Ryzen won't be competitive for gaming?! Like I have said countless times, 5Ghz on a 7700K is NOT a guarantee, I've seen the owners club on OC,net. Eurogamer's review there's was not stable at 5Ghz etc etc.
Eurogamer's tests always seem to show lower clockspeeds than are typical for whatever chip they're testing.
I know some people who are still in the mindset of anything more than 60C being "too hot" for the chips, and so they will scale back their overclocks - despite modern Intel CPUs being fine up to 100C and throttling long before any damage can occur.
Even if that's not the case, and they've just been unlucky with every CPU they've had, their review has a 4.8GHz overclock scoring 207 points - which is still 35% faster than the 1700X score.
Also, you're comparing the fastest and 5Ghz overclocked 4-core CPU to an 8-core at stock frequencies, of course it will be faster in cases where frequency is important. By your methods the $1500 i7 6950X comes out poorly too. Having said that, like you said, once you overclock the Ryzen's as well, the gap will narrow to something unnoticeable, which was my point, for the vast majority of gamers.
Well we were talking about gaming performance, where >4 cores barely matters for most games today.
In the majority of games, a fast 4c/8t CPU will perform best.
Most people wouldn't be buying a 6950X for gaming either, as it's generally a poor choice and will perform worse than a 7700K in the majority of games.
That's why I'm surprised to see people freaking out over the multi-threaded scores.
We knew that Ryzen would do well there, since it's an 8c/16t CPU with good IPC.
Something like the $8898 Xeon E7-8894V4 would also be a terrible choice for gaming, despite it being ridiculously fast in certain types of workload.
You don't see people freaking out when a $350 7700K outperforms a $9000 Xeon in gaming - it's expected.
I reckon the R7's will overclock to 4.5Ghz at least, possibly up to 4.8Ghz, which would be beyond Intel's HEDT chips.
That seems
very optimistic - but I hope that is the case.
If you want the best bang for buck Ryzen chip just for gaming then you're probably best going for the 1400X 4c8t model.
The highest boost figure of all the (cheaper) models (3.9Ghz) suggests it might be the best overclocker as well.
It depends on whether there's a significant difference in clockspeed, but I'm thinking that the 6c/12t CPUs are going to be the best value for money - especially if your focus is gaming.
While most games today don't need more than four cores, some newer games do benefit from it.
I'm not expecting there to be a significant difference in clockspeed between a 4c/8t Ryzen CPU and a 6c/12t CPU as long as you have decent cooling.
If you plan on keeping the CPU for more than a year or so, those two extra cores is likely to make a big difference in how long the CPU will still be competitive.
With Intel, there are significant differences between the HEDT chips (≥6 cores) and the mainstream chips (≤4 cores).
To go from a quad-core CPU to a hex-core CPU means moving to a completely different architecture and sacrificing both IPC and clockspeed.
With Ryzen, they just add another CCX unit to the chip. There are no significant differences between them.
The 6c/12t Ryzen chips are also going to have the most cache available per-core, which might even have them outperform the 8-core chips running at the same clockspeed in some tests.
That's why I'm also interested to see what Intel's response to Ryzen is going to be.
Coffee Lake is said to have mainstream 6-core CPUs, which means two more cores without the sacrifices that you have to make to move to the HEDT platform.
But that's potentially a year away, and not relevant to someone wanting to buy a CPU for gaming today.
I certainly need to upgrade my 2500K now, and not wait another 12 months. The question is whether it ends up being a 6c/12t Ryzen, 8c/16t Ryzen, or a 7700K.
Who says it was running at that frequency during the test?
That CPU-Z screencap doesn't mean much like that
When the boost clock for that CPU is 3.8GHz, running at 3.9GHz suggests that XFR is enabled.
Since XFR dynamically overclocks the CPU based on the workload and the cooling solution attached, it means that we can't draw any real conclusions from these results.
While that Cinebench test shows the CPU running at 3.5GHz, if XFR is enabled then it may have boosted much higher than that during the single-threaded test.
If we're being optimistic, 154cb is a good score for 3.5GHz and hopefully the chip will reach speeds of 4.5GHz or higher. That would be very competitive with what Intel is offering.
But since XFR seems to be enabled, maybe the chip was boosting to 3.9GHz or higher during that test, which would mean that 154cb is not that impressive. My old 2500K at 4.5GHz scores 150 in that test.
On this '7700K is the best gaming CPU Ryzen won't be able to compete with' trope that has quickly formed, let's examine that a little shall we. Because it is a claim that makes no sense for the vast majority of gamers or the 'average gamer'. And this is an important point so I've bolded it so I don't get a load of replies where I'm shown benches with Titan X's which nobody really owns, let's face it.
Anandtech doesn't understand how to benchmark games. Their tests are meaningless.
Look to sites like DigitalFoundry, GameGPU, and Techspot for proper CPU performance testing in games.
The argument that the "average gamer" is always GPU-bottlenecked and that CPU performance doesn't matter doesn't make a whole lot of sense.
A CPU performance test should always be done with the fastest GPUs you have available, and at a low resolution like 720p specifically to try and eliminate the GPU as a factor.
You might not have a Titan X, but that level of performance will be available for less than $1200 very soon with the release of the 1080 Ti (maybe $800), and likely around $400 later this year/early next, with the 1170.
Most people tend to hold onto CPUs far longer than GPUs, so that's why you need to eliminate the GPU as a factor in these tests.
Not only that, but you don't
have to be GPU-bottlenecked just because you're running something less than a Titan X.
If you choose appropriate settings for your GPU instead of always setting everything to Ultra, then you shouldn't be GPU-bottlenecked at all the majority of the time.
The pertinent question is really what frame rate a player is targeting. Anyone with a 60Hz monitor is very unlikely to see any difference between a 7700K and a significantly less powerful processor
That really isn't true any more.
Even targeting 60 FPS requires a very fast CPU in most new games. Practically everything I bought in 2016 was bottlenecked by my CPU.
And every time a game comes out which has high CPU requirements, you now have people complaining about how it's "unoptimized" or doesn't run well despite their GTX 1080 - paired with an older, slower CPU.
Part of the difference might be explained by different RAM bandwith, though. The Witcher 3 has shown that it scales pretty damn good with faster RAM.
Quite a few games do now - which is why I'm somewhat concerned about the highest-end ASUS board only listing 3200MHz DDR4, and boards from other manufacturers being ≤3000MHz.
Z270 boards support 4266MHz RAM now.
