ADATA XPG SX8200 Pro has notable performance variations between its versions
Yesterday, the hardware website Tom's Hardware reported performance differences in ADATA drives. Specifically, they tested three versions of the ADATA XPG SX8200 Pro drives, revealing significant performance variations. It appears that these XPG drives, designed for gaming, have changed their controller or the memory they use. We have also contacted ADATA for more information.
It is usually quite common to modify some of the components of the units to achieve lower prices. These modifications do not normally imply noticeable changes in the performance of the final products. But in this case, it appears that the changes made to the ADATA drive have had a significant impact on the performance of the drive.
[amazon box=»B07K1J3C23″ title=»ADATA XPG SX8200 Pro » star_rating=»none» template=»table»]ADATA XPG SX8200 Pro SSD suffers heavy performance losses depending on drive
The first version of this ADATA XPG SX8200 Pro drive uses the SM2262ENG @ 650MHz controller and IMFT 64L TLC memories. To our knowledge, the second version uses the SM2262G @ 575MHz controller with Micron 96 TLC memories. Finally the third version makes use of 65-layer Samsung TLC memories.
Under CrystalDisk, the ADATA XPG SX8200 Pro v1 drive performs the same in reading as the v3 drive, but there are discrepancies in writing. Drive v1 performs the same in writing as drive v2 but performance drops in read. But the biggest discrepancies are in 4K random performance, which ranges from 15-21% at this point.
We contacted ADATA for more information regarding this data. They told us that performance variations can depend on test conditions and platforms used. They also indicated that Silicon Motion's SM2262EN and SM2262G controllers offer the same performance according to ADATA's tests.
The thing is that SSD manufacturers do not mark read and write speeds with a clear margin of variance. They are usually marked with speeds where 'up to' is indicated which are theoretical laboratory maximums. The usual thing is usually a performance loss of less than 10% of what is marked by the manufacturer. It is still time to analyze the performance marked by the manufacturer and the actual performance.
