I'm still waiting for a more diverse set of synthetic and real-world benchmarks. It'll be interesting to see how IPC performance holds up with Threadripper, however I think the most interesting debate will be whether the 1920x or lowest end Epyc CPU are a better buy.
Unfortunately, even as an enthusiast $799 is more than I'm willing to spend on a CPU. I'm also still hard pressed to build a Ryzen 1700 System since I can purchase an i7 7700 from MicroCenter for about $10 less than the Ryzen part (and have equal or better general performance with notable better IPC).
The i7-7700k will give you a slightly better performance in games, but mostly only in games. In Multi-threaded workloads - and isn't a cpu in this league mostly relevant for those? - the Ryzen 7 is a bit faster than the i7-7700, and very close to the 7700K[0]. This is before overclocking, which one would definitely do with the smaller R7.
Also, the IPC of the i7 is not that much better, as evident by the good gaming performance of the Ryzen line. The 7700K however with its higher clock arrives at a higher single threaded performance, making it reach higher FPS.
You can have the same or better performance when running a single app, for sure.
Lots of people who've moved to Ryzen saying how it's changed the way they user their computers, not having to worry about multiple CPU hungry apps running at the same time.
Cinebench scores are fantastic for speculating on how these CPUs will work in the context of 3D rendering/simulations. That personally has me very excited.
Agreed - 3062 for the TR-1950X and 2431 for the TR-1920X are colossal results. 2167 for the i9-7900X is still huge!
At first I thought that (given it's AMD running the test) they'd hamstring the Intel CPU with an under powered cooler, but it looks like they gave it a big Corsair H100i 240mm water cooler. If that's not enough for a 140W CPU, I don't know what is - it was probably on turbo the entire time. They gave their own CPUs an "EK Prototype SP3" cooler, though - is this it? https://www.ekwb.com/shop/ek-kit-s360
Regardless, the message seems to be that if you want top performance, get a big cooler.
The "EK Prototype SP3" cooler was probably a prototype of a water block designed for the X399 platform. The chip is massive, so the existing solutions probably don't fit it.
My only qualm is that purchasing an i7 7700 system in comparison to a Ryzen 1700 system is almost the same price (since I'm buying my CPU from MircoCenter).
The use of my desktop is maybe 5% gaming, the rest is software dev and browsing, isn't there still a reason to prefer an IPC advantage for compilers and other system tasks?
I mentioned this, because I don't think a 12 core CPU would really give me that much more productivity or value gain than an i7 7700 or Ryzen 1700.
I only really need a single linux compatible nVidia GPU, enough cores and ram to comfortably handle 20 browser tabs, an IDE and a relatively hefty docker dev flow.
Unfortunately you don't see a whole lot of benchmarks of dev stuff on CPU review sites. Phoronix at least does Linux compile time. Not sure how representative that is of your "relatively hefty docker dev flow" but it shows the 1700 as 10% faster than the 7700K, which itself has a 17% higher base clock and 7% higher boost clock than the 7700.
Higher single thread performance is higher single thread performance, no matter if achieved via higher clock, doing more per clock ("IPC"), or both.
If you run one non-parallel task, the 7700K will be faster. If you run many non-parallel tasks in… parallel (make -j16), the 1700 will be seriously faster.
Unfortunately, even as an enthusiast $799 is more than I'm willing to spend on a CPU. I'm also still hard pressed to build a Ryzen 1700 System since I can purchase an i7 7700 from MicroCenter for about $10 less than the Ryzen part (and have equal or better general performance with notable better IPC).