The AMD Ryzen 7 1800X Performance Review

[SKYMTL]

Productivity Benchmarks: GIMP / Handbrake

GIMP

While it may be open source, GIMP is actually one of the most popular free photo editors available right now. It uses both CPU and GPU acceleration for certain tasks. In this test we use an 8K image and use a script to run eight different filters in succession. This is considered a lightly threaded workload since the memory, CPU and storage drive can all play a role in performance.

Handbrake

Video conversion from one format to another is a stressful task for any processor and speed is paramount. Handbrake is one of the more popular transcoders on the market since it is free, has a long feature list, supports GPU acceleration and has an easy-to-understand interface. In this test we take a 6GB 4K MP4 and convert it to a 1080P MKV file with a H.264 container format. GPU acceleration has been disabled. The results posted indicate how long it took for the conversion to complete.

Once again we are seeing Ryzen struggle in a lightly threaded photo manipulation task but it doesn’t have exclusivity on those performance challenges. The Broadwell-E chips which don’t pack in the very specific IPC increases from Intel’s Skylake and Kaby Lake architectures also post lower than expected results. Honestly, if you are pushing filter-heavy photography workloads in GIMP, Photoshop or another similar program, the Kaby Lake chips represent money well spent.

Handbrake on the other hand plays to Ryzen’s strengths in a big way and, much like Adobe Premier Pro, the 1800X posts some extremely respectable conversion times.

 

[SKYMTL]

Productivity Benchmarks: POV Ray / WinRAR

POV Ray 3.7

POV Ray is a complex yet simple to use freeware ray tracing program which has the ability to efficiently use multiple CPU cores in order to speed up rendering output. For this test, we use its built-in benchmark feature which renders a high definition scene. The rendering time to completion is logged and then listed below.

WinRAR

WinRAR is one of those free tools that everyone seems to use. Its compression and decompression algorithms are fully multi-core aware which allows for a significant speedup when processing files. In this test we compress a 3GB folder of various files and add a 256-bit encryption key. Once again the number listed is the time to completion.

Rounding out our real world benchmarks are two very different programs. On one hand POV Ray represents a very typical rendering environment whereas WinRAR utilizes low level algorithms to compress multiple files. As we’ve seen over the last few pages, the Ryzen 7 1800X provides an awesome platform upon which you can build a rendering station. However, in a program like WinRAR that fluctuates between light CPU workloads and heavily threaded situations, performance relative to Intel competitors slips a bit. Given the fact this is a $500 processor, there’s absolutely nothing to be embarrassed about here.

 

[SKYMTL]

Single Thread Performance / Memory Bandwidth

Single Thread Performance

Even though most modern applications have the capability to utilize more than one CPU thread, single threaded performance is still a cornerstone of modern CPU IPC improvements. In this section, we take a number of synthetic applications and run them in single thread mode.

One area in which the Zen architecture has made significant strides over previous AMD CPU generations is within single threaded workloads. To be honest, Bulldozer and its derivatives featured laughable performance metrics when used outside multi-threaded environments unless they were pushed to insane clock speeds (like the FX-9590’s 5GHz).

The differences are on full display in the charts above. It’s like night and day. While Ryzen has a tough time competing against Kaby Lake in these situations (which is a concern if Intel gets around to releasing a Kaby Lake DT offshoot), it does overcome both Broadwell-E chips.

Now some of you may find these results at odds with the observations I’ve been making about lightly threaded workloads but let’s be clear: lightly threaded doesn’t necessarily refer to purely single thread workloads like those above. While AMD’s XFR and Precision Boost technologies allow Ryzen to have a frequency advantage in single threaded situations, Intel’s chip design tends to flex its muscle in the void between one and sixteen threads. Platform maturity certainly factors into this equation as well and I’m sure as AMD rolls out optimizations for Precision Boost through BIOS updates, we will see this gap narrow.

Memory Bandwidth

If you actually managed to read page 5 of this review, my inclusion of this section should come as no surprise. If you haven’t read it, do so now. Simply put, the Zen architecture has some serious memory limitations that will certainly affect the way people buy their modules and adds confusion to an otherwise straightforward launch. With those specific characteristics in mind, I wanted to see how Ryzen handled memory-specific workloads and the fact I’m using identical DDR4 memory on every platform in this review (minus 990FX of course) allowed me to test in a controlled environment.

The results are far from the “gotcha” moment I was somewhat expecting given the hiccups I encountered with compatibility. Instead, when operating at a relatively sedate speed of 2666MHz with 14-14-14-32 timings, Ryzen performed well within the norms of a modern dual channel platform. It was never expected to compete against the quad channel Broadwell-E platform in terms of raw bandwidth. Whether or not these results hold true as we push beyond 2666MHz remains to be seen and is outside the scope of this review anyways.

 

[SKYMTL]

Gaming Performance (Synthetic)

3DMark Fire Strike (DX11)

I’m sure that most of you skipped to this section, so let’s start it out with some light synthetic tests which should set the stage for the next few pages of real-world testing. When using DX11 Ryzen suffers which is likely due to the benchmark’s focus on dual to quad thread workloads; an area where AMD’s new architecture has trouble keeping up with Intel.

3DMark Time Spy (DX12)

Switching things to DX12 and its multi-threaded environment sees the 1800X climb back into contention but oddly enough, the CPU-focused side of the Time Spy benchmark shows no love for AMD. Its result is within 10% of Intel’s i7-6900K but that’s a fair bit off from the 5-15% wins we were seeing in some previous tests. Will this hold out into in-game testing?

 

[SKYMTL]

Gaming Performance (Battlefield 1 / COD: IW)

Battlefield 1

Battlefield 1 will likely become known as one of the most popular multiplayer games around but it also happens to be one of the best looking titles around. It also happens to be extremely well optimized with even the lowest end cards having the ability to run at high detail levels.

In this benchmark we use a runthough of The Runner level after the dreadnought barrage is complete and you need to storm the beach. This area includes all of the game’s hallmarks in one condensed area with fire, explosions, debris and numerous other elements layered over one another for some spectacular visual effects.

Call of Duty: Infinite Warfare

The latest iteration in the COD series may not drag out niceties like DX12 or particularly unique playing styles but it nonetheless is a great looking game that is quite popular.

This benchmark takes place during the campaign’s Operation Port Armor wherein we run through a sequence combining various indoor and outdoor elements along with some combat.

With a mighty Titan X beating at the heart of this test system, I was hoping to eliminate any GPU bottlenecks but obviously that didn’t happen in some cases for various reasons. When there wasn’t any GPU bottlenecks, Ryzen did fall behind its immediate competition though. With its latest patch, Battlefield 1’s DX12 API path seems to be perfectly tailored for high frequency quad core architectures like Kaby Lake and as a result Ryzen ends up midpack, trailing every one of the Intel processors. Infinite Warfare on the other hand has a framerate cap at 125FPS and none of the newer processors had much of a problem reaching that level.

 

[SKYMTL]

Gaming Performance (Deus Ex / DOOM)

Deus Ex – Mankind Divided

Deus Ex titles have historically combined excellent storytelling elements with action-forward gameplay and Mankind Divided is no difference. This run-through uses the streets and a few sewers of the main hub city Prague along with a short action sequence involving gunplay and grenades.

Doom

Not many people saw a new Doom as a possible Game of the Year contender but that’s exactly what it has become. Not only is it one of the most intense games currently around but it looks great and is highly optimized. In this run-through we use Mission 6: Into the Fire since it features relatively predictable enemy spawn points and a combination of open air and interior gameplay.

The Deus Ex result came as a bit of a shock to me since I expected AMD to have in-place optimizations for their Ryzen architecture; they were very close development partners with Square Enix on this title. Once again however, Ryzen fell behind even the 7600K. As for Doom’s Vulkan implementation….well, what’s there to say? 200FPS is more than enough and we’re smashing right into the game engine’s framerate cap. Obviously more testing will need to take place at higher resolutions but in those situations the GPU rather than the CPU will become a bottleneck.

 

[SKYMTL]

Gaming Performance (GTA V / Overwatch)

Grand Theft Auto V

In GTA V we take a simple approach to benchmarking: the in-game benchmark tool is used. However, due to the randomness within the game itself, only the last sequence is actually used since it best represents gameplay mechanics.

Overwatch

Overwatch happens to be one of the most popular games around right now and while it isn’t particularly stressful upon a system’s resources, its Epic setting can provide a decent workout for all but the highest end GPUs. In order to eliminate as much variability as possible, for this benchmark we use a simple “offline” Bot Match so performance isn’t affected by outside factors like ping times and network latency.

Neither of these games allow Ryzen to catch much of a break but they do highlight why I’ve repeated time and again that Intel’s i5-series processors are absolute gangbusters in the gaming price / performance category. They may not have all those fancy cores but their lack of hyper threading leads to substantially better resource allocation in many games. Perhaps DX12 will change this situation in some way but right now, buying an 8-thread or higher processor exclusively for gaming is a phenomenal waste of money.

 

[SKYMTL]

System Power Consumption

System Power Consumption

I don’t typically dedicate a whole page to power consumption but there’s a pretty substantial story lurking behind the numbers you see below and how they directly relate to TDP claims from both Intel and AMD. Without getting too technical, the way these two companies go about measuring TDP is fundamentally different from one another. Intel themselves published a very comprehensive and quite neutral White Paper (PDF download) about some of the nuances of power testing a few years ago and its worth a quick read if you have a chance.

What you need to know is that TDP values are a universally poor way to determine actual power consumption for end users since they are simply thermal design guidelines that are given to system integrators. As I say in every review, TDP is not actual power consumption so don’t take it as such.

In this case it looks like AMD is publishing something quite different from Intel. As such, the Ryzen 7 1800X’s claim of 95W TDP isn’t directly relatable to the 140W of Intel’s own i7-6900K. It sure looks bloody impressive and make no mistake about it, Ryzen 7 is a relatively efficient 14nm 8-core design but there’s no secret black magic that would make its process node that much more efficient than Intel’s.

As both Intel and AMD recommend, the best way to measure true power deltas between processors is via a simple (yet calibrated) power meter plugged into a wall outlet. That’s exactly what we do but add in a controlled 120V power input to eliminate voltage irregularities from impacting the results.

The testing here is with a constantly high load from AIDA64 alongside idle conditions. At idle Ryzen’s enhanced p-states kick in and it is able to sip down an impressively low amount of juice.

Kicking things up a notch, our load results measure 900 separate logged data points (one every second) to determine a true average power consumption for the system. Here the Ryzen 7 1800X provided respectable results but nothing that aligns with the 45W “TDP” delta between it and the i7-6900K. Unlike the official TDP values would have you believe, the ultimate winner in a raw performance per watt dogfight will be highly variable, often boiling down to an application to application baseline.

One other thing I wanted to point out is just how far AMD’s CPU architecture has matured since the Piledriver-based FX-series. From a performance per watt standpoint, Ryzen is in a completely different league.

 

[SKYMTL]

Overclocking Results

Overclocking Results

Coming to grips with a new platform is never easy, particularly when it comes to overclocking and system stability. The AM4 platform and in particular the Ryzen 7 1800X I have on hand can be charitably called a challenge to overclock and there's a number of reasons for that.

First and foremost is the fact that voltages implementation is a bit different this time around with both the CPU and its associated SoC requiring voltage boosts to attain stability. Cooling is also a challenge since at 1.4V (the maximum I'd push through Ryzen) the processor puts out an epic amount of heat. And I mean a lot heat since even a Corsair H110i GTX I have on hand struggled to keep up with the burgeoning heat load.

As you can imagine based on our coverage of Ryzen's memory limitations, overclocking the DRAM is like navigating through a minefield while wearing snowshoes. AMD and their motherboard partners haven't opened up a lot of sub-timings so getting modules above 2933MHz requires you jump through a few hoops. According to several overclockers I spoke to, in order to get higher frequencies you will need DIMMS with Samsung B-die ICs whereas Hynix-based DDR4 is limited to about 3000MHz. It's increasingly looking like Ryzen has an extremely weak memory subsystem.

So what does this all mean for overclocking? I'm not quite sure yet since I'm still dialing things in. Right now it looks like overclocking will settle somewhere between 3.85GHz and 4GHz. To be clear, that's a particularly poor amount of additional overclocking headroom given how our year-old i7-6900K had no trouble going above the 4.4GHz mark.

UPDATE: So the overclocking trials have finished for the time being and the results are actually quite respectable, especially on the memory overclocking front.

Unlike with Intel platforms, right now overclocking Ryzen is an "all or nothing" affair wherein all the cores need to be overclocked or none at all. In addition, while trying to push above 4.1GHz there were over a half dozen times when the processor literally shut itself down, crashing the system without a BSOD or anything else that's typically representative of an unstable overclock. This was not temperature or voltage related, so it could be due to it hitting a hard power limit or some other speedbump along the way.

Click on image to enlarge​

The final overclock was a constant 4.1GHz while the memory was able to run at DDR4-3200 levels. One thing to note is that CPU-Z reports an incorrect vCore on the board being used here (an ASRock X370 Taichi) since the Gigabyte and ASUS boards we have on hand are reporting correctly. Also note that in the results above we changed out the GPU.

A few other things to note is that to accomplish this, the CPU voltage sat at 1.40V, the SOC voltage was sitting at 1.20V and memory voltage needed to be set at 1.35V. As for temperatures, even with the Hydro H110i at maximum pump and fan speeds the CPU hit 77C.

Right now overclocking Ryzen isn't for the faint of heart but the end result is immensely rewarding. Many of the hiccups we encountered are likely due to platform maturity rather than any massive limitations of the CPU or chipset. Temperatures are of course a concern but when aren't they when overclocking any 8-core, 16-thread part? Once the six and 4-core CPUs are launched sometime in Q2, there may be some interesting results indeed!

 

[SKYMTL]

Conclusion; AMD Rises Again

Conclusion; AMD Rises Again

I’ve been in this industry for the better part of 15 years now and in all that time, I can’t remember an instance when there’s been so much excitement building over a product launch. AMD didn’t help the situation out by layout out an endless trail of cookie crumbs for the rumor mill to actively pick over. The whole carefully laid plan could have gone horribly wrong if Ryzen didn’t live up to the stratospheric amount of hype it generated in the lead-up to today.

I use the term “hype” to loosely describe Ryzen’s long, drawn-out birth but a more accurate term would likely be “hope”. There’s hope Ryzen will bring the plucky AMD back into lock step with Intel’s CPUs, hope that competition will drive costs down and hope for that little something that’s simply called customer choice. Competition is a great thing and without AMD in contention over the last five years, it just hasn't felt the same. On a deeply personal –and perhaps unprofessional- level I have been cheering them on as well; we all like to see the underdog win. And did this particular underdog “win”? Well, that really depends on how you look at things.

Without a shadow of doubt the Zen microarchitecture is the big winner here. In nearly every situation it proved itself to be the equal to the best Intel has to offer. It is fast, efficient and almost infinitely scalable to provide a platform upon which AMD can build a wide range of enthusiast, professional and mainstream solutions. Zen is everything that people hoped Bulldozer would be and then some. Considering where they were just two short years ago this is an achievement of monumental proportions for AMD.

The foundational strengths of Zen have led to the Ryzen 7 1800X becoming what it is today: one of the best processors that money can buy. Right now it is impossible to find a CPU that can power through heavily threaded workflows so well while costing so little. Professionals and prosumers alike will appreciate everything that the 1800X can bring to the table, especially when you consider how much cash can be saved versus the i7-6900K.

To be perfectly candid with you guys, throughout the course of this review it’s been tough to step away from the subliminal need to have AMD win this round. That separation of church and state so-to-speak is essential here since it’s important to look at Ryzen from an unbiased critical perspective. Indeed there are some speedbumps lurking below the surface.

In-game performance is just about the only performance-driven metric that doesn’t fall into AMD’s happy Ryzen narrative. In titles that weren’t capped by the game engine or bottlenecked by the TITAN X, Ryzen’s winning streak came to an end. The 1800X still provided highly competitive results but in many situations it fell behind the less expensive 7700K and 7600K. This issue isn’t unique to Ryzen since even the once-mighty Broadwell-E processors had problems keeping up as well. As I said when Broadwell-E was launched, mammoth 8-core, 16-thread CPUs are great for people who need that excess horsepower but they go largely underutilized in gaming rigs.

The standouts here were the Kaby Lake processors whose high clock speeds and quad core architecture is well tuned to deliver optimal gaming performance whereas low-clocked 8-core CPUs simply aren’t. Even DX12 -which is inherently multi core aware- doesn’t change this equation in any appreciable way. Perhaps a closer look will be required, particularly at higher resolutions but in those situations the GPU will become more of a bottleneck, thus negatively impacting top-level results.

Memory compatibility could also be a major hurdle to overcome since right now. It is both confusing and frustrating. For example I tried two different quad channel 1.35V 64GB DDR4-3200 kits and the system flat out refused to POST on two 370X motherboards from ASUS and Aorus. Even ASUS’ otherwise-flawless MemOK! feature failed to save the day. These hiccups may point towards some basic teething issues for a brand new platform or they could herald a very prickly situation for end users who are trying to piece together an AM4-based system. Either way, right now the best thing to do is stick to memory that features a JEDEC-approved frequency of 2666MHz or less.

Ease of use isn’t high on the optimization front either. AMD’s suggestions for squeezing optimal performance out of Ryzen involved a process more akin to doing the Macarena than the straightforward plug-and-play approach end users crave. That process meant disabling all monitoring tools, insuring the Windows High Precision Event Timer was disabled, seting Windows to its High Performance power plan, toggling on / off SMT functions and a few other things. Many of these recommendations can be chalked up to AMD wanting their processors to be seen in the best light possible but newcomers to the PC Master Race just can’t be expected to jump through those hoops.

Pricing will of course be a large part of the Ryzen story but we can't fall into the trap of believing that a $500 CPU is a great value proposition for everyone. For people who live with parallel multi-threaded workflows on a daily basis the Ryzen 7 1800X is a no-brainer which makes Intel’s current pricing structure look absolutely preposterous. On the other hand, if you are a gamer some great bang for buck alternatives can be found with the i5-series SKUs from Intel and likely future Ryzen 5-seires and 3-series parts as well. Basically what I'm trying to say is don't let hope turn to hype which leads to a poor buying decision.

For an immature platform still in its infancy, Summit Ridge and by extension Ryzen is a remarkably well rounded solution and like any fine wine, it will only get better with age. While the 16-thread parts are just now blazing a trail that will ultimately shake the CPU industry’s pricing foundations, they won’t be right for everyone. However, after seeing what the 1800X can accomplish excitement for those six and four core derivatives. But what is clear right now is that Ryzen is indeed the real deal and Intel has been put on notice that AMD is back in the game.