1

I recently built a system for my daughter for her birthday using the following primary components:

  • AMD RYZEN 5 1500X 4-Core 3.5 GHz (3.7 GHz Turbo) Socket AM4 65W YD150XBBAEBOX Desktop Processor
  • ASRock AB350M-HDV AM4 AMD Promontory B350 SATA 6Gb/s USB 3.0 HDMI Micro ATX AMD Motherboard
  • Samsung 850 Evo 256GB SSD
  • SAPPHIRE Radeon RX 460 DirectX 12 100409-2GOC-2L 2GB 128-Bit GDDR5 PCI Express 3.0 CrossFireX Support Video Cards
  • EVGA 430 W1 100-W1-0430-KR 80+ WHITE 430W Power Supply
  • CORSAIR Vengeance LPX 8GB 288-Pin DDR4 SDRAM DDR4 2666 (PC4 21300) Desktop Memory Model CMK8GX4M1A2666C16

I ran Passmark's PerformanceTest 9 and all tests came back with relatively good results with the exception of my memory. All numbers were extremely low (in the bottom 2% in latency and bottom 20% in all other categories compared against the same memory model in PerformanceTest). I know this system isn't a super high performance machine, but I feel like I should see much better numbers versus my exact same memory module that are probably in much older systems, or that at least the latency number indicates a serious issue. What could cause this?

I installed the AMD All-In-One drivers that came with the MB, as well as the newest Radeon software and all the drivers it recommended. The MB reports the ram as running at the rated speed (2666). Is there a potential BIOS setting or driver install I missed, or does this just indicate a bad stick of RAM?

Improve this question
4
  • Memeory compatability is a known problem with Ryzen but if it's running at the stock frequency then it might just be the quality of said memory.
    – Ramhound
    Commented Jun 3, 2017 at 14:49
  • @Ramhound So maybe the reason is that the benchmarks of the same model were run on non-Ryzen systems, and that's why my numbers are so low in comparison? How would that be addressed in future updates? Through an AMD chipset driver?
    – Nicholas
    Commented Jun 3, 2017 at 22:51
  • Ryzen (Zen) with multiple CCX have funny performance issues when program runs on multiple CCX. You can try cheap trick of setting process affinity to single core to see if it makes difference. (Can't test in practice, happy Intel user here, but there is bunch of info on reddit and I see people testing CCX variations on youtube)
    – PTwr
    Commented Sep 18, 2017 at 11:16
  • Sounds like it's probably a consequence of using only one stick of RAM, whereas other systems may use 2 or 4. This article reports that they found using 2 DIMM's instead of 1 resulted in ~17.7% performance increase, whereas 4 DIMM's is expected to be even better. So if you're using just 1, then comparing against systems that use more, it makes sense that your configuration is testing as being slow relative to the others.
    – Nat
    Commented Nov 17, 2018 at 6:18

1 Answer 1

Reset to default
1

I was thinking about the same thing (Ryzen 1600), I haven't seen any info about it but I think I have a logic explaination.

If I run Aida memory tests the results are good against other "competing" systems with 2 memory channels, for example system based on Intel I7 7770k. If I look on the Passmark results however the results are grouped and compared to systems with the same memory module rather than CPU (or chipset). This means system with two memory channels will "compete" in the charts agains 4 channel systems, for example Threadripper or premium Intel platforms.

This will make the system come out pretty bad on the charts of Passmark. By looking at the charts you can also see that many benched clients are residing in the upper segment where you can suspect a 4 channel memory configuration has been used. This is also very logical beacuse the people that has paid high price for a quad memory configuration and a beefy CPU is likely more interested in doing benches on their new build beasts.

Edit: The Passmark charts can also be a bit confusing at first glance since the lower end of the segment don't start at zero but rather on the results of the lower end scores of a particular memory module. This will make the high end configurations look several times stronger than the lower ones. This is not the case if you study them a little bit more.

Improve this answer

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .