With the release of the new Ryzen processors, a new era begins for AMD. It brings the company back to competitiveness against behemoth Intel. Whether Ryzen is for you or not, this restored rivalry will benefit us all, be it through performance, features, or pricing. The first processors released are part of the Ryzen 7 Series, which are all eight-core models, featuring a new HyperThreading-like technology called SMT, which was long overdue in AMD's lineup. We also found huge improvements to single-thread performance, which was a major issue on the company's previous Bulldozer architecture. Another improvement worth mentioning is that power efficiency is incredible, beating Intel processors in many scenarios. Overall, this makes the Ryzen 7 1800X an excellent alternative to processors from the Intel HEDT lineup (eg: Core i7-6900K) - at much more attractive pricing.

The caveat: You must use software that makes heavy use of multiple cores, and frankly, I don't see that much software out there that really benefits from it significantly. Nearly all professional/workstation/productivity applications now have a GPU-accelerated work path, which will produce results much faster than any CPU you can throw at it while being more energy efficient at the same time. There are some notable exceptions here. For example, media encoding can be done on the GPU, but it is often performed using the GPU's own media-encoding hardware, which creates lower-quality videos than CPU-based encoders, like x264 or x265. This is where Ryzen will shine. If you look at the bulk of the other applications we tested, like Office, Photoshop, or Web Browsing, you will find that the Ryzen 7 1800X performs comparable to Intel processors that are one or two generations old. This is still a tremendously impressive feat considering how far back AMD was before Ryzen on applications that are mostly single or dual-threaded. But it's something you should be aware of. Ryzen will not magically make everything you do faster. On the other hand, these applications usually run very quickly on all CPUs on the market; when was the last time you've waited more than two seconds for an operation in Microsoft Office to complete?

We also tested a large selection of games, in both 1080p and 1440p, comparing it to the Intel Core i7-7700k, which is the typical gamers' choice and $150 cheaper than the Ryzen 7 1800X. Depending on the game, the results either nearly match the Intel CPU or are vastly slower, especially at 1080p. The only exception is Civilization VI where Ryzen's additional cores show a clear performance improvement. The most simple explanation why some games run at lower performance is because these games tend to be more CPU limited at the framerates tested. Every game has a certain CPU load it generates to calculate a single frame (almost independent of resolution), some games more, some less.
When paired with a fast graphics card, and running at the standard resolution of 1080p, the framerate will be high, which results in a high CPU load. At some point, the CPU will be running as fast as it can, which will cap the framerate. This is worsened by games that don't properly scale across multiple cores, or only run at up to four cores, for example. While this is of course the game developer's "fault," it is a reality of today's game market that isn't going to change soon.
Our second set of tests at 1440p shows performance losses getting smaller because the GPU can not drive such high framerates anymore; it is busy processing more pixels due to the higher resolution. This trend will continue at 4K resolution.

What does this mean for gaming on Ryzen? I would say that if you have a graphics card of up to $300 and game at 1080p, you should be fine because the GPU will be the performance-limiting factor. When you use faster cards, like the GTX 1070, GTX 1080, or GTX 1080 Ti at 1080p, though, Ryzen will result in lower game performance. One solution is to game at 1440p or higher to ensure the GPU becomes the bottleneck again, or to buy an Intel CPU.

What I'd also like to mention is how unfinished the whole motherboard ecosystem feels. AMD sent me a Gigabyte Aorus motherboard with Corsair memory, so I assumed they properly tested that combination for optimum user experience. Not really. Once you setup the system, your memory will run at 2133 MHz, which is extremely low and will severely restrict performance in both applications and games. You want to be running 2666 MHz at least. So, off I went into the BIOS, set 2666 MHz, but nothing happened. The damn motherboard BIOS just didn't apply the memory frequency. At this point, many novices would RMA the memory, motherboard, or CPU, or everything altogether, claiming "it doesn't work." The magic bullet (on my Gigabyte board at least) is that every single memory timing and memory voltage has to be configured to a manual value - not "auto" (this works fine on Intel of course, where you can leave most settings on auto or just select "XMP3000," and boom, you are ready to go). After this change, the Gigabyte Ryzen board would boot at 2666 MHz memory and run fine all day. We got 3000 MHz memory, though, so 2933 MHz was tried, which ended up being unstable no matter what I did. I ended up buying a bunch of memory kits with same-day delivery, and oh wonder, the newly bought Corsair 3000 MHz memory kit works fine (AMD sent me the exact same model, but apparently never tested its 2933 MHz stability). Several 3200 MHz memory kits that work fine on Intel at even higher clocks barely worked at 2666 MHz, and 2933 MHz remained a no-go. Once you've mastered the memory hurdles, you'll find various posts online by users, reviewers, and AMD themselves recommending you turn off HPET and SMT, use the Windows High Performance power profile and more tweaks. I'm not sure if this is a solid buying argument to professional workstation users who just want a system setup quickly and ready to go because every hour they spend tweaking costs them money.

Some people will now claim that it's not AMD's fault that the motherboards aren't ready. In my opinion it is. Why not give motherboard manufacturers all the hardware and support they need with plenty of time to spare?

Despite the motherboard issues, overclocking the CPU itself works well (functionally), also thanks to AMD's Ryzen Master utility, where you can make changes in real time and apply them without a reboot. On the 1800X model specifically, I'm not sure if overclocking is worth it. AMD's Boost and XFR will work in tandem to maximize CPU performance automagically. For example, in workloads with few threads, the CPU will boost up to 4.1 GHz without any manual setting changes, a frequency you'll probably not reach with manual overclocking unless you use watercooling or super-high voltage. Manual overclocking can still be beneficial for heavy workloads that use all cores, where XFR and Boost might not be active or running at lower frequencies.

For years, Intel has spoon-fed us incremental improvements to their architecture, especially on the power efficiency side, claiming that their processors are reaching the climax of what is possible for x86 power-wise. Then in comes AMD Ryzen with a much lower R&D budget and at least matches and sometimes beats Intel power-efficiency, often delivering better performance at the same time. Even with all its eight cores stressed with Prime95, the Ryzen 7 1800X draws just as much power as the quad-core i7-7700K.

If you compare Ryzen 7 1800X pricing to Intel's highest-end offerings, AMD has a clear winner on their hands once the platform's issues are ironed out. The 1800X comes at much better pricing with very similar performance. Looking at the Ryzen 7 lineup, though, I feel that most people will not need a 1800X and could probably go for the 1700 non-X, which is much more affordable with only little lower performance (we'll test this in a future review). If I were in the market for Ryzen personally, I'd wait till the platform is more mature and the major bugs are fixed.