Cool your jets
Do you need or even want to buy a heatsink PCIe adapter for M.2 drives?
Cool Your Jets: Can the Angelbird Wings PX1 Heatsink-Equipped PCIe Adapter Tame M.2 SSD Temps?
Introduction to the Angelbird Wings PX1
PCIe-based M.2 storage has been one of the more exciting topics in the PC hardware market during the past year. With tremendous performance packed into a small design no larger than a stick of chewing gum, PCIe M.2 SSDs open up new levels of storage performance and flexibility for both mobile and desktop computing. But these tiny, powerful drives can heat up significantly under load, to the point where thermal performance throttling was a critical concern when the drives first began to hit the market.
While thermal throttling is less of a concern for the latest generation of NVMe M.2 SSDs, Austrian SSD and accessories firm Angelbird wants to squash any possibility of performance-killing heat with its Wings line of PCIe SSD adapters. The company's first Wings-branded product is the PX1, a x4 PCIe adapter that can house an M.2 SSD in a custom-designed heatsink.
Angelbird claims that its aluminum-coated copper-core heatsink design can lower the operating temperature of hot M.2 SSDs like the Samsung 950 Pro, thereby preventing thermal throttling. But at a list price of $75, this potential protection doesn't come cheap. We set out to test the PX1's design to see if Angelbird's claims about reduced temperatures and increased performance hold true.
PX1 Design & Installation
PC Perspective's Allyn Malventano was impressed with the build quality of Angelbird's products when he reviewed its "wrk" series of SSDs in late 2014. Our initial impression of the PX1 revealed that Angelbird hasn't lost a step in that regard during the intervening years.
The PX1 features an attractive black design and removable heatsink, which is affixed to the PCB via six hex screws. A single M-key M.2 port resides in the center of the adapter, with mounting holes to accommodate 2230, 2242, 2260, 2280, and 22110-length drives.
Angelbird specifically cites compatibility with Plextor and Samsung SSDs (both AHCI and NVMe), although there's nothing to indicate that any x2 or x4 PCIe M.2 SSD with the "M" or "B" connectors wouldn't work (with the latter limited to x2 speeds). The only clear limitation is the absence of support for SATA-based M.2 drives, like the Samsung 850 EVO, as the PX1 lacks SATA passthrough support found in some other M.2 adapters.
Installing an M.2 SSD in the PX1 is simple, which is good because Angelbird doesn't include a setup guide or manual in the box, requiring users to scan a QR code or visit the company's website to obtain a PDF. Users need only to insert the mounting screws into the marked hole corresponding to their SSD's length (almost all of the consumer M.2 SSDs currently use the 2280 form factor), optionally affix the second thermal pad (one thermal pad is pre-attached to the heatsink and the second included pad can be placed on the PCB if the SSD lacks chips on its underside), and then screw everything together.
Once assembled, simply install the PX1 in a x4 or larger PCIe slot and make sure that your BIOS/EFI is configured properly to recognize and provide the correct bandwidth to the newly-occupied PCIe slot.
While the PX1 looks and feels solid out of your case, it also pops when installed, thanks to a line of white LEDs around its edge that light up when the PC is powered on. There's no way to control the LEDs in software or via a user-accessible switch, but they're not terribly gaudy and should fit relatively well in the case designs of many users. Still, the lack of user control over the lighting is an early mark against the PX1.
Test Setup & Methodology
Out tests aim to compare the potential advantage of the PX1 against not only standard motherboard placement of an M.2 SSD, but far less expensive "standard" PCIe M.2 adapters as well. A number of PCIe M.2 adapters can be found online for around $20, but several motherboard manufacturers also include adapters in the box, either to add M.2 support to layouts that lack them, or enable the use of multiple M.2 devices on motherboards that feature only a single M.2 port. In our case, we're using the X4 M.2 adapter included with our ASUS motherboard.
The configuration for our tests is as follows:
|Test System Setup|
|CPU||Intel Core i7-5960X @ 4.0GHz|
|Motherboard||ASUS X99-PRO/USB 3.1|
|Memory||Corsair Dominator Platinum DDR4-2400 32GB|
|Storage||Samsung 950 Pro 512GB
Samsung 850 EVO 1TB
Samsung 850 EVO 500GB
|Graphics Card||NVIDIA GTX Titan X 12GB|
|Power Supply||Corsair AX860i|
|CPU Cooler||Corsair H110iGTX
Noctua NF-A14 iPPC-2000 PWM
|Operating System||Windows 10 Pro x64|
We performed the tests in three configurations: (1) with the Samsung 950 Pro installed in motherboard's single on-board M.2 port, (2) with the 950 Pro installed in a x16 PCIe slot via the standard ASUS adapter, and (3) with the 950 Pro installed in a x16 PCIe slot via the Angelbird Wings PX1 adapter. For the tests involving the PCI adapters, the cards were positioned beneath the GPU and without any special modification of the case to increase airflow to that location. Other sites which have covered the PX1 ran their tests in the absence of additional PCIe devices, and with the PX1 installed to ensure maximum airflow. We feel that this does not represent the configuration that is likely to be used by PC hardware enthusiasts interested in this product, hence our placement of the PCIe adapters beneath a high-powered GPU.
Performance testing was conducted with ATTO Disk Benchmark and Iometer, while stress testing to determine maximum temperature was provided by AIDA64 Extreme, which was configured to stress all system components including the 950 Pro. Temperature values were measured via a 2-pin thermal sensor affixed to the 950 Pro's controller and monitored with the ASUS AI Suite 3 software in Windows. The results of specific tests are explained beneath their graphs in the following section.
The primary purpose of the Angelbird Wings PX1 is to reduce the operating temperature of an installed M.2 SSD, so we first set out to measure thermal differences between the three installation configurations. In the chart below, Idle values were determined by measuring the 950 Pro's temperature after 20 minutes of remaining idle at the Windows desktop. This provides a baseline temperature for the 950 Pro while the PC is powered and awake but not engaged in anything other than standard background Windows tasks. On the other side of the equation, Max temperatures were measured after 20 minutes of stress testing all system components with AIDA64, providing a challenging situation in terms of both the heat produced by the 950 Pro itself as well as the ambient temperature of the case.
We can see immediately that the PX1 has a significant effect on the 950 Pro's thermals. The adapter's heatsink keeps idle temperatures 7 degrees lower and maximum temperatures a noteworthy 30 degrees below the standard motherboard placement. The use of a less expensive PCIe M.2 adapter proves its worth as well, however, with an 11 degree reduction in max temperature compared to the motherboard configuration.
It's important to note that the temperatures measured by our tests are heavily influenced by our specific hardware. The ASUS PCIe adapter features a design which elevates the SSD quite far above the PCB for good airflow, while the M.2 port on our ASUS X99-PRO motherboard places the SSD significantly closer to the PCB, which restricts air movement around the drive.
The location of the port on the motherboard itself also isn't necessarily ideal for optimal airflow in the NZXT S340's standard configuration, with the motherboard's SATA ports and the case's wire management column blocking direct airflow from the H110i's intake fans.
Therefore, while our initial results clearly show that the PX1 has no problem dissipating heat, your results will vary based on the unique combinations of your motherboard layout, PCIe adapter design, and case airflow characteristics.
ATTO Disk Benchmark
It's clear that the PX1 performs as advertised when it comes to lowering the temperature of your M.2 SSD, but the real question is how much, if at all, these lower temperatures affect the drive's performance. To answer that question, we'll first look at some sequential numbers provided by ATTO Disk Benchmark.
For the majority of transfer sizes, there is virtually no difference in performance between any of the three configurations, with the exception of the PX1 pulling just a hair ahead of the pack in read speeds. However, at the higher end of the test, we see a consistent drop in performance from the motherboard configuration only, suggesting a throttling issue due to the higher temperatures revealed earlier. Although the standard ASUS PCIe adapter was also quite hotter than the PX1 in our temperature tests, it's notable that it had no problem keeping up with the PX1 throughout the transfer sizes in the ATTO benchmark.
Iometer Random & Sequential Performance
Turning next to Iometer, we'll first look at random performance. For this test, we let the 4KB random read and write tests run for 20 minutes at a queue depth of 4 and reported average performance in megabytes per second over that period.
While we see virtually no difference in random read performance between the three configurations, the PX1 pulls ahead in random writes, besting the motherboard and ASUS PCIe adapter configurations by 18 and 3 percent, respectively.
The 256KB sequential test is where performance limitations due to temperatures really come into play, although it appears that simply moving the 950 Pro off of the motherboard and into a position of better airflow, regardless of additional heatsinks, is the key. While the 950 Pro installed in the PX1 is about 87 percent faster in both sequential reads and writes than it is when installed on the motherboard, there is an insignificant difference in performance between the PX1 and the open-air ASUS PCIe adapter.
For a passively cooled design, we were surprised to see how far the Angelbird Wings PX1 lowered the Samsung 950 Pro's operating temperatures compared to the alternative configurations. However, while the reduced temperatures may have some unknown long-term effect on the 950 Pro's longevity and reliability the reality is that virtually all of the real-world performance benefits that the PX1 offers compared to the 950 Pro's installation on the motherboard can be obtained from a much cheaper "standard" PCIe M.2 adapter like the one in our test. In fact, your motherboard box may contain an unused adapter at this very moment.
Keep in mind, however, that our ASUS PCIe adapter may have performed as well as it did thanks in no small part to the way that it holds the SSD well above the PCB. Many other PCIe M.2 adapters, including the PX1 without its heatsink, provide far less airflow underneath the SSD, so be sure to confirm an adapter's design before purchasing in order to ensure the best thermal results.
In the end, the main takeaway is that temperature, and by extension thermal throttling, may still be a factor you need to consider with M.2 SSDs. Just how much high temperatures affect you, however, will depend on the airflow design of your case, the location of the M.2 ports on your motherboard, and your willingness to reconfigure your case layout to address the issue. Adding a $75 Angelbird Wings PX1 to your setup is certainly one way to go, but you could also easily save some money with the use of a standard PCIe adapter that is designed to allow good airflow around the SSD. At worst, you could simply rig up a case fan to blow directly across your motherboard's M.2 ports.
In summary, the PX1 looks good, features great build quality, and performs as advertised. But the real lesson from these tests is to save the $75 and try a $20 PCIe adapter first. If you still need to go further after that, you can find the Angelbird Wings PX1, complete with its 10-year warranty, available now from RamCity via Amazon.