Reading Guide: AS SSD will first continue to write to the tested partition in terms of the size of 16MB, generate a file that reaches the size of 1GB, then read the file with the same unit size, and finally calculate the average score and give the result.
The test file will be deleted immediately after the test.
We often use AS SSD Benchmark when testing SSD.
Through this software, we can get some general information about the reading and writing of SSD.
However, some friends with high skills can judge the performance of SSD by the 4K reading and writing performance tested by this software.
Today we’re going to take a look at AS SSD Benchmark and the so-called “4K performance.”
AS SSD Benchmark is the software commonly used by many of us to measure the speed of SSD.
But many of us just look at the test data but do not understand the meaning of each test project, let alone the reading, writing, and application of SSD, which leads to some misunderstandings.
Let’s first take a look at the specific meaning of several test projects of AS SSD Benchmark.
- Seq (continuous read and write): AS SSD will continue to write to the tested partition in terms of the size of 16MB, generate a file that reaches the size of 1GB, then read the file with the same unit size, and finally calculate the average score and give the result. The test file will be deleted immediately after the test.
- 4K (4k single queue depth): random single queue depth test. The test software generates the 1GB size test file with the unit size of 512KB and then carries on the random 4KB unit size write and read test within its address range (LBA) until it runs through this range, and finally calculates the average score to give the result. Due to the generation steps, this test will generate a total of 2GB data writes to the hard disk, and the files will be temporarily retained after the test.
- 4K-64Thrd (4k 64 queue depth): random 64 queue depth test, the software will generate 64 16MB size test files (total 1GB), then write and read in these 64 files in the unit size of 4KB at the same time, and finally take the average score as the result to generate the data write the amount of 2GB. The test file will be deleted immediately after the test is completed.
- Acc.time (access time): that is, the data access time test, which reads the entire address range randomly with 4KB as the unit size, (LBA), with 512B as the write unit size, randomly writes within the reserved 1GB address range, and finally gives the test results with average scores.
- Score: something like your test score in the school.
Most rookies look at the continuous reading and writing speed during the test.
The rookie with higher skills is to look at 4K and 4K-64Thrd here, looking at the so-called “4K performance” to judge whether SSD is good or bad.
Let’s find out what 4K is.
In SSD, 4K is the smallest read and write unit of SSD.
For example, if we need to write 2K of data, we actually have to write 4K; if we need to write 13K of data, we have to write 16K of data (write magnification is not considered here).
In view of this, the data we write is made up of countless 4Ks.
4K performance includes 4K random and continuous read and write performance.
The performance of 4K also shows the read and write performance of SSD.
Therefore, it is necessary to test the performance of 4K.
But the source of what we often call “4K performance” is AS SSD Benchmark, in a narrow sense, but it is actually 4K random read and write performance, which represents the data throughput capacity of the hard disk (measured in IOPS), and relative to continuous read and write performance).
So what is continuous read-write performance and random read-write performance?
Continuous read-write performance is the performance of reading and writing sequentially, while random read-write performance is the performance of random reads in a range.
Random reading and writing, wide range and scattered.
Our SSD is also written sequentially when it is used for the first time, but with the longer the use and the larger the capacity, its writes will become more scattered.
Therefore, testing the random read and write performance of 4K is helpful for us to understand the read and write performance of SSD and the stability of performance.
In daily applications, the writing of web page cache and the update of system files, including the loading and response of programs and games, are all closely related to the random 4K read and write performance. it can be said that the speed of 4K read and write determines the operation experience of the system.
So it is so important for 4K performace.
However, we mentioned above that the “4K performance” measured by, AS SSD Benchmark, is a random read-write performance with a single queue and a 64-queue depth.
However, the applications we use on a daily basis cannot only read and write randomly under a single queue depth, nor can they always reach 32 or 64 queues (this depth is often used on servers). In-home and small office environments, random reads and writes with a depth of 4-16 queues are generally dominant.
Therefore, the two random 4K performances (single queue and 64 queues) measured by ASS SSD are of little significance in ordinary home environments.
The 4K random read and write performance measured by AS SSD Benchmark is too one-sided.
In addition, the performance of SSD includes read-write performance and security performance.
The read-write performance of SSD includes continuous read-write performance and random read-write performance.
Therefore, it is one-sided and narrow to judge the read-write performance of an SSD by a random read-write performance, or even to judge the performance of an SSD.
Generally speaking, 4K performance is important, but the narrow sense of “4K performance” measured by AS SSD Benchmark is not very important and is for reference only.
Therefore, it is unreliable to see whether the performance of SSD is good or bad only by the so-called “4K performance” measured by AS SSD Benchmark.
When we come to this conclusion, some folks with only 4K theory say, “the random performance of the group RAID0,4K has not increased, so the read and write performance of the hard disk array has not increased, and the group RAID0 is useless.” there is also an answer to this question.