Ryzen 5 3600 Wraith Stealth Vs. DeepCool Tower Cooler

Out of AMD’s first batch of new 3rd-gen Ryzen processors, the Ryzen 5 3600 has become a best-seller due to its excellent price/performance ratio. It clearly offers better value than the more expensive Ryzen 5 3600X, which is less than 5% faster. On the downside, the fan included with the R5 3600 is the smaller Wraith Stealth model instead of the slightly more capable (though not by much), Wraith Spire that ships with the 3600X.

On the other hand, as for overclocking the new Ryzens, AMD seems to have squeezed the most out of them through aggressive optimization, leaving little room for manual overclocks. What you do get though, among other things, is the Precision Boost Overdrive (PBO) feature. This doesn’t boost the clocks outright but indirectly, by allowing the CPU to use more current if the temperatures allow it. You enable the feature via your motherboard’s BIOS and/or by using the Ryzen Master software from AMD.

All other things being equal, sufficient cooling will thus give your R5 3600 CPU the best possible conditions for boosting its own performance, even in the absence of traditional overclocking. It all depends on how the included Wraith Stealth performs compared to a decent aftermarket solution – which is what we intend to take a brief but hopefully revealing look at here.

Considering the budget-oriented nature of the Ryzen 5 3600, we’ve picked one of the most affordable tower-style CPU coolers on the market to compare with the Wraith Stealth. We ended up with the DeepCool Gammaxx 400 (Newegg). It’s one of the cheapest full-size tower coolers, with a 120 mm (i.e. case-size) fan and a nice-looking blue LED thrown in as a bonus. It uses a simple design that’s been around for many years and sockets, but it’s still going strong in terms of sales and popularity.

So, what sort of gains can you expect from just swapping the Wraith Stealth for this entry-level aftermarket cooler – without any manual tweaks whatsoever? Let’s find out.

Test Setup

• Motherboard: MSI B450 Gaming Plus (BIOS 7B86v1A)
• RAM: Crucial Ballistix Sport 16 GB DDR4 (2×8 GB dual channel) 3000 MHz
• Storage: Corsair MP510 M.2 PCIe 3.0 SSD
• Graphics: Gigabyte Aorus 1080 Ti Extreme 11G
• OS: Windows 10 Pro version 1903
• CPU: Ryzen 5 3600
• CPU cooler: Wraith Stealth/DeepCool Gammaxx 400

Benchmarks are run five times to produce an average. Other than swapping CPU fans and shifting between stock and PBO modes, we make no changes of any kind to the system.

Maximum Temperatures and All-Core OC Behavior

When running any sort of stress test it quickly becomes obvious that the Wraith Stealth is about as efficient as you’d expect a run-of-the-mill boxed cooler to be. In AIDA64 Extreme (CPU/FPU/Cache), the Ryzen 5 3600 reaches its 95C temperature ceiling in a matter of seconds and hovers just below that limit throughout our 10-minute test run. Meanwhile, clock speeds stabilize at around 3850 MHz on all cores.

In comparison, our affordable aftermarket cooler cuts those temps down by a spectacular 20 degrees C, i.e. from near the boiling point well into the green zone. It also adds about 100 MHz to the clocks in this high-load situation and the tower cooler allows the R5 3600 to use 10W of additional power. All settings are identical and Precision Boost Overdrive is activated in both cases.

Maxing out the CPU like this is not representative of the average gaming load, but for us, it’s a useful indicator of relative cooling performance. Unsurprisingly, the large heatsink and full-size fan from DeepCool cool the Ryzen 5 3600 much better than the Wraith Stealth. More so than expected, mainly because the Wraith Stealth performs even worse than expected. In any event, it’s clear that the added cooling headroom should allow the CPU to perform better in any situation involving heavy loads.

Cinebench R20

Moving on to something a bit more grounded in reality. Cinebench simulates rendering performance and provides scores for all cores and a single core, respectively.

As expected, our Cinebench R20 scores are slightly better when using the DeepCool tower. However, activating Precision Boost Overdrive actually lowers the multi-core score compared to just running the R5 3600 in default mode. The single-core score is still higher, but as with all single-core scores, the difference is negligible. This is all well and good since temperatures are less of an issue in single-core loads.

3DMark Time Spy CPU Score

Futuremark’s popular benchmarking software 3DMark is mainly intended for evaluating gaming graphics performance. Consequently, GPU performance weighs heavily into the total scores. However, the DX12 Time Spy test also contains a separate CPU score, which is what we’re looking at here.

The improved cooler increases the score ever so slightly, but yet again, activating PBO reduces the score by a very small amount. It’s worth mentioning that, to my understanding, Precision Boost Overdrive does not push the clocks beyond the rated maximum boost specs. It only adds current to improve the CPU’s “boost potential” if such potential is available. Consequently, this “boost potential” is lower when the CPU is already sufficiently cooled. It’s nevertheless surprising to see a consistent (if slight) performance drop when PBO is activated.

Gaming: Civilization 6 AI/Turn Times

CPU performance affects gaming in a number of ways, and it’s not just frame rates at resolutions lower than 1440p. A case in point is Civ 6 turn times, and there’s a built-in benchmark to measure those. Although it doesn’t look overly demanding, the game will tax your CPU heavily in the late stages of larger maps with many AI participants. In other words, a fast CPU will reduce your waiting time between turns.

In this test, PBO worked as intended by lowering the average turn times. There was quite a bit of fluctuation between runs, though, which is important to mention since it’s a rather short benchmarking sequence.

Gaming Frame Rates: Grand Theft Auto 5 Benchmark

We’ll stick to games with an in-game benchmark for the purpose of CPU experiments. In GTA V, we just turned off V-Sync, set the resolution to 1080p (full-screen), and left the other settings as they were. So it’s not a particularly demanding situation for an Nvidia GTX 1080 Ti, which leaves room for the Ryzen 5 3600 CPU to shine …

… and it did, once AMD’s Wraith Stealth cooler was left out of the equation. With the DeepCool tower cooler installed, GTA V frame rates improved by no less than 14% at the CPU’s default setting. At higher resolutions and settings it won’t be as pronounced, but it’s still a surprisingly large effect from just better cooling.

Conclusion

So, is upgrading the boxed Wraith Stealth cooler to something slightly better worth the effort and money? For gamers and other demanding users, the answer is most definitely yes. The sizzling temps during high loads with the Wraith Stealth are all the reasons you need. In addition, you get a slight performance boost – without even touching the clocks manually.

For any sort of overclocking, it seems that replacing the boxed cooler is an absolute must. And you don’t need anything more sophisticated than a basic tower cooler such as this one to bring the temperatures down considerably.

It does make sense that AMD ships its more affordable mid-range CPUs with a small boxed cooler, and for many users, it might be good enough. However, any mid-range gaming or workstation PC build based on the Ryzen 5 3600 will benefit from something better.

The price/performance calculation is, as of now, affected by the price difference compared to the Ryzen 5 3600X, which includes the slightly better Wraith Spire. But you will likely want to upgrade that one too, in which case it evens out.

Based on the above, it’s easy to recommend the DeepCool Gammaxx 400 because it’s affordable and does the job, even if the build quality can’t compete with a Noctua for three times the asking price. The actual performance benefit in relation to the CPU cost would probably also be questionable. For a high-end build with a corresponding budget, the story might be different.