Test System
| Test System Specs | ||||
| Mainboard | Gigabyte X670E AORUS Master Bios Version 122 |
|||
| CPU | AMD Ryzen 9 7950X3D | |||
| GPU | Asus Rog Strix Gaming OC RTX 4090 | |||
| NVMe | XPG GAMMIX S50 Lite 1TB | |||
| RAM | XPG Lancer DDR5 (2 x 16GB) 6000MHz | |||
| Power Supply | Seasonic Vertex 1200W (Cybenetics Platinum) |
|||
| CPU Cooler | NH-D15S chromax.black | |||
| Case | DimasTech Bench | |||
| Ambient Temperature | 22°C ±3°C | |||
| Drivers | NVIDIA: 531.18 | |||
I use a powerful test system to have zero bottlenecks.
Real-Life Testing
I use a custom-made application named Pegasus to simulate ten different real-life data transfer tests and measure the average MB/s speed for each. The disk cache is flushed between tests, and each test runs multiple times, with reboots between each cycle. I use the average as the final result.
Synthetic Testing
Besides my benchmark, I use Crystal Disk Mark, AS SSD, and HD Tune Pro.
I also use DiskSpd by Microsoft, a highly flexible storage test tool capable of accurately simulating different workloads. I wrote two advanced scripts, one that simulates an On-line Transaction Processing (OLTP) system and another that simulates an On-line Analytical Processing (OLAP) system. The OLTP scenario consists of many short transactions, with IOPS (Input/Output Operations Per Second) playing a pivotal role. The number of transactions is low in my OLAP scenario, but the queries can be complex. Response times are crucial for an OLAP system, and the maximum throughput speed is reached in this scenario because the block size is quite large.
