The Team Group MP32 is a cost-effective SSD design built upon a Phison E8 controller and Toshiba TLC flash. For cost reasons, a DRAM chip is not available, and the interface to the host system is limited to PCIe x2 instead of x4 like on some competing drives. We talked about DRAM cache in depth earlier in this article, and it can have a huge impact on synthetic testing. Many standard benchmark apps like AS SSD, CrystalDiskmark, and ATTO use small test area sizes that yield extremely unrealistic results for drives without DRAM cache as compared to what you will see when testing with larger sizes. The underlying reason is that the controller itself has a small portion of memory that is used in the same way as the DRAM cache on DRAM-less drives, but since the capacity is too small to store the whole mapping table, performance will progressively get worse with larger test sizes.

Generally, and the Team Group MP32 is no exception here, the lack of DRAM mostly affects performance of random writes; i.e., writing small data chunks spread out over a large area of the disk (multiple GB). When benched with our regular test area size of 128 GB, we saw pathetic numbers (that are of course still much better than with HDDs) that are nowhere near what we've come to expect from modern SSDs. After some consideration, we added a second data point for these results, testing at a smaller size of 16 GB, which should be more in-line with what the MP32 will see in actual usage. At this smaller test size, the SSD does a lot better, but the synthetic results are still not as fast as competing drives with DRAM. Nearly all of today's write-intensive consumer applications fall into just two categories: large sequential writes (ISO/game/disc image/video copy and large software installations) and small random writes (small software installation, extracting or copying lots of small files).

Sequential performance of the MP32 is excellent, but slightly held back by the interface which is only x2. The differences are relatively small, though, and only matter when it comes to sequential reads. Our real-life testing shows results that are "close enough" to competing drives, especially when you consider that the MP32 is more affordable than some of those drives. Some applications do write to larger disk areas, but are constrained by the rate at which they can generate data (for example, BitTorrent downloads). In the enterprise world, things are different, of course. Large databases with many concurrent writers are the norm as are busy file servers with many connected clients for which the MP32 is definitely not an option.

On average, in our real-life testing, we see the MP32 right in the middle of the bulk of our NVMe drive test group—most of these drives are PCIe x4 with DRAM, so not much performance is lost because of the transition to x2 without DRAM. The fastest drives we ever tested are about 7% faster, which isn't a whole lot but could still be significant, depending on your main applications.

The Team Group MP32 did extremely well in our thermal testing. Even without a heatsink and any airflow over the drive, we never saw it throttling, which is great, especially if you are looking for an M.2 NVMe drive that will see serious load in a badly ventilated case. Sequential writes are good for a TLC-based drive. The pSLC buffer has a reasonable size with 12 GB, but we wouldn't mind it being a few GB bigger. When writing directly to flash, the MP32 delivers a solid 450 MB/s, which is faster than many SATA TLC SSDs.

What really matters most in this market segment is pricing. With $120, Team Group has priced the MP32 reasonably, much lower than the drives from Samsung, but strong competition comes from ADATA and Crucial. The recently-released Crucial P1 is built using QLC flash, which unlocks a whole new world of achievable price points. For example, the Crucial P1 500 GB costs only $90 and comes with a large pSLC buffer and DRAM. On the other hand, sustained writes are much lower on the P1 due to the additional slowness of QLC when writing to it directly. The other major competitors are the ADATA SX8200 ($110) and the new ADATA SX8200 Pro ($115). At those price points, both are much faster on paper, a few percent faster in real life, and more affordable—all at the same time. This means pricing needs to come down for Team Group's MP32, clearly below $110, possibly below the psychologically important $100 mark, which should bring in massive sales and would help potential customers look over the lack of DRAM and PCIe x2.