There are many things to consider when measuring performance, here are some relevant pieces of information we can provide based on your description:
- CrystalDiskMark is not a tool we officially use to measure the Intel® SSD 750 Series performance, however, you will get accurate results as long as you have Crystaldiskmark 4.0.3 or higher. You will need to change some options to match the tests specified in page 8 of the Intel® SSD 750 Series Product Specification. The configuration options are in the menus, under “settings” and Queues and Threads. You can refer to the specification for details about each test, but basically sequential uses 128KB transfer size and random tests use 4 KB. Maximum performance requires multiple threads (4+) and Queue Depth 32.
- Reviewing the Asus* Sabertooth Z97 Mark 2 specifications, it has 2 x PCIe 3.0/2.0 x16 (Single at x16, dual at x8/x8) and 1 x PCIe 2.0 x16 (max at x4 mode, black). Please make sure the drive is connected to one of the PCIe 3.0 slots.
- Contact the Computer Manufacturer Support and obtain the most recent BIOS for your Motherboard.
- Make sure to Download Intel® Solid-State Drive Data Center Family for NVMe Drivers, version 18.104.22.1687.
If the issue persists, please check the following thread, as it explains many performance implications and specific information regarding CrystalDiskMark: Intel 750 PCIe SSD 1.2TB Slow Benchmarks on CrystalDiskMark??
If you go to File -> Save, the tool will create a file that includes the results and settings used. You can attach this file to the thread using the Advanced Editor.
Thx for the suggestions. It's definitely in a 3.0 slot. I think I need to go through Intel's procedure for performance testing next.
Intel alumni (1980 - 1985)
I ran the 750 through all the standard IOMeter tests. The results are great: 128K Seq Read: 2360MB/sec, 128 Seq Write: 970MB/sec, 4K Random Read: 452K IOPS, 4K Random Write: 202K IOPS (only slightly low - it varied quite a bit as the test ran). I found out I was using an older version of Crystal Disk Mark (3.0.3) that couldn't really properly evaluate NVMe drives. Upon trying a later version (5.1.0) the numbers improved substantially.
Thanks again for the suggestions!
We are glad to know you are now getting better results in your benchmark tests.
I would like to remind you that using additional Data Treads (4 or 8) improves performance results, specially for Random data. The "T1" in your test indicates it is using a single thread, this is the default in CrystalDiskMark, however, it does not take full advantage of your drive's capabilities.