Gigabyte Z390 Designare Test Setup, Thunderbolt And PCMark Benchmarks
Test System Configuration Notes: When configuring our test system for this article, we first updated the BIOS and set the board to its default settings. Once we were sure all settings were at their stock values, we saved them, re-entered the BIOS and ensured the memory speed was set to XMP DDR4 3200MHz -- identical to all of the other boards featured here.
Our test rig is powered by an Intel Core i9-9900K Coffee Lake Octa-core CPU, Thermaltake Floe Ring 360mm AIO water cooler, 32GB of Corsair Vengeance DDR4 3200MHz CL16 memory, a 1600W Corsair AX1600i PSU, an NVIDIA GeForce RTX 2080 Ti GPU, and a Samsung PM961 256GB NVMe SSD (boot) and a Samsung 860 Evo 4TB SSD (Game).
Once the Windows 10 installation was complete, we fully updated the OS, and installed the drivers necessary for our components. We then installed all of our benchmarking software, performed a disk clean-up, cleared any old prefetch and temp data, and ran the tests. The system was restarted and allowed to hit an idle state when necessary before initiating any tests. All tests were run 3 times to ensure a standard variance was accounted for and results were averaged. Any outlier tests would be discarded and another run would be engaged to ensure continuity.
One of the major features of the Gigabyte Z390 Designare board is its dual Thunderbolt 3 ports. We wanted to test these to see exactly how well they perform with a full PCIe x4 storage device attached. For this test we disassembled our Patriot EVLVR Thunderbolt 3 SSD and replaced the PCIe x2 M.2 drive with another Samsung PM961 M.2 NVMe SSD. Since the Thunderbolt ports are routed via the PCH (DMI), we know that the PCIe x4 DMI link will be strained when pushing for full performance...
As a baseline this is how the main boot drive, a 256GB PM961 will perform through DMI via a M.2 slot on the motherboard.
Here we can see Thunderbolt port 1, and there is some performance lost as there is likely congestion passing test data in conjunction with the OS drive trying to talk through the same DMI space.
Here we see the same with the drive moved to port 2. It takes a performance hit, but is still super speedy when compared to a USB storage solution.
And here the proof is in the pudding. We ran the on-board M.2 test run at the same time as the Thunderbolt run, and you could watch the test passes fluctuate as they fought over available bandwidth.
The reason we tested this is to show why Gigabyte implemented the ability to run CPU lanes to the slots for add in card SSDs. That means data passes directly to the CPU and bypasses the DMI bottleneck, which we see here. Do keep in mind that any Z390 board you use that has Thunderbolt will experience the same results or similar, which is another reason why we are illustrating it here. This is normal and nothing is wrong with the system; it is simply a limitation of the interface when running through the crowded DMI. This is one of the reasons that many workstation builds are HEDT, as they have a lot more available PCIe lanes from the CPU, so you can run more devices with less chance of conflict.
That being said, the results above are still plenty speedy and your work will not grind to a halt. Though, if transferring large 4K/8K video footage from Thunderbolt to an M.2 device or something like that, you may get slower than expected speeds. All things considered, even with both drives running the test, we still saw speeds far exceeding USB 3.0 thumb drives by almost double in most cases.
The Essentials test - The Essentials group covers the common, everyday ways that people use a PC. The workloads include Web Browsing, Video Conferencing, and App Start-up time.
The Productivity test - The Productivity test group measures system performance with everyday office applications. This test group includes the Spreadsheets and Writing workloads.
The Digital Content Creation test - This test group's workload reflect the demands of working with digital content and media. The tests include Photo Editing, Video Editing, and Rendering and Visualization.
The Gigabyte Z390 Designare has a strong showing at stock, since the default BIOS settings and ICCMax guidelines are followed differently between the ASUS boards and Gigabyte. This results in decent gains as the Gigabyte board has sustained higher clock speeds due to not rolling back after the ICCMax turbo duration limit is met. The ASUS board more closely follows these guidelines and suffers a bit in comparison for following the defined spec here.
The overclocked results at 5.0GHz show the Gigabyte board losing the lead to the Maximus XI Extreme board. Because of the manual tuning here, any perceived performance benefits from the auto / stock board optimizations are eliminated.
Lets move on to a few more tests and see what else shakes out as we put the Z390 Designare through its paces.