Solid state drive benchmarking tests SSD performance under a variety of workloads. SSD owners that want to monitor their drive can use benchmarks to observe, read, and write rates and other performance metrics.
For enterprises such as Samsung and Western Digital, which sell SSDs, benchmark tests reveal the speeds at which their SSDs read and write data under different conditions. Those read and write speeds can then be marketed to potential customers who are searching for persistent storage devices that provide quick access to data. An ideal benchmark test gives customers an accurate picture of how quickly they’ll be able to access that data using the SSD.
Different types of benchmarking test different things:
- Synthetic benchmarks test the performance of a particular drive feature through an artificial, computer-created environment.
- Application-based benchmarks test one piece of software at a time.
- Trace-based benchmarks test drive use during drive runtime.
Benchmarking can sometimes be inaccurate or give misleading results, especially if performed too quickly. All SSD benchmark preparation should involve appropriate preconditioning — preparing the drive to be steady. Proper testing can take hours. Tests for enterprise SSDs should give consistent information about how the drives will perform, rather than misleading consumers with inaccurate speeds.
SSD benchmarks test both sequential and random read and write speeds. A thorough benchmark result should be replicable: someone should be able to reliably repeat the test environment. It should not be a one-time test that can never be recreated; that suggests inaccuracy and uncertainty about how the test actually occurred.
- Common SSD Terminology
- Top SSD Benchmark Tools
Common SSD Terminology
If you aren’t familiar with the concept of benchmarking or the details of how SSDs perform, the following terms are very commonly used to describe SSD specifications.
- Throughput: simply a word that describes data transfer rate for drives. Throughput is measured in bytes per second — in SSDs, more commonly megabytes or gigabytes.
- IOPS: SSD input/output operations per second. High input and output are ideal. Input and output refer to writing and reading data to and from a drive.
- Latency: the length of processing time that an input or output requires.
- 4K: the number of bytes in a sector on a drive; this amount differs from a 512-byte sector, which is more common in hard drives. 4K is often referred to as 4K blocks.
- 64 thread: a queue depth of 64. Queue depth measures how many I/O requests can wait at one time on an SSD.
Top SSD Benchmark Tools
The following products test SSDs under a variety of conditions. Some also test hard drives and other data storage devices.
ATTO Disk Benchmark tests HDDs, SSDs, and RAID arrays; it can also benchmark controllers, including RAID controllers. ATTO tests maximum transfer speed, as well as a variety of transfer sizes and test lengths.
ATTO supports overlapped I/O requests and a variety of queue depths. Users can:
- Choose to run a test continuously. If the software detects any errors during the test, it will halt the process.
- Make an I/O comparison between data from a test file and data written per block.
- Add a description of the test
ATTO’s Windows version is more commonly known, but ATTO also offers a macOS version with features such as snapshots and custom transfer sizes.
AS SSD, for Windows 10, uses uncompressed data for its benchmark operations. AS SSD offers three test options for benchmarking different file sizes:
- Read and write speeds for 1 GB of data
- Read and write speeds for 4K blocks
- 4K testing for 64 threads, which places the read and write operations throughout the 64 threads
AS SSD allows users to put the software on a USB drive, which they can then use with a variety of SSDs. Users can also test the access time of their SSD and observe their drive’s performance while making copies depending on file size and amount.
CrystalDiskMark offers SSD benchmarking for a variety of drives and scenarios. It can also help users decide when to optimize their SSD if needed. CrystalDisk software solutions have a very simple interface but are popular among hard drive and solid state drive owners.
Users can measure sequential read/write speeds and random read/write speeds. CrystalDiskMark supports a queue depth of up to 32.
Iometer, initially developed by Intel, is now an open source SSD benchmarking software. Because it’s open source and flexible, users can configure the software if they wish. Iometer tests and reports multiple SSD performance features, including:
- Input and output
- Mixed I/O
- Random read and write speeds
- Server load
Iometer can create a variety of tests for I/O workloads. It’s often used by professionals who need to regularly test their SSDs. Iometer is a good benchmark tool for users that want to apply a variety of stressful workloads to test a drive.
PCMark10 gives drives an overall score across all tests. PCMark drive benchmark options include:
- Drive performance consistency test for lengthier, heavier workloads and reporting
- Full system drive benchmark, which performs “traces”, simulations of normal computing tasks or common applications
- Quick system drive benchmark for shorter tests, smaller drives, and lighter trace loads
- Data drive benchmark, which tests file storage drives, including a variety of external storage options
Samsung’s Magician tool tests an arsenal of Samsung SATA and NVMe SSDs. Compared to many other benchmark tools, the Magician has a modern interface with a variety of SSD management features — benchmarking is just one. Users can:
- Test sequential read/write speeds
- Test random read and write IOPS
- View the previous history of the drive’s read and write operations
- Visualize drive performance with colorful graphs.