there is a concept called sandbagging which AMD has already done with Zen.
AMD is giving away the information which does not reveal PR's competitive positioning imho. The base clocks will probably be up 10-12% . So a 2800x with 4 - 4.1Ghz is expected. The killer is going to be the turbo clocks. Do not be surprised to see max turbo of 4.7 4.8 Ghz. So AMD are not technically wrong with what they are saying. btw there is a concept called sandbagging which AMD has already done with Zen.
For Zen 2, maybe. For Zen+ I would be shocked if turbo clocks are north of 4.5GHz.
We already have 4.2GHz in Threadrippers. Granted they're the top 5% of dies, but a whole year of silicon optimization and a new process won't go above what can already be done? I doubt it.I doubt any will even reach 4.5Ghz. The biggest improvement will likely be the lower clocked CPUs like the 1700. 4.2Ghz seems more likely.
I think people pay way to much attention to Single core turbo clocks and overclocking that takes the CPU past the TDP window AMD has set. The biggest thing we will see is a small uplift in base clocks, a decent all core turbo clock, and probably a disappointing SC turbo clock. All because AMD has power windows they want to maintain. That said I do think these will be much better overclockers.We already have 4.2GHz in Threadrippers. Granted they're the top 5% of dies, but a whole year of silicon optimization and a new process won't go above what can already be done? I doubt it.
We already have 4.2GHz in Threadrippers. Granted they're the top 5% of dies, but a whole year of silicon optimization and a new process won't go above what can already be done? I doubt it.
XFR still means that four cores out of 16 are doing 4.2GHz as long as you give it sufficient cooling. It would be hilariously pointless if Pinnacle Ridge couldn't do better than that.The boost clock is 4Ghz, XFR is 4.2Ghz. I just don't see a full 10% boost at a high frequency range. Base clocks should see around 10% though.
https://www.computerbase.de/2018-01/amd-ryzen-threadripper-2000/
I fully expect Pinnacle Ridge to blow past even 4.6 Ghz. AMD has given hints that GF 12nm is very close to competitor 14+ process at the CES Tech day. Its obvious the competitor is Intel 14+ process. KBL clocked 5 Ghz easily. I think AMD will get very close to KBL with around 4.8 Ghz for max clocks. In fact I think everybody is going to be surprised with max clocks of PR.
I doubt any will even reach 4.5Ghz. The biggest improvement will likely be the lower clocked CPUs like the 1700. 4.2Ghz seems more likely.
Fully agree. The 4.2ghz was pretty much a hardwall with Summit ridge due to the process. this new 12nm process is juts a tuned 14nm one and from GF. I don't expect much in terms of max clockspeed. The real benefit will be in higher all-core turbo. Still, as krumme said what is really needed is an improvement in the memory subsystem. That is were the main bottleneck is and more clock speed won't really help to up the mins in games in fact it will only make it more variable.
I did some testing on a downcored 1700 (3+3) to try and find where the memory was clocked on the Pinnacle Ridge test, by comparing bandwidth. I believe it was 2666.
After that I made some charts to compare the individual test results, and here they are:
Here is the Geekbench link to my result and the Pinnacle Ridge comparison: https://browser.geekbench.com/v4/cpu/compare/6690641?baseline=7255496
Taktraten für den Ryzen 7 2700X liegen bei 3,7 GHz in der Basis und bei 4,1 oder gar 4,2 GHz im Turbo.
so 4.5/4.6 OC is definitely not out of picture