Testing thermal limits

When I unboxed the MEG X299 CREATION and saw the M.2 Xpander-Aero, I knew I had to try to set up something special. I can't look at a "graphics card level thermal solution" and not see how far it can actually go. Given the size of the single monolithic heatsink that covers the four slots, my lonely Samsung 950 Pro NVMe test drive was not going to cut it. I reached out to our rep over at ADATA with a proposal.


ADATA delivered in spades: not one, or two, or three, but FOUR of their XPG SX8200 Pro 512 GB M.2 NVMe SSDs for the purpose of seeing what the MSI M.2 Xpander-Aero is actually made of. You can check out W1zzard's recent review of the 1 TB version of the XPG SX8200 Pro here. Lets dig into the Xpander-Aero for a bit before we unbox and populate it with the XPG SX8200 Pro drives.


The MSI M.2 Xpander-Aero looks, on first inspection, just like an SFF graphics card. The short (compared to a standard GPU at least) PCIe x16 card features a black plastic shroud with faux carbon fiber patterns and a single fan centered over a cast aluminium heatsink. "AERO" is printed left of the fan, and "MSI" on the fan hub.

The back of the card has four screws secure the heatsink, as well as "M.2 Xpander-Aero" and printed certification badges. The top of the card features a 6-pin PCI power connection.



Of course, a quick glance at the rear of the card shows that it isn't a GPU, featuring a distinct lack of video outputs. Instead, the bracket is dominated by a hexagonal mesh as well as two groups of distinct holes, four circular and six hexagonal. These ten holes are LED cutouts; the four circular LEDs display drive activity for each of the four slots, while the six hexagonal LEDs are for displaying drive temperature.


Removing the four Phillips screws from the back of the card frees the heatsink. Don't forget to unplug the shroud fan as well (if you are careful, you can get away with leaving it plugged in to just fold the heatsink next to the card over). Underneath the heatsink, the MSI M.2 Xpander-Aero has a pretty interesting setup.


There are four M.2 slots, aligned vertically, and each has its own thermocouple to feed data to the six temperature LEDs on the back of the card. Other than the slots, there are two headers for case HDD LEDs, the four pin header for the shroud fan, and two small switches, one for the rear LEDs and another for the fan.


The heatsink itself is quite hefty and features the distinct granular silver texture cast aluminium often has. The heatsink interfaces with the drives through four (or maybe one large) thick thermal pads.


Back to the XPG SX8200 Pros; again, if you want a full review please check out W1zzards write-up. I will just do a quick unboxing and overview here. The SX8200 Pro is the fastest SSD the XPG brand has to offer, featuring max reads of 3500 Mbps and max writes of 3000 Mbps; the Pro essentially offers blazing performance at a great price.


Now, as is well known of M.2 drives, with great speed comes great heat. The XPG SX8200 Pro is no exception. Of course, ADATA has included a stylish DIY heat shield with each drive, and as per W1zzard's thermal testing, the included shield does mitigate thermal throttling in extreme scenarios. But for this test, we need that speed, and thus heat, to put the MSI M.2 Xpander-Aero through its paces.


I will be using three different configurations for this test, and the methodology is relatively simple. First, I will run a sequential write test without the heatsink installed. This will give me a baseline to work with and lets me see the thermal characteristics of the XPG SX8200 Pro without any mitigation. Next, I will install the heatsink and run the same test, but with the shroud fan disabled, which will show if the heatsink is capable of passively radiating the heat from all four drives away. Last, I will run the same test with the fan actively cooling the heatsink.


I am using SMART to gather temperature data for these tests; the thermocouples on the MSI M.2 Xpander-Aero are neat, but only display the average temperature across all four measurements using 10 degree Celsius increments displayed as a color on the six LEDs. This is really neat and potentially useful for general use, but isn't exact enough for this test. I would love to see these readings as individual options within the MSI Dragon Center software.

Since I will be using SMART to log temperature data, a RAID configuration is ruled out (they don't play nice together). Instead, I will start a sequential write test for each drive in sequence, running in the same order every time. For simplicity's sake, and for my sanity, I am only showing the temperature data here. If you want a more thorough look at what these drives are like in real world uses and synthetic tests, please check out W1zzard's full review.


As expected, with no mitigation for heat, the drives quickly throttle. Remember, ADATA includes a DIY heat shield with the XPG SX8200 Pro 512 GB to help mitigate thermal throttling. The XPG SX8200 Pro 512 GB has a relatively low throttling temperature of 70 degrees Celsius to probably extend its lifespan.


With the heatsink installed but the fan disabled, things fare much better than I expected. Frankly, I was hoping I could load the heatsink to capacity by running all four drives. While I did see significantly higher temperatures over the final test, the drives never came anywhere near thermal throttling. This test was conducted on an open air test bench, and a hot case with poor airflow may yield a different result.


With the fan active, temperatures never got higher than ten degrees above ambient. The fan is quite audible under heavy load, but does have a temperature-based curve, so it's barely audible at idle. I believe it would pretty much be inaudible in an enclosed case instead of an open test bench two feet from my head.


A chart of all the tests combined really shows the difference the heatsink and active cooling make under extreme scenarios. Fully utilized, the M.2 Xpander-Aero kept max drive temperatures under 40 degrees, while the bare drives idled in that same range. The second test line (green; heatsink with no fan) looks like it will continue to rise, but in fact levels out at around that final temperature; I wanted to keep the charts as short as feasible.

Of course, if you think for one second I am going to go through this whole test with four blazing fast M.2 drives without sticking them in RAID 0 to see what they can really do...


The XPG SX8200 Pro can put up some impressive numbers, although there are serious diminishing returns on multi-drive Raid configurations.

The MSI Xpander-Aero proved to be even more capable than I had originally assumed. Even when fully populated with very high performance drives, the heatsink doesn't even break a sweat. In cases with decent airflow, the fan isn't even necessary as the heatsink is large enough to passively radiate the heat away.

The MSI M.2 Xpander-Aero is a solid inclusion for the MEG X299 CREATION and lends some very real utility to the board. I want to thank ADATA once again for providing the four XPG SX8200 Pro 512 GB drives used in this test; it proved to be a very interesting addition to the normal benchmarking suite.