AMD Ryzen Threadripper 1920X & 1950X Review

[SKYMTL]

Productivity Benchmarks: Blender / 3ds MAX Corona

Blender

Blender is a free-to-use 3D content creation program that also features an extremely robust rendering back-end. It boasts extremely good multi core scaling and even incorporates a good amount of GPU acceleration for various higher level tasks. In this benchmark we take a custom 1440P 3D image and render it out using the built-in tool. The results you see below list how long it took each processor to complete the test.

3ds MAX Corona Renderer

Autodesk’s 3ds MAX is currently one of the most-used 3D modeling, animation and rendering programs on the market, providing a creative platform for architects to industrial designers alike. Unfortunately its rendering algorithms leave much to be desired and third party rendering add-ons are quite popular. One of the newest ones is called Corona.

In this test we take a custom 3D scene of a room with global illumination enabled and render it out in 720P using Corona’s built-in renderer.

If your job involves rendering complicated scenes or HDR images, Threadripper is definitely the way to go from a price to performance perspective. Both Blender and Corona show that more threads can be used in an efficient and time saving manner.

 

[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.

With GIMP and many other photo manipulation programs preferring high clock speeds over multiple cores, it was obvious the Threadripper processors would suffer. Our custom photo manipulation test actually saw the 1950X and 1920X come right up against the Ryzen 5 1600X rather than the faster Intel processors.

Luckily Handbrake saved the day by allowing Threadripper to flex some incredible muscles and beat down Intel’s highest end desktop processor.

 

[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.

The hits continue with another rendering program POV Ray and Threadripper showing that it may be the best choice out there for professionals. Unfortunately WinRAR takes some wind out of AMD’s sails since the 1920X and 1950X barely match Ryzen 7 processors. This is likely due to memory latencies in the default Distributed Mode.

 

[SKYMTL]

Gaming Performance (Synthetic)

Gaming Performance (Synthetic)

3DMark Fire Strike (DX11)

3DMark Time Spy (DX12)

The initial synthetic gaming results in 3DMark look pretty promising with the Ryzen Threadripper processors remaining near the top of the charts in the DX12 benchmarks. Unfortunately, the Fire Strike benchmark brings things crashing back down to earth with neither processor being able to match Intel’s higher end chips.

 

[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.

In actual games the results don’t really fare any better with the 1950X and 1920X struggling to compete with the 1600X in Battlefield 1. However, this also goes to show that none of the HEDT processors do particularly well since the i9-7900X is just a few frames ahead. Like we keep on saying: buying a 16 thread or higher processor exclusively for gaming is just a waste of money.

Even Call of Duty Infinite Warfare backs that up with some low results for Threadripper but as usual its not too far back from the i9-7900X.

 

[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.

Deus Ex shows pretty much the same results as the last two games. Basically the Threadripper chips are just as fast as the Ryzen 5 1600X and within 7% of the i9-7900X. Meanwhile, Doom’s performance is capped by the game engine so all of the processors perform identically given the fact that we are running these games with a Titan XP.

 

[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.

Finally we start to see some separation in GTA but once again the real stars of this show are the lower priced four to eight core models from AMD and Intel. Those little CPUs are just amazing values for gaming! Threadripper keeps it close to the i9 though, which is great news. Finally Overwatch is just more of the same as it doesn’t care if you have a $200 processor or one that costs $1000 to deliver the best performance.

 

[SKYMTL]

Benchmarking Threadripper’s Memory Modes

Benchmarking Threadripper’s Memory Modes

Closer to the beginning of this review I mentioned that AMD implemented two different memory modes in an effort to give end users the opportunity to fine tuned their systems based on different usage scenarios. On one hand Distributed Mode prioritizes raw bandwidth while Local Mode is supposed to lower latencies by retaining memory transactions close to a controller’s associated core. In addition, the ASUS Zenith Extreme has something called “Auto” mode which isn’t really automatic at all….it simply utilizes the system default which is Distributed Mode.

But what does this all mean when the rubber hits the road? Let’s find out.

AIDA Memory Benchmarks

The AIDA memory benchmarks show exactly what AMD claimed: higher memory read bandwidth for Distributed Mode and much lower latency when NUMA is activated. Somewhat ironically, Write bandwidth isn’t touched at all.


Synthetic System Benchmarks

As we can see, the memory subsystem setting has very little to no appreciable impact upon synthetic testing. This actually surprised me since I was expecting an application like Cinebench to take advantage of the additional bandwidth being offered but that didn’t happen. It could very well be that the 2666MHz memory speed I have set in these tests is somehow influencing differentiations.


Real World Testing

In longer real world benchmarks the Memory Mode obviously has an effect, at least in the two applications I tested. Corona sees the advantage fall to Distributed Mode while After Effects, a notorious memory hog, obviously sees some benefits with higher bandwidth.


Game Testing

Just like Distributed Mode shows net positive reactions in real world multi threaded scenarios, games obviously see benefits from the Local Mode. The uplift isn’t all that much but the results were repeatable so there is definitely some method behind AMD’s “madness”.

The question I have to pose is pretty simple: is all of this flicking of switches, reboots and other detractors from an otherwise fluid platform experience really worth a 2-3% uplift in framerates? Granted there will be outliers that benefit even more from a NUMA setup but I personally I don’t think the fuss is worthwhile. I do however have to applaud AMD for opening up this feature, I just wish it was integrated through application detection rather than direct user interaction.

 

[SKYMTL]

Memory Speeds & Their Effect On Performance

Memory Speeds & Their Effect On Performance

Whenever I get to the comment thread of these reviews, someone always asks why we use a standardized memory speed of DDR4-2666 for every platform. The next question will always be about how higher memory speeds would affect overall performance. Well this time I decided to include a section dedicated to just that.

Ryzen’s memory configuration and inherent limitations haven’t changed since the Ryzen 7 1800X was launched a few months ago. Even with Threadripper, the CPU’s memory controllers are certified for default operation at DDR4-2400, with DDR4-2666 and DDR4-3200 being only “unofficially” supported through overclock speed bins and built-in XMP profiles. But that got me thinking: what if I had actually pushed the frequencies to DDR4-3200 or decided to keep the reference design DDR4-2400? I grabbed my trusty set of G.Skill memory, kept the timings at an aggressive 13-13-13-32 and got to work.


Synthetic System Benchmarks

The first set of benchmarks was actually pretty interesting since there really isn’t much difference in performance when going from DDR4-2400 to DDR4-3200. There are some slight performance benefits but really not all that much.


Real World Testing

Switching gears to the real world of Corona and Adobe actually shows that the Corona rendering container loves faster memory with every uptick in speed being directly correlated to a lowering of output times. Adobe meanwhile didn’t show anywhere near the same separation but there was a slight improvement when going from DDR4-2400 to DDR4-2666.


Game Testing

Gaming ended up pretty flat as well, except for that pesky GTA which seems to throw a wrench into the works whenever I benchmark it.

With all of this said and done, I can’t conclusively say that Threadripper users will benefit from spending extra money on a DDR4-3200 memory kit. At least from the testing you see above the effect is minimal at best, likely due to the fact that the quad-channel memory interface already delivers a ton of bandwidth at reasonable memory speeds. However, if I had to choose between DDR4-2400 and DDR4-2666, I wouldn't hesitate to recommend the faster of the two.

 

[SKYMTL]

Power Consumption

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.

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.

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 the wall outlet. That’s exactly what we do but add in a controlled 120V power input to eliminate voltage irregularities from impacting the results.

Naturally, these results really do require a bit of context. Even though Threadripper is built on an efficient manufacturing process, the sheer number of cores housed within each CPU package guarantees these will be among the most power hungry desktop processors ever created. And yet, both the 1920X and 1950X consume less power than AMD’s now-ancient FX-9590.

When compared against Intel’s i9-7900X, both processors may actually have a slight performance per watt lead in some highly multi threaded applications. Remember, that 1920X boasts four more threads than Intel’s current HEDT flagship and yet consumes about the same amount of power.

As for temperatures, that’s always a tricky thing to comparatively measure on processors and doing so with Threadripper is even more difficult. The problem is that none of my air coolers that I used on other processors fit onto its mammoth IHS. Hence, I was forced to use a Corsair H90 and hope for the best. It worked but the temperature results weren’t anything to be proud of. With the fans spinning in the Auto mode, both processors idled between 29.6°C and 31.1°C. Hitting the 1950X with a multi core load saw temperatures spike to 80°C while the 1920X ended up hitting about just south of that at 78.5°C. Remember, these are stock numbers running on one of the best dual fan AIOs you can but. You need some serious cooling for Threadripper, there's no two ways about that.