Annoying flash swap: retesting Crucial’s P2 SSD after QLC downgrade
Crucial’s P2 did not become a strong contender for the title Best SSD When it was first launched, but now it is firmly on our non-recommended list. Recently, Crucial decided to replace P2’s TLC flash memory with slower and less durable QLC flash memory, making the new P2 much worse than the original version. For example, the QLC version of the drive is nearly four times slower than the original drive when transferring files. In the actual test, the reading speed is half of the original, and the continuous writing speed has dropped to a level similar to USB 2.0, only 40 MBps . This is slower than most hard drives. Unfortunately, Crucial made changes, but did not change the product name or number, and did not issue an announcement. Crucial claims that P2 will meet its specifications because the company included the performance of QLC flash memory directly in the specification sheet when it was released. But these specifications do not match the performance you see in many reviews of the original shipped drive, causing the SSD manufacturer to misleadingly replace the components again.
Sound familiar?Just like Adata is under fire for deception tactics, Crucial is now in a popular position to do so. Unlike Adata, Crucial keeps the P2 SSD controller the same, but replacing the TLC flash memory with a cheaper, slower, and less durable QLC may be more detrimental to performance.
Replacing components in any product can happen for a variety of reasons, such as ensuring component supply or cutting costs, but now even more reputable SSD manufacturers are becoming more active. Although these companies stated that the modified SSD will meet or exceed the original specifications, this is mainly only applicable to a part of the SSD performance-the comprehensive measurement sequence performance of the SLC cache. This does not reflect the true performance loss that may occur during actual use.
Based on modifications to the initial design, if you don’t know what to look for, the performance consequences can be very deceptive. Although the sequential read and write performance can meet the specification table, the random performance and the continuous write speed may be very different, which seriously affects the actual performance. This is bad news for new buyers looking for the next storage upgrade, especially if they read reviews of drives of the same name with higher benchmarks that are different from the drives they will buy in retail stores.
For Crucial’s P2, we usually don’t want companies that manufacture their own NAND to participate in this practice. However, Crucial did set a precedent last year when it released the second SKU of the BX500, replacing TLC with QLC NAND. The problem this time is that the company has not changed the model or brand at all. Instead, these new P2 SSDs secretly replaced the original models we received for review.
Crucial stated that P2 meets P2’s specifications because it considers lower QLC performance when launching the drive. In our initial review, we did notice an unusually low performance specification: the drive consistently exceeded its specification performance, leading us to believe that something was “off”. The 500GB model can only provide a write speed of up to 940 MBps, which is inconsistent with what we saw in our tests, as our sample managed to issue a write speed of more than 1.8 GBps.
This gave us a sense of anxiety when talking about the future of P2, and now we know why. Although these new QLC P2 variants may meet the company’s specification sheet, as we saw on Adata’s XPG SX8200 Pro, they definitely do not meet the performance of the original version. Unfortunately, this means that the benchmark tests you will see in every evaluation of the original P2 SSD paint a misleading picture and do not represent the actual performance you get with these drives.
Take a closer look
From the outside, the two versions of the SSD look almost the same (the original version is at the top, and the replacement version is below). The packaging, the label on the top of the drive, and even the PCB and components placed on our new samples are basically the same. The only difference is that the new QLC variant has UK/CA printed on the packaging near the model number, as well as the new firmware version. Two NAND flash packages are missing from our new sample, but this is well hidden under the drive label. The reduction in the number of chip packages is because each package of QLC NAND stores more data, thereby reducing the number of packages required. This can save Crucial’s money.
As mentioned earlier, the controller remains the same in both variants. Both use Phison’s E13T, which is a single-core, DRAM-free design optimized for cost-effectiveness and low power consumption. However, the original version uses Micron’s TLC flash memory, while the new version uses Micron’s 1Tb 96L QLC (N28A) flash memory. However, the problems that plague P2 are manifold. The performance issues we see in this new sample are not only due to the type of flash memory. It is also related to the die density, the number of die used, and based on the SMART data of the new SSD, the power limit may also be reduced.
After further digging with our new SSD analysis tool, we found that even our original sample did not have the 96L TLC claimed by the company. Our original sample was equipped with the company’s 256Gb 64L TLC (B16A) flash memory instead of the 512Gb 96L TLC that was supposed to be used. Presumably, this strategy is to use up the company’s last flash memory supply before replacing it with the 512Gb 96L flash memory that was already in mass production at that time. This means that there may be another variant of the drive with performance somewhere in between.
Since the chip density, number of chips, and interface speed directly affect the performance of the SSD, it is necessary to pay attention to these statistics. Although using higher-density flash memory can help reduce cost per GB and the number of components compared to drives built with lower-density flash memory, it can also have a detrimental effect on performance.
SSD uses an algorithm that increases data interleaving to achieve fast and responsive data access. Therefore, if the SSD controller can use more flash memory chips to obtain greater interleaving capability or faster interface speed, higher performance can be obtained. The reverse is also true, so fewer packets means slower performance.
The interface speed of Crucial’s new 500GB P2 flash memory is similar to the original version (800 MTps), but only four chips are four times denser (now 1Tb instead of 256Gb), and the QLC-powered P2 cannot be compared with the original. There are a total of 16 die, especially in continuous write workloads.
more: The best solid state drive
more: How do we test HDD and SSD
more: All SSD content