TechyInAZ :
If this is true, this is going to be the biggest performance boost for games we've seen in a long time!
Considering that the performance of most of today's games tends to be much more limited by graphics hardware in most setups, and relatively few games make heavy use of much more than a handful of threads, I wouldn't expect these to increase current gaming performance all that much. They might help reduce performance dips in parts of games that are heavy on the CPU, but probably not really any more so than Intel's current i7s and unlocked i5s. These might potentially be faster, but probably not enough so to make the existing higher-end hardware seem obsolete. If they're bringing that kind of performance to lower price ranges though, and increasing core counts at the high end for other tasks, then that could be great. As far as noticeably boosting performance in today's games goes though, I would be more interested in what their upcoming 7nm graphics cards have to offer. Those will probably have a bigger impact on gaming performance within the price ranges most people spend for graphics hardware.
Aspiring techie :
10nm will come out at the end of the year, but 1st gen 10nm is likely to perform worse than 14nm. It will likely take one or maybe even two 10nm updates to bring it to par with curent 14nm+++. Even worse, the first architecture to come out on 10nm will be Cannon Lake, which will probably be only a little better than Skylake.
It's kind of difficult to say on this. Intel's 10nm has been delayed so long that by the time they launch they should have already been on the 2nd or 3rd-generation hardware. It's possible that part of the reason for the delays has been that 10nm wasn't performing as expected, and they didn't want to release something that performed worse than their existing hardware, so they may have been redesigning things to make the CPUs better for their debut. I would suspect that at the very least, Intel will make sure that their first 10nm desktop parts perform at least a little better than their most recent 14nm parts. I'm pretty sure they'll have a high-end chip that performs at least a bit faster than their 9900K. It's difficult for one to say how much better though, without having inside details about the current state of their 10nm production.
tim.hotze :
I want to believe that this is true, but I'm skeptical: I don't see how you keep 16 cores fed with 2 channels of DDR4 memory, even if its overclocked (and DDR5 remains an "on-paper" product, and will likely start out in severs), and since you need dedicated pins for memory access (and they're sticking with AM4).
That puts the entire leak into question, IMO, though it might make sense if they released Ryzen 9 as a niche product (somewhat akin to Intel's 8086k or something).
Running 16 cores on 2 channels of memory probably isn't much different than running 32 cores on 4 channels of memory, as their Threadripper 9900WX currently does on the TR4 platform. That processor does run into performance issues at certain tasks though, but is still a viable product for many heavily-threaded tasks. I would only be a bit skeptical due to the fact that it encroaches on the TR4 platform, but since Intel is now matching AMD's core counts on their consumer platform (albeit at a huge price premium), I could see AMD pushing AM4 to 12 or 16 cores just to have something available at the high-end that is notably more capable at heavily-threaded tasks than what Intel is offering.
kidfusion3000 :
I vote this is bunk. I'm an AMD fan, always have been, which is why I think we should be expecting less dramatic increases on these CPUs. AMD made huge waves in the last 2 years but it doesn't mean they can again increase core counts, and up to 16 cores on the AM4 platform? even with a 135w tdp, I'm just not sure that's possible with a concurrent increase of 1 ghz, that's a 20% increase in frequency compared to the last gen and 2x the core count, with a 12% increase in IPC per previous reports, this lineup just isn't plausible.
While I don't know how accurate these rumors might be (at least the prices for a given core count seem a bit questionable) 16 cores on AM4 doesn't seem impossible. The upcoming 7/10nm node should bring big efficiency gains over the existing 12/14/16nm node, and their current 8 core Ryzen 2700 can already boost to 4.1 GHz on a 65 watt TDP. Keep in mind that the 4.7 and 5.1 GHz boost clocks listed in that graph would not be for heavily-threaded tasks, so at stock you would only see clocks that high with more lightly-threaded workloads. With all 16 cores and 32 threads active, you would likely be looking at clocks closer to the base-clocks of those chips.
The Threadripper 2950X already does 16 cores with 32 threads and a 4.4 GHz boost on a 180 watt TDP. A 135 watt TDP is just 25% lower than that, and I could see the significantly increased efficiency of 7nm leaving enough headroom for them to also boost clocks. The rumored 125 watt 3800X only has about 11% higher base clocks and 7% higher boost clocks than that processor with about a 30% lower TDP, making it sound rather feasible. The 3850X, which is rumored to be coming later in the year, pushes those limits more, but they would undoubtedly be using the best binned chips to manage something like that, so it might not be too far-fetched either. In any case, we might know in another week how much truth these rumors have to them, if these chips actually launch at CES.