- Meet Ryzen 5
- First up: Productivity benchmarks
- Gaming performance: The big question
- Why you don't need to freak out
First up: Productivity benchmarks
Our benchmarking begins with a battery of productivity tests. First, Cinebench for multithreaded performance, then Blender and POV-Ray for image-rendering chops. We add Handbrake and Adobe Premiere CC 2017 for video encoding.
Our first test is the ever-reliable Cinebench R15, which is made by Maxon and based off a real rendering engine used in its Cinema4D parts. You might recall that this benchmark was fingered by the FTC over alleged Intel benchmark improprieties, but Maxon has claimed its innocence.
The first result you see is Cinebench R15 restricted to a single thread. The Core i5-7600K comes out on top, which is to be expected given its higher clock speed and higher instructions per clock. For the most part, the Core i5-7600K sits at 4.2GHz at almost all loads all the time. The Ryzen 5 1600X bounces around, hitting 4.1GHz clock speeds only on occasion. You’re looking at maybe a 12-percent difference, which ain’t bad, but in the end, still second-best.
We retested on Cinebench R15, but this time with all the CPU cores and threads available. You’re seeing roughly an 80-percent difference in performance when all cores are hot on Ryzen 5 and the Core i5. Let’s say that again: An 80-percent difference. That’s just a crushing blow to Core i5 and pretty much frames how this battle is likely to shape up: Give up a little single-threaded performance for a huge bump in multi-threaded performance.
Blender is a popular open-source 3D renderer that’s used in a lot of indie movie productions. Like most 3D production apps, it loves cores, but I’ve found it not to scale quite as well as Maxon’s rendering engine. Still, Ryzen 5 is clearly in front by a huge margin and finishes about 50 percent faster than Core i5.
POV Ray performance
Our last rendering test uses a professional application called Persistence of Vision Raytracer, or POV Ray. It actually hearkens back to the days of the Amiga. Its built-in benchmark spits out a score that tells you how many pixels per second are rendered by a CPU.
First, single-threaded performance, which looks an awful lot like Cinebench R15’s single-threaded results. The Kaby Lake Core i5 has the advantage. Haters, say what you will, but those 7th-gen chips are mean little CPUs.
With POV-Ray, like all professional applications, the number of cores and threads available boost performance. The multithreaded results say it all: The Ryzen 5 is pushing out almost 50 percent more pixels per second during its render job. It’s another crushing blow to Core i5, showing you the weakness of a straight quad-core chip with no Hyper-Threading.
If you’re like us, you immediately uninstalled the annoying brand-name Zip program on your PC and pointed your browser at the free and fast 7-Zip. The free app features a built-in benchmark that measures the speed of a machine at compressing and decompressing files. Ryzen 5 picks up Core i5, throws it into the audience, then jumps out of the ring to smash a break-away chair over its quad-core head.
I haven’t always used Geekbench in CPU reviews, as the test has been somewhat controversial in the past. The latest version still takes algorithms its creator (Primate Labs) thinks are relevant in image editing, encryption and the like, and measures how well a CPU can run them. Primate Labs has changed how the test looks at CPU performance, however, and most consider that to be an improvement.
The latest version specifically addresses Ryzen performance, so I was interested to see the results. Ryzen 5 outpaces Core i5, but even though this is a multi-threaded test, I’m not seeing the scaling result I expected. That means either the test’s algorithms don’t particular favor Ryzen, or it just doesn’t scale with the core and thread counts that well. I’m inclined to believe the latter, as the difference between a Core i5 and Core i7 (which adds Hyper-Threading) is maybe 10 percent.
Perhaps more relevant is the single-threaded result from Geekbench 4.1, which mostly mirrors the single-threaded performance we’ve seen from Cinebench and POV-Ray. In fact, Ryzen 5 seems to do even worse, trailing Core i5 by just over 20 percent. If Geekbench’s choice of algorithms are a reliable indicator (and it uses a dozen different algorithms), the world runs just slightly better on Intel.
PCMark 8 Performance
PCMark 8 tries to replicate real-world use by not just using small algorithms, but also creating small simulated applications to measure performance.
First up is the PCMark 8 Creative Conventional test, which throws various workloads at a system, including encoding, video conferencing, and photo editing. In the end, Ryzen 5 has a slight advantage. Although there is an encoding portion, I’d take that to mean the creative mode is pretty lightweight and doesn’t favor multiple cores much. The takeaway from this is, if you’re editing photos and doing quick video edits with the free software that came with Windows, both CPUs will do the job.
PCMark 8 Work focuses on productivity tasks like browsing, video web conferencing, and driving a spreadsheet. You could do this work on just about any CPU, and it wouldn’t make working in a row of cubicles any better.
To measure encoding performance, we use the popular and free Handbrake to convert a 30GB 1080p MKV file using the Android Tablet preset. Handbrake loves, loves, loves CPU cores, and it shows. Ryzen 5 smokes Core i5 by about 70 percent in encoding time. Again, 12 threads vs 4 threads is not even a contest.
Adobe Premiere Pro Creative Cloud 2017
For a video test, we took an actual published project from our video colleagues at IDG.tv that was shot on a 4K Sony camera, and we tasked both machines to encode it using the Blu-ray preset in Premiere Pro CC 2017. Premiere Pro lets you encode using the GPU, which tends to offer a huge performance decrease in render time, or the CPU, which is still generally considered to offer the best visual quality. Using the CPU for this first chart, you can see the Ryzen 5 flexing its core count over Core i5, offering a massive decrease in render times. If time is money, Ryzen 5 gets you more money.
Realistically, however, many people will give up visual quality for the speed of a GPU render. Although the CPU matters a little less, we’re still seeing roughly a 40-percent bump in GPU rendering speed by using Ryzen 5 vs. a Core i5.
We know you want to know about gaming performance, so keep reading.