Question Benchmark shows computer not performing to specs, i9-9980XE results are terrible.

Jul 15, 2019

I am very inexperienced, but was in the market for a new computer and decided to have a go at building my own. After an extremely long weekend, I have got everything up and running. Thank you of course to all the forum contributors on this site for your help along the way. However, running some benchmark tests shows some disappointing results.

My system is as follows:

Motherboard: Asus Prime X299 Deluxe II
CPU: Intel i9-9980XE
Memory: Corsair Vengeance 4x16GB DDR4 3600MHz
Storage 1: Intel Optane 905P NVMe PCIe 960GB
Storage 2: Samsung 860 Pro 2TB
Graphics: AMD Radeon Pro WX 9100
Cooler: Corsair Hydro Series H100i Pro 120mm Liquid Cooling System

The Novabench results and a link to UserBenchmark results are below.

  1. I'm concerned with the Read / Write speed of my Disk, which is supposed to be up to 2600 / 2000.
  2. Shouldn't the RAM speed be a fair bit higher? I loaded the XMP profile in BIOS to utilize the 3600MHz, and it seemed to take, but translating that to the benchmark MB /s it doesn't appear so.
  3. Novabench picks up Trigger 6 External Graphics. Does this mean it's not testing my graphics card?
  4. UserBenchmark shows my CPU and both drives performing way below expectations. I have not overclocked the CPU, so not sure if that has an effect on the benchmark expectations. I just want to get to or near the performance I should be at for my hardware.
What now? Thanks

Verified Novabench Score: 3853
Tested on Jul 15, 2019 at 16:06
CPU Score

Show details
RAM Score - 64GB DDR4
RAM Speed: 20883 MB/s
GPU Score
Direct3D 11: 102 FPS
OpenCL: 13015 GFLOPS
Disk Score
Write Speed: 2007 MB/s
Read Speed: 1459 MB/s
I'd go check what RAM speeds are supported by the mainboard when all 4 slots are populated (I'd be surprised if 3600 MHz was in the supported list with 4 slots populated, no matter what the RAM kit claims, and you might want to check at 2933 and 3200 MHz first, and confirm RAM bandwidth scaling...)

You might want to check Intel's XTU and/or HWMonitor to see if any thermal throttling is occurring under load...(Intel's own XTU has a built in stress test that will run a CPU 85-90% of it's max. (not sure what mainboard VRM/power limits might be imposed , if any with a stock 9980XE, but, a sharp decrease in clock speeds under load when/if temps are under control might imply some VRM limits and/or throttling...

CPU-Z single/multithread bench should quickly tell you if something with processing/threads is out of line...

No idea on the Trigger6/graphics reference...
With all of the Meltdown,Spectre,MDS ETC................... patches applied the Hyperthreading performance of Intel processors has dropped quite a bit in some workloads.
Your numbers will never match those run before the patches were applied.
Check your memory speed and CAS to make sure they are correct for your processor and motherboard(regardless of speed purchased).
Your scores should end up closer to the bottom of published scores.
You did not state power supply ,so that could also be your problem.
Jul 15, 2019
Thanks for the responses. I am truly a novice so please forgive me if my follow ups are simplistic.

From the spec sheet for the motherboard:

8 x DIMM, Max. 128GB, DDR4 4266(O.C.)/4200(O.C.)/4133(O.C.)/4000(O.C.)/3600(O.C.)/2666/2400/2133 MHz Non-ECC, Un-buffered Memory
Supports Intel® Extreme Memory Profile (XMP)

So does that speed possibly not apply if I have more than one stick slotted?

Intel stats the Memory type for that chip is DDR4-2666, but they also have XMP profile to allow me to change it, and some reading online suggests that the chip will keep up if the motherboard does, which based on the above it should ?

I ran CPU-Z after a restart, but was not sure how to interpret the results. Comparing the specs to benchmark results I found to be a little more straightforward.

I will check out XTU on next reboot, although monitoring the temp display on the motherboard suggests that the temp has not been getting very high. Perhaps related to this, I only plugged in one of the 8 pin connectors to the CPU. Should I plug the other one in?

CPU-Z results were 424 (Single) / 9222.6 (Multi) vs 543 / 10,082 Benchmarks on the site. This test was not done directly after a fresh reboot though, so perhaps would be improved marginally?

The power supply is an EVGA Supernova 750 G3, 80 Plus Gold 750W.

The most bothersome thing about these results is the Intel Optane 905P NVMe PCIe 960GB read / write speeds. Why is this falling so far short ?
The power supply is fine. But definitely plug in the other 8 pin EPS connector for the cpu. Without it the cpu will not boost to full clocks, it can not get enough power.
Memory is another matter.
Generally the more dims or ranks you install the slower the memory speed becomes. Memory has many more timings, than the first 4 stated on the package . So you may or may not receive the full speed unless it is listed on the motherboard memory QVL. Also check the motherboard manual for dim placement , it can make a huge difference in memory performance.
If this is a fresh install of windows 10 give it a couple days to settle in and finish all updates and moving things around on the disk.
Indexing , driver updates, OS updates all take away from system performance.
Make sure you have all of the motherboard drivers installed, plus Intel's Storage management software.
Jul 15, 2019
I plugged the other connector in, and ran another test on each. Performance was near identical.

Ran XTU benchmark and a brief stress test and it read "No" for Thermal Throttling as well as Power Limit Throttling.

As far as the memory goes, I will have to learn some more about the topic. Everything I understand and interpret from Asus about the board says it should support that memory speed. I did ensure the memory was slotted appropriately. It is a fresh install of Windows so I'll give it a couple days.