DRAMeXchange> Weekly Research> 2013 SSD Performance Test- 128GB

【Performance Test】2013 SSD Performance Test- 128GB

Published Aug.09 2013,17:40 PM (GMT+8)



With the 64 GB SSDs no longer able to fulfill the demands of next-gen operating systems, the 128 GB models have emerged as the new mainstream SSDs in the market. Below, we take a look at some of the major pros and cons of the different models available.

In benchmark tests of SSDs, random read and write speeds can be a very revealing indicator of performance. Following the testing of AS-SSDs with Thrd, we found that enabling NCQ through AHCI can make a noticeable difference on speed. The SanDisk SDSSDXP-120GB Extreme II, as well, showed impressive scores, and is apparently able to display improved read-write performances following necessary adjustments.


When testing data copying functions, traditional memory systems almost always perform well with ISO files. One of the major reasons for this is the generally linear nature of the saving locations. KingFast and Tigo tend to utilize similar types of components, and as such, the benchmark scores for the two turned out to be nearly identical. For Apacer Turboll AS610, the scores earned are also worthy of mentioning: Whether in linear, random, or interlaced mode, the said SSD's speeds tend to show very little difference from one another. Such results are indicative of the various adjustments that were implemented by the hardware engineers beforehand.


Under the assumption that the external variables are equal, we have found a general tendency for the performance of SSDs --even those coming from the same company-- to differ significantly from one another. An evidence for this can be seen in the varying performances of A-DATA's two distinctive SSD models. The discrepancy is attributable not only to the difference in the kinds of components used, but also to the specific settings of the internal controllers.


Plextor M5S, SanDisk SDSSDXP, and Transcend SSD740 all showed relatively weak benchmark performances; this appeared to be the case regardless of whether the linear "0 and 1" mode had been used.


Contrary to our initial expectations, the writing speeds of Tigio and Kingfast turned out to be vastly different from one another. Testing the SSDs through the random "0 and 1" mode turned out to be helpful in that a more accurate picture of the models' differences is shown. This is important considering the tendency of the reading speeds to be similar among different SSDs.


Through the CrystalDiskMark test, we found that Transcend SSD740 showed similar scores regardless of whether QD had been activated. Taken together, the scores suggest the model could be ranked at somewhere above the mid-entry level. For the Plextor M-SATA SSD, both the reading and writing speeds have been notably impressive. The QD enabled M5S's performances were mostly consistent, whether the SSD was tested using the linear or random read/write modes.


The QD enabled Plextor SSD, as can be seen above, came up with astounding results in the random reading test, which signifies that the model may have been designed with the activation of AHCI in mind. With regard to the mSATA-based PX-128, on the other hand, the test results were no worse than those of the other mainstream SSDs on the market. This goes on to show that when it comes to reading and writing performances, having SATA 6Gb/s isn't necessarily always the most important; what's equally crucial is having the right settings for the controllers.


Aside from Transcend SSDs, which showed noticeably different reading and writing speeds, all the other vendors' products displayed similar scores in the reading and writing departments. KingMAX SMU35, among others, showed a surprisingly good performance overall, with a C/P value that is decent when all the separate values are added together.


SATA 6.0Gb/s SSDs are generally able to read up to 90 thousand I/Os per second. SanDisk Extreme II memory, on the other hand, has an even higher rate, and this is obvious among the SSDs.


When it comes to the non-sorting tasks, the performance differences among vendors tend to be smaller. TeamGroup's Dark L3 is still able to perform reading and writing tasks at up to 15 thousand IOPS, which is rather impressive.