Codenamed Sandy Bridge, the second-generation Core architecture was released way back in early 2011 and may represent Intel's most impressive enthusiast processor series of the last decade. Positioned as a replacement for 2009's Core i5-750, the 2500K was priced at $216 yet it outpaced the $1,000+ Core i7 975 Extreme Edition and consumed considerably less power. It's no surprise that so many system builders have invested in the chip.
Although five years have passed, if we're being honest, the difference between today's seventh-generation Core i5-7600K and the 2500K isn't that big when they're matched clock-for-clock. Nonetheless, many Core i5-2500K owners have been wondering if it's time to upgrade, specifically asking whether they should move to the ultra-affordable Ryzen 3 1200 or the higher-threaded R5 1400. Likewise, many of you still making do with AMD's FX-4000, 6000 and 8000 series CPUs have also been asking what you can expect from an upgrade to Ryzen 3.
To answer your questions, we're back with the figures from another nine games benchmarked at 1080p. For this round of testing we overclocked the Core i5-2500K and FX-8370 to 4.4GHz (a particularly mild boost for the Core i5) while the Ryzen 3 1200 and Ryzen 5 1400 were set to 4.0GHz. We also included the Pentium G4560 but it ran at its locked speed of 3.5GHz. Right then, on with the results...
Ryzen System Specs
Kaby Lake System Specs
AMD FX System Specs
Sandy Bridge System Specs
When playing Battlefield 1 on the GTX 1060, every CPU except the FX-8370 was able to max the card out. I've found previously that the FX series really struggles in Battlefield 1 and frame drops are a common occurrence. The Core i5-2500K on the other hand had no trouble matching the Ryzen CPUs, though neither did the G4560.
Moving to the GTX 1070 we see the Pentium G4560 starts to fall behind and interestingly the FX-8370 managed to pull ahead now though the old AMD CPU does trail the Ryzen and Core i5-2500K CPUs by a noticeable margin.
Installing the more powerful GTX 1080 we start to see a real weakness in the FX-8370 and shockingly this eight-core/four-module CPU struggles to beat the dual-core Pentium G4560. On average, the 2500K delivered 28% more frames with the GTX 1080. The 2500K even edged out the R3 1200 with a few extra frames and it wasn't much slower than the R5 1400. This here is exactly why so many people have so much love for the 2500K and of course its bigger brother, the 2600K.
F1 2016 has even more interesting results to go over. With the GTX 1060 installed, the FX-8370 again struggles to extract maximum performance and falls a few frames behind the pack. Not a big deal, that is until you install something with a bit more oomph.
A card like the GTX 1070 for example. Here the FX-8370 hits the same 55fps minimum as the Pentium G4560 though the average frame rate was 14% lower. You might think with a tighter margin between the minimum and average frame rates the FX-8370 provided a smoother experience, but that wasn't the case.
Meanwhile, the Core i5-2500K matched the Ryzen 3 1200 and Ryzen 5 1400 with what was a very impressive showing of 94fps on average. The 2500K wasn't done yet either. With the GTX 1080 installed we see that the 2500K actually pulls ahead of the Ryzen CPUs, offering a minimum frame rate of 87fps. So if you have an overclocked 2500K and you play F1 2016, I'd probably hold off upgrading to a modern sub-$200 CPU.
Far Cry Primal is a heavily GPU-bound game and even so the FX-8370 still manages to stand out like a sore thumb. Intel always seems to have an advantage in this title so it's probably not hugely surprising that the Pentium G4560 outclasses the FX-8370 every step of the way. The Core i5-2500K also gives the Ryzen CPUs a hard time and with a fairly sensible configuration using the GTX 1070 the Ryzen 3 1200 at 4GHz couldn't keep pace.
Benchmarks: Total War Warhammer, Overwatch, The Witcher 3 Wild Hunt
Total War Warhammer is the destroyer of CPUs though this time the FX-8370 does pretty well to hang in there with the G4560. That said, even with the GTX 1060 handling rendering, the R3 1200 was 20% faster than the FX chip, so a big upgrade there. That said, as we've seen multiple times now, the Core i5-2500K also does well for itself.
In fact, once we increase the GPU power with the GTX 1070, the 2500K is now seen matching the Ryzen 5 1400, making it a little faster than the R3 1200. The FX-8370 is still stuck averaging around 60fps along with the G4560.
We're pretty much CPU bound with the GTX 1070, so upgrading to the GTX 1080 offers nothing here.
Folks with a GTX 1060 or slower will be served very well by any of these CPUs when playing Overwatch. All of the chips pushed over 100fps at all times.
Moving to the GTX 1070 shook things up a little though interestingly the FX-8370 managed to stick with the Ryzen 3 1200 and turned in an impressive average of 185fps. The 2500K was faster again, though not by much this time. Meanwhile the Ryzen 5 1400 was able to boost the minimum frame rate by 12% over the 2500K.
Then with the GTX 1080 we find what looks to be the limits of the FX-8370. Its minimum frame rate came up quite a bit though the average was much the same, so when using more powerful GPUs the Ryzen 3 CPU will offer more headroom in this title.
The Witcher 3: Wild Hunt is mostly GPU-bound in our testing, even at 1080p. Please note that I'm using the maximum in game quality settings which includes HairWorks. Using the GTX 1060 we see much the same performance on all five CPUs.
It was a similar story with the GTX 1070 though the G4560's minimum frame rate does slip behind here. When testing with the GTX 1080, the Ryzen CPUs and the 2500K were still able to get the most out of this high-end GPU.
Benchmarks: Rainbow Six Siege, World of Tanks, Counter Strike
Rainbow Six Siege shows competitive performance using the GeForce GTX 1060. The Ryzen 3 and Ryzen 5 CPUs deliver the best minimum frame rates, but performance is much the same across the board. Even with the GTX 1070, results remains quite similar across the five CPUs tested, though we did see more variance here and the Pentium G4560 and FX-8370 are starting to lag behind. Meanwhile, the 2500K matched the R5 1400's average frame rate but did dip slightly lower than the Ryzen 3 1200.
Moving to the GTX 1080, we saw further performance gains with all five processors though the G4560 was already giving us everything it had. You might have noticed that although the FX-8370 is now averaging 154fps we still saw frame dips down to 118fps and that made the FX CPU much slower than the 2500K and Ryzen CPUs.
World of Tanks isn't really that demanding on the CPU or GPU, though the FX-8370 makes it look like a real chore, even with the GTX 1060. I realize that we've seen some pretty poor performances so far but I was pretty surprised by how poorly the FX processor does here. Granted we are pushing over 100fps at all times, but when compared to the competition it surely putts along.
We see that the FX-8370 finds its limit with the GTX 1060 and it's much the same for the GTX 1070 and GTX 1080. The G4560 maxes out with the GTX 1070 and likewise really with the Ryzen and Core i5 CPUs.
World of Tanks favors single thread performance, or at least frequency, and this hands the Core i5-2500K a slight performance advantage in this title.
Wrapping up the benchmarks we have Counter Strike: Global Offensive and like World of Tanks strong single thread performance really counts for a lot here. As a result the Core i5-2500K looks really strong, it does only match the minimum frame rate of the Ryzen 3 and 5 CPUs, seems odd saying only, but that’s because the average is at least 8% higher with the GTX 1070 and 1080.
Taking the Ryzen Plunge?
The point of this comparison was to see what kind of upgrade Ryzen 3 might provide for those still gaming with much older AMD FX or Intel Core i5-2500K processors. Again, we've tested these chips in nine strategically selected games using several tiers of graphics cards. It's worth noting that I'm generalizing with the FX-8370 as it really covers that entire series.
Since we're already talking about it, let's start with the FX-8370's results, and again this information should really apply to anyone using the AM3+ platform. It is my opinion that Ryzen 3 offers a solid upgrade over the FX series, but how big of an upgrade will depend on the games you play and what kind of graphics card you're using.
With a card that's equivalent to the GTX 1060 or slower, the gains for most games will be quite slim. That said, if you play titles such as Counter Strike: Global Offensive, Total War Warhammer or even World of Tanks then you will see a noteworthy gain even with the GTX 1060. The gains will be even greater for those using an FX-6000 or 4000 series CPU.
As the GPU power increases there is a good chance for Ryzen 3 to make noticeable improvements so if you plan to get yourself a GTX 1070 or maybe a custom board partner Vega 56 card in a few months time then I highly recommend switching to Ryzen 3. Of course the good news here with this upgrade option is that you can still increase the CPU power quite substantially down the track by getting a six or even eight-core Ryzen processor, or quite possibly even a second-generation Ryzen part.
If I were still using an FX series processor today, I'd be jumping on the Ryzen bandwagon faster than you could call me a fanboy.
The Core i5-2500K, however, simply refuses to die. It's the little engine that could and by the looks of it, always will. If you're still holding out for a cost effective upgrade, well it’s business as usual. Unless you're jumping to a modern Core i7 or maybe the Ryzen 5 1600, the gains simply aren't there.
There were few instances where the 2500K trailed the Ryzen CPUs, though when it did the margins were small. After seeing this, I'm keen to do an in-depth 2500K revisit and compare it to CPUs such as the R5 1600 and i5-7600K. Maybe I'll make time for that once I'm done with Vega.
A few side notes... If you want to see how Ryzen 3 stacks up against higher-end CPUs please check our day-one coverage and the gaming-focused followup. Likewise for price vs. performance data and power consumption figures, though you won't find data for the 2500K and FX-8370 there.
In short, the 2500K is still quite fuel efficient by today's standards whereas the FX-8370 looks sluggish by comparison. Now we've covered Ryzen 3's gaming performance from just about every angle, so potential buyers should have a good idea of what they're getting.
- Ryzen 3 1200 Amazon, Newegg
- Ryzen 5 1400 Amazon, Newegg
- Core i5-7400 Amazon, Newegg
- Pentium G4560 Amazon, Newegg
Later this week I have a 30-game benchmark piece looking at the Vega 56, 64 and 64 liquid-cooled, which will be compared against both custom and Founders Edition models of the GTX 1070 and 1080. It's going to be epic!