Date: Thursday , March 02, 2017
When was the last time we reviewed an AMD CPU? It has been a long time ago. Today we remedy that! And we are certainly excited about Ryzen, but let's take a moment and separate the wheat from the chaff. It is not as if AMD has not produced a new CPU since 2012, but we have not reviewed any AMD CPUs because it has not produced one that has not sucked. OK, maybe "suck" is too harsh of an overall term, but I assure you when it comes to desktop PC enthusiasts, "suck" seems to fit just fine. Intel left AMD in the dust a long time ago when it comes to top end processor performance, but Intel has gotten soft. Without AMD to jab the giant in the ass to keep it moving, Intel has become lethargic. Intel has become fat. Intel has become complacent. And AMD looks to take advantage of situation.
Ryzen represents a "clean slate" approach to building a CPU, something that Intel has not done for quite some time. AMD tells us that its goal for Ryzen was a 40% Instruction Per Clock (IPC) gain over its failed Excavator processor. It also has told us last week that it beat that goal with reaching 52% improvement in IPC. That is all great, but what does it mean to the PC enthusiast? We are going to find that out here today.
The Ryzen CPU we are using here today is the 1700X model. It has Turbo speed of 3.6GHz. We have overclocked it to 4GHz and 4.1GHz for our testing today as most HardOCP readers would be using it.
We have overclocked our Intel CPUs we are using as well. All of these CPUs have been overclocked to speeds that are what we consider stable for the application. There will be plenty of review sites covering stock clocks today, but we realize that those stock clocks are not what most HardOCP readers will be wanting to know about. What we show here are not "crazy" high overclocks and voltages. You should be able to come close to reproducing these at home with good water or air cooling configurations.
Our AMD Ryzen 1700X was stable for all sorts of long term encoding and stress testing programs at 4.0GHz. It would actually run a long time at 4.1GHz as well during encoding and stress testing, but it was not bulletproof however. In gaming scenarios, it was rock solid at 4.1GHz so I decided to give it that small bump for VR testing. This is not something I usually do, but I am a bit unsure of my cooling setup. Even pushing up to 1.5v vCore I could not get the system stable above 4.1GHz for any sort of load. Also during all this testing, I did not have access to a desktop temperature monitoring program either. It was not until last Sunday morning that I got access to temperature monitor software, and then I was bogged down in finishing up data and getting graphs and charts together for this review. I had to fly out to GDC on Monday, do the AMD Casaicin livestream on Tuesday, and finally make it back here by midnight on Wednesday, to get the final touches put on this for publication today. So....the point is that I will be getting back into Ryzen overclocking from here out for a while. Worth mentioning, the Integrated Heat Spreader is soldered on, which is a good thing from a big picture perspective.
We are not going to be doing a deep dive on Ryzen technology here today, as there are certainly others that are better equipped at doing that. However if you want to give the AMD Ryzen Press Deck a look, you can check that out at the previous link.The 1700X and 1800X AMD Ryzen Reviewers Guide are there for you to look at as well.
Our testing system configurations are outlined below. The X99 / 6900K processor is actually a 6950X CPU that has had two cores turned off in the BIOS. This 6950X processor has 25MB of cache compared to the 6900K's 20MB, so our results should have a tiny bit of advantage, but I do not truly expect this to impact scaling in any way. The X99 system also has a 32GB memory footprint as well, but none of our benchmarks actually push into this footprint, so no advantages are given based on the amount of memory.