Getting this cooler working was a bit of a struggle. It was some parts human error, on our end, and some parts mechanical error. This thing is a $100 cooler from Aliexpress, and it uses both open loop liquid cooling for a few of its pipes while also using traditional air cooling and heatpipes. We had some small (read: significant) leaks during our livestream, and after the stream, we discovered that the screws securing the inlet manifold to the tower were loose, causing significant leakage as the water filled the pipes. After fixing this, we were finally able to fully test this truly unique hybrid water-air cooler.

The cooler is an interesting one. We’re planning a separate tear-down of the cooler to see what’s going on under the coldplate – likely not much – but for now, we’ve done exhaustive thermal testing under various conditions. Some tests were just straight pump/reservoir hookups to the cooler, while others included a 360mm radiator and 3 high-end fans. The W120 has been sitting on shelves for a while, clearly, as it was first shown at Computex 2011, and the box we received had dried thermal paste and yellowing on the product box. We still wanted to test it, as the unique combination of G-1/4” fittings, open loop support through 4 water pipes, and traditional air cooling meant the cooler could perform peculiarly. You’d assume that there’s a reason this isn’t really done, but we still wanted to find out why.

Cooler Master’s H500M is the 18th addition to our “Cases Named H500” chart. The H500M was shown at CES 2018, and follows-up the initial H500P, the H500P Mesh, and the unrelated H500 and H500i cases from NZXT. This is Cooler Master’s high-end solution, shipping at $200 and including user-swappable glass or mesh front panels, with the mesh panel pre-installed in a default configuration. Today, we’re reviewing the Cooler Master H500M enclosure.

Cooler Master’s H500M officially launches for product availability to consumers in the second week of June, just after Computex ends, and carries an MSRP of $200. For clarity, this is a different product than the H500P Mesh that we previously reviewed, although it does ship with a mesh front by default. The H500M also includes a swappable glass front, and otherwise primarily differentiates itself with additional gloss and ARGB support and controllers.

From the ARGB side, software is still to come, and immediate compatibility includes ASUS motherboards. Cooler Master is working with other vendors on further integration. For our purposes today, we’re more focused on overall build quality and thermal performance; besides, we’ve got Computex and flights to Asia breathing down our necks, so we’ll stick with what we’re good at.

NZXT opened their revamped H series of cases a few months ago with the H200i, H400i, and H700i, which are all mostly differently sized versions of the same case. The H500/H500i is a brand new addition--no, not that H500--and NZXT has made some tweaks since the first batch. The NZXT H500 is an S340 replacement, priced at $70 MSRP for the H500 and $100 for the H500i (which includes a “smart” device and RGB LED strips).

We liked the H700i overall, but we had some criticisms, mostly about the “i” representing the included smart device. NZXT told us they listened, so let’s start by checking off those earlier complaints.

It’s been a long time since we’ve reviewed any mini-ITX cases. The standard system that we use for testing ATX cases includes a full-sized GPU, PSU, and CPU cooler, which may or may not fit in small form factor cases, as well as an ATX motherboard that definitely won’t. Even if our components were small enough to fit, ATX and mini-ITX enclosures are like apples and oranges--SFF cases often have specific uses and different priorities than standard mid-towers.

Enough time has passed that it’s worth it to put together a separate ITX benchmarking system with a separate table of results to compare. To start off our database, we’re doing a roundup of three not-so-new cases from our backlog: the Thermaltake V1, Silverstone SG13, and the Cryorig Taku. This will start our charts, and we intend to work toward expanding those charts with the full suite of cases, as usual, including several upcoming products at Computex.

The NZXT M22 is one of the stranger liquid coolers made by a relatively large liquid cooling manufacturer. NZXT dumped Asetek for this 120mm closed-loop cooler, instead opting for a pump-in-radiator design that circumvents Asetek patents and permits sale in the US. The M22 is a complement to NZXT’s Asetek products at the high-end, but comes in at $100 and 120mm. That’s a bit high for a 120mm liquid cooler, particularly considering that competition from EVGA’s CLC 120 comes in at $70 and is made by the familiar Asetek, but its performance may make up for the price differential. Today, we’ll find out.

Primary competition in this price class includes NZXT’s own Kraken X42, a 140mm Asetek-made design, and 240mm units from the same price class. NZXT’s M22 ships for $100 MSRP, and at that price, it’s competing (strictly in price) with the likes of the EVGA CLC 240, the Corsair H100i V2, and NZXT’s units. If we look strictly at size class, the EVGA CLC 120 competes most directly at $70. Despite its low price, that’s still a modern Asetek unit; it uses the same pump as any higher-end cooler, just has fewer fans. It’s not cheap garbage – it’s not something we recommend, either, but it’s not going to fall apart.

It’s a fierce market at $100. Even air coolers would reach equivalence or superior performance than NZXT’s M22. They’re going for one demographic, and one only: Has RGB LEDs and is exactly 120mm. That’s it. That’s the demo. If you’re not that, it’s really not worth the time or money to grab the M22.

To NZXT’s credit, the LED integration is the best-in-class for a 120mm liquid cooler. It’s also expensive, so that makes for an odd combination of size and price.

Intel’s Pentium G line has largely managed to hold-on as one of the better buys of the past few years. There was a brief period where the G3258 made a lot of sense for ultra budget-minded buyers, then the G4560 recently – particularly at the actually good price of $60 – and now Intel has its Pentium G 5000 series. The G4560 had stunted growth from limited stock and steep hikes on MSRP, forcing people to consider i3s instead, up until R3s shipped. The 4560 remained a good buy as it dropped towards $60, fully capable of gaming on the cheap, but it is now being replaced by the units we’re reviewing this month.

We’re starting with the Intel Pentium G5600, which is the most expensive of the new Pentium Gold line. At $95, it’s about $40 more than the G4560, $10 more than the G5500, and $20 more than the G5400. The R3 1300X is about $105, and the R3 1200 is about $95.

NZXT’s Kraken X72 closed-loop liquid cooler is another in the XX2 series, following the 280mm X62 that we previously reviewed. The X72 is a 360mm cooler, putting it in more direct competition with the Corsair H150i Pro (the first to feature a 6th-gen pump) and Fractal S36, and indirect competition – in performance only – with the EVGA CLC 280.

NZXT’s X72 costs $200, making it one of the most expensive CLCs on the market. The Floe 360 lands at around $184, the EK Phoenix 360 – a semi-open solution – is the only one that lands significantly higher. The X72 still uses the same pump design as when we tore-down the X42, running Asetek’s 5th Gen pump and a custom, NZXT-designed PCB for RGB lighting effects. Functionally, 5th Gen has proven to be marginally superior – technically – to its 6th Gen for outright cooling performance. We’re talking nearly margins of error. The newest generation is presently only used on Corsair’s H150i and H115i Pro products, as Corsair largely dictated what went into the 6th generation. Major differences are made-up by the metal impeller, similar to the one used by Dynatron in old Antec Kuhler products, rather than a 3-prong plastic impeller. These don’t perform differently in terms of thermals, but there should be reduced susceptibility to heated liquid, and theoretically reduced hotspots as a result of the new 6th Generation design. That doesn’t manifest in outright performance, but might manifest in endurance. We won’t know for a few years, realistically.

Our primary tests for the NZXT Kraken X72 review and benchmark include the following:

  • 100% fan / 100% pump
  • 100% fan / silent pump
  • 63% fan (40dBA)

We’ve previously tested custom copper integrated heat spreaders (IHS) for Intel, primarily the unit sold by Rockit Cool for LGA115X CPUs. Our findings of the custom copper IHS (sold here) for the i7-8700K were that, generally, it was a fun, worthwhile project at $20, but that the thermal improvement was not game-changing. It was still impressive, though, as we monitored between 4-5 degrees Celsius improvement from the IHS replacement on the 8700K, partly benefiting as a result of the increased surface area over the stock Intel heat spreader. That’s a lot of uplift for something that isn’t a CPU cooler, and if you’re up against hard requirements for noise in your system, it could allow for just enough headroom to slow-down the case fans a bit more.

Ryzen is different, as its heatspreader is one large block, as opposed to a machined block with cut-outs and dips and generally smaller surface area. Rockit Cool improved on Intel IHS performance by increasing surface area, but had little to improve on with AMD’s. Both Intel and AMD use copper IHS units, but all of them are nickel-plated. This shouldn’t impact performance significantly and helps with cleaning.

Today, we’re benchmarking a custom copper IHS for AMD Ryzen CPUs and APUs, using the Rockit Cool copper IHS on an AMD R3 2200G that we previously delidded and benchmarked.

The AMD R5 2600 and 2600X are, we think, among the more interesting processors that AMD launched for its second generation. The R5 1600 and 1600X received awards from us for 2017, mostly laying claim to “Best All-Around” processor. The 1600 series of R5 CPUs maintained 6 cores, most the gaming performance of the R7 series, and could still capably stream or perform Blender-style production rendering tasks. At the $200-$230 price range, we claimed that it functionally killed the quad-core i5 CPU, later complicated by Intel’s six-core i5 release.

The R5 2600 and 2600X have the same product stack positioning as the 1000-series predecessors, just with higher clock speeds. For specs, the R5 2600X operates at 3.6GHz base and 4.2GHz boost, with the 2600 at 3.4/3.9GHz, and the R5 1600X/1600 operating at a maximum boost of 4.0 and 3.6GHz, respectively.

Reviewing the AMD R7 2700X was done outside of normal review provisions, as AMD didn’t sample us. We’ve had the parts for a month now, and that has meant following development, EFI updates, and more as they’ve been pushed. We have multiple chips of every variety, and have been able to cross-validate as the pre-launch cycle has iterated. Because of the density of data, we’re splitting our content into multiple videos and articles.

Today’s focus will be the AMD R7 2700X and R7 2700 reviews, especially for live streaming performance versus the i7-8700K, gaming performance, and production (Blender) performance. Most importantly, however, we dedicate time to talk about the significant improvements that AMD has made in the volt-frequency department. At a given frequency, e.g. 4.0GHz, Ryzen 2000 operates at a heavily reduced voltage versus Ryzen 1. We’ll dig into this further in this review, but check back later for our R5 2600X and 2600 reviews (combined in one piece), including 2600X vs. 8600K streaming benchmarks. We’re also looking at VRM thermals, motherboard PCBs and their VRM quality, memory overclocking and scalability (in this content), and more.

There is a lot of confusion about AMD’s branding – Zen 2 vs. Ryzen 2 vs. Zen+. We’re calling these CPUs “Ryzen 2,” because they’re literally called “Ryzen 2X00” CPUs. This is not the same as the Zen 2 architecture, which is not out yet.

Note: For overclocking, we only OC one CPU of each core count – so just the R7 2700X or R7 2700, but beyond validation of maximum frequency, there’s no need to OC both and run each through 20 hours of testing.

Page 1 of 12

We moderate comments on a ~24~48 hour cycle. There will be some delay after submitting a comment.


  VigLink badge