9800X3D, XPower Zenith 6000mhz cl30

helmetboii · 3 Sep 2025 at 12:35

Specs
CPU: Ryzen 9800X3D
MOBO: MSI X670E GAMING PLUS WIFI
RAM: 2X16G X Power Zenith RGB, 6000 MT/s, CL30-38-38-96 1.35V (EXPO), (SP032GXLWU60AFDF)
GPU: MSI 3080 SUPRIM X 10G

To begin, I have recently upgraded to AM5, I purchased my ram from amazon, advertised as the kit above. SP032... which on X Powers website is advertised as a 2x16GB. Upon upgrading I ran into issues so I just wiped my drives clean and reinstalled windows from a USB Drive. This got rid of the blue screens that were occurring. However, running EXPO on this kit has come up with issues, with sometimes the memory failing instantly on memtest5 and even cinebench not running. I reset CMOS and booted into windows with JEDEC timings, 4800MT/s and the issues were gone.

I thought that one of the ram sticks could be damaged, however first I reseated the ram and tried again with EXPO but was getting similar errors so I tried 1 stick at a time, both ran at EXPO, with no memory issues, Cinebench ran etc., which leads me to believe the sticks are fine. I put both sticks back in and they began working fine, it did take a while to train the memory. Everything was fine, I decided to try some undervolting using PBO, turned on motherboard limits, worked my way up to -25 CO, +200Mhz. Booted into windows fine, ran memtest and it starts throwing out errors. What I'm trying to understand is if my ram is unstable, or my CPU is unstable and if the memory controller isn't functioning as it should. I'd also like to know if change PBO values can throw out memory test errors, and if my problems indicate any specific issues that someone may have encountered before. It may well be that my CPU isn't a crazy good bin and that the negative CO just throws out every test as Cinebench also crashed again at these values.

I then tried to find what dies my RAM has using the serial code on my sticks, just to find out that the serial codes on the physical sticks dont match that which I ordered of amazon,
Amazon Page: SP032GXLWU60AFDF
Physical Stick: SP016GXLWU60AFSF

I checked some information on X Powers website and it shows that the serial code for my kit, is the one shown on amazon, and the physical sticks serial code is for a singular stick of 16GB, however both of my sticks came together in a singular plastic cover. This makes me wonder if I actually got a kit that was tested together or if I got 2 separately binned ram sticks that barely function together at EXPO.

I'd appreciate if anyone can advise on any checks I should do to find the root cause and whether the serial codes are even an issue or not. (I sent an email to X Power about the serial codes, waiting to hear back in a couple days)

Thanks

JollyJamma · 3 Sep 2025 at 16:50

Might want to read the forum rules there.

pastymuncher · 3 Sep 2025 at 18:58

Hi and welcome to the forums.

First thing to do if you haven't already done it is update the motherboard to the latest bios as there are always updates for memory compatibility.

Individual sticks will have a different code to the actual kits box as they are 16Gb sticks. The retailers site will be listing the kit part number which is why it has 32 in it.

Tetras · 3 Sep 2025 at 19:12

helmetboii said:
I'd appreciate if anyone can advise on any checks I should do to find the root cause and whether the serial codes are even an issue or not. (I sent an email to X Power about the serial codes, waiting to hear back in a couple days)

Difficult to isolate the problem really, you just have to try one at a time.

From what I can gather, the sticks work totally fine at 4800, which would suggest they're probably not faulty in the typical sense that RAM is, because faulty memory often just chucks out BSODs and errors.

You did at one point have these sticks running stably with EXPO enabled, right? Was it actually stable, or just stable in use?

Undervolting can reduce the general stability of the PC and it can impact the memory controller, but as for: how much should you expect memtest errors due to that, I don't know. From what I'm aware, memory errors due to instability are not that common, whereas memory errors due to faulty memory are expected to be a much higher frequency. WHEA errors (event viewer) are also evident with CPU/memory controller instability (including due to PBO), whereas not with faulty memory.

It is possible that your memory can't run at the rated speed, or the CPU can't operate that high, though most AM5 CPUs can hit those speeds fairly comfortably.

Finners · 3 Sep 2025 at 20:26

Definitely take off the curve offset and boost. You are trying to over clock and undervolt at the same time, very high possibility that isn't stable.

To many variables with that on to try and narrow down what is wrong.

I would load up expo timings and take a screen shot of Zentimings and post it here and we can look and see if anything looks odd.

Have you definitely got the RAM in the primary slots? Usually slots 2+4 these days

helmetboii · 3 Sep 2025 at 21:17

pastymuncher said:
Hi and welcome to the forums.

First thing to do if you haven't already done it is update the motherboard to the latest bios as there are always updates for memory compatibility.

Individual sticks will have a different code to the actual kits box as they are 16Gb sticks. The retailers site will be listing the kit part number which is why it has 32 in it.

Thanks for the response, first thing I tried, bios is officially up to date.

I see, thankyou for the clarification

helmetboii · 3 Sep 2025 at 21:20

Tetras said:
Difficult to isolate the problem really, you just have to try one at a time.

From what I can gather, the sticks work totally fine at 4800, which would suggest they're probably not faulty in the typical sense that RAM is, because faulty memory often just chucks out BSODs and errors.

You did at one point have these sticks running stably with EXPO enabled, right? Was it actually stable, or just stable in use?

Undervolting can reduce the general stability of the PC and it can impact the memory controller, but as for: how much should you expect memtest errors due to that, I don't know. From what I'm aware, memory errors due to instability are not that common, whereas memory errors due to faulty memory are expected to be a much higher frequency. WHEA errors (event viewer) are also evident with CPU/memory controller instability (including due to PBO), whereas not with faulty memory.

It is possible that your memory can't run at the rated speed, or the CPU can't operate that high, though most AM5 CPUs can hit those speeds fairly comfortably.

I didn't do extensive testing regarding just the EXPO profile, I ran memtest5, I also booted memtest86 on a usb to check for any issues but everything was clear. I tried this without any PBO changes, which leads me to believe the RAM is okay running at EXPO, but reading your point about memory errors due to instability, that does make sense so I'm not sure what to think about it really.

Will have a look at event viewer now. I also did a windows memory diagnostic test, that came up clear although I have no clue about the reliability of that diagnostic test.

helmetboii · 3 Sep 2025 at 21:28

Finners said:
Definitely take off the curve offset and boost. You are trying to over clock and undervolt at the same time, very high possibility that isn't stable.

To many variables with that on to try and narrow down what is wrong.

I would load up expo timings and take a screen shot of Zentimings and post it here and we can look and see if anything looks odd.

Have you definitely got the RAM in the primary slots? Usually slots 2+4 these days

I'll start with that.

RAM is definitely in A2, B2 on my board and I re-seated the sticks a couple times

https://imgur.com/hkmrG7N

Hope this link works.

Thanks for the help

sativa · 3 Sep 2025 at 22:00

EXPO is overclocking and not guaranteed.. Could be the CPU memory controller failing (albeit, unlikely)

Finners · 4 Sep 2025 at 08:43

helmetboii said:
I'll start with that.

RAM is definitely in A2, B2 on my board and I re-seated the sticks a couple times

https://imgur.com/hkmrG7N

Hope this link works.

Thanks for the help

The only thing that stands out to me is the VDDP voltage is on the low end. I'm sure people can run 6000Mhz at 0.95v but I'm surprised it's as low if that's the mother board set at auto

Try it at 1.05v and see if things improve, dont go higher than 1.1v as it should be 0.1v below the soc voltage.

JollyJamma · 4 Sep 2025 at 09:44

Finners said:
The only thing that stands out to me is the VDDP voltage is on the low end. I'm sure people can run 6000Mhz at 0.95v but I'm surprised it's as low if that's the mother board set at auto

Try it at 1.05v and see if things improve, dont go higher than 1.1v as it should be 0.1v below the soc voltage.

I agree with this and would suggest manually setting voltages, SOC voltage is super important.

helmetboii · 4 Sep 2025 at 22:05

Finners said:
The only thing that stands out to me is the VDDP voltage is on the low end. I'm sure people can run 6000Mhz at 0.95v but I'm surprised it's as low if that's the mother board set at auto

Try it at 1.05v and see if things improve, dont go higher than 1.1v as it should be 0.1v below the soc voltage.

I will try this and come back here, but in the meanwhile, everything is currently stable with stock configuration. Everything is also stable with +200 Boost Clock, this is with no undervolt (including y-cruncher). The main test that didn't pass, or didn't even run to begin with when applying any undervolt below -10 CO was Y-Cruncher, everything else went smooth including a 10 hour memtest5 config from anta, cinebench

I will re-test with stock config and +200Mhz Fmax, with your suggestions once I'm home later.

XPower got back to me, saying its likely not the RAM, which makes sense since its stable in its own right with stock config, however they suggested just in case, to increase voltage to 1.4V. I'll try this separately.

With all tests above considered, the cpu is stable at stock settings, its also "stable" with +200 Fmax and EXPO enabled, I used OCCT, Prime95, Y-Cruncher.

I had to reset CMOS at some point, this was my own fault and forgot to change ram settings, but anyways I tested with a negative CO around -20, but still issues. Which just leads me to believe its not the best binned 9800x3d. No WHEA errors so not sure what to make of it all.

Thanks for all the input, I may consider a replacement through overclockers if its a quick turn around and try my luck again.