One dimm then the rest seems the most sane. They’ll make it less quirky aft er bios 0404 these are just teething issues. Annoying teething issues.
9 Likes
Just arrived via UPS, I finally have the last component I need for my own build… the RAM!
I appreciate what has been discovered in this thread so far, and I’ll share my own experience as I progress ahead with my own ASUS WRX90 / G. Skill Zeta R5 Neo setup.
2 Likes
Excellent - also curious if you can get all 8 sticks running above base clock. I’m seriously doubting they would run at 6400 as rated… maybe 5600?
1 Like
I am stable at 6000 but it might be able to do 6400 at 2:1
2 Likes
Well, running with all 8 sticks!
that 4800MHz speed was a little disappointing to see though.
Apparently when you try to increase the ram speed, you still get the message that your threadripper cpu will be marked for overclocking? Even if it’s just for the ram?
3 Likes
Awesome – did you start with 1 stick, let it POST, and then add the remaining 7? Also, didn’t realize that re: overclocking for the CPU to use EXPO.
Does it warn about the overclocking fuse even if you try setting it to 5200, which I think should still be stock?
4800 is jedec baseline that is totally normal. You can run 5200 or possibly beyond with the expo/xmp profile.
1 Like
Yeah, I think I got the warning when I just happened to pick 6400 in the list, to see what would happen. But I never tried to boot with those settings, nor have I tried any other speeds yet.
I’ve decided to just get the whole system assembled, up and running, before I start fine tuning those settings.
It’s good to know that lower speeds may not give that overclock warning though, I will probably try that tomorrow.
2 Likes
Yes, I did start with just one memory module, and it booted fine. Then I shut it down, added the rest of the ram, and did it again. No issues. I don’t think that initial boot took much longer with 256GB than it did 32GB either.
Though, it’s not the speediest machine on a restart, but it’s not too annoying either.
3 Likes
This looks like a rush to release a motherboard that is not ready for prime time. Today I received the motherboard with Threadripper Pro 7985WX. It will only post with four 64GB modules. If I install 6 or 8 modules, the post code stucks at P0. The board comes with 04.04 BIOS pre-installed.
Take a look at two BMC screen shot. The 1st photo shows the CPU model mismatch. It reports 7965WX (incorrect) while the motherboard BIOS reports 7985WX (correct)
The 2nd photo shows only one 32GB DIMM module in slot DIMM_7. But the board actually has four 64GB modules in slots DIMM_0,1,2,3. The motherboard BIOS does report the correct size (256GB).
This looks like a rush job by ASUS. 
2 Likes
That’s post code 0d.
Leave the dimm out of slot h and try again. I bet eventually you get 7/8 dimms. Clear cmos and 10 min wait is your friend. This was a similar bug on tr5k too.
I can replicate you issue, post code 0d, on another board so it’s likely not your board.
The ipmi reporting is what cpu the board was tested with from factory. It won’t update till you enter bios then save the first time.
Ipmi doesn’t inventory the system until the dust settles into a mostly working system, counter to most peoples intuition.
I agree it’s half baked.
2 Likes
Thanks Wendell. I reached the same conclusion as you regarding the inventory information. After the post, I reset BMC to factory defaults. Eventually it reports the correct CPU and memory information.
I have used countless Supermicro and ASUS boards with onboard IPMI. This is the 1st time I see such report from the initial boot. You would think the BMC would just report what is currently configured.
I have the board running with 512GB memory now using eight Samsung 64GB DDR5-5600mhz modules. The BIOS set the memory bus speed to 5200mhz (auto mode).
3 Likes
What did you do to get all 8 sticks picked up?
I didn’t do anything special. Just started with 4 modules. Then added 4 more later.
Alright, here’s my update after ordering a replacement ASUS WRX90 and a Kingston 256GB kit (just in case the G.Skill kit was the issue):
-
Installed new mobo, CPU/heatsink, 2xNVMe drives, 1 piece of memory from the 32GB G.Skill kit, and 1 4090 FE (no other PCIe cards). Result: Same thing - Code 92!
-
Cleared CMOS, tried again. Result: Same thing - Code 92!
-
Tried 1 piece from the Kingston kit (on the QVL). Result: Same thing - Code 92!
-
Connected both supplemental PCIe_8p power connectors and tried again. Result: Same thing - Code 92!
-
It was at this point I decided to try the only thing I had not tried before. I pulled the 4090 FE and replaced it with an ARC A750 I had lying around. Result: Immediate POST (with 1 G.Skill DIMM).
-
I put the 4090 FE back in and tried again. Result: Immediate POST!
-
I added all 7 of the remaining DIMMs from the G.Skill kit. Result: After a couple minutes of training, POST!
-
I increased the clock to 6000 MT/s for the 8 G.Skill DIMMs (kit is rated at 6400, but I only went to 6000). Result: Immediate POST!
So far, everything is stable.
Takeaways:
- The original board was likely NOT defective.
- The GPU was the issue, but I don’t know why. And, to clarify, I am NOT using either supplemental PCIe power connector at this point – should be no need with a single GPU, but more on that in a sec.
- The system appears to be stable and has been running for the past few hours without issue.
Remaining Issues:
- I have not tweaked anything in the BIOS with the exception of the EXPO clock to 6,000 MT/s. Curious to see what others find to be the most impactful changes.
- Sleep is not an option in Windows – it’s as if that power state is unavailable. Perhaps something in the BIOS can fix that.
- I added two (2) additional PCIe cards – an Elgato 4K capture card and a Sound Blaster Both consume VERY little power (well under the 75W slot limit, obviously). However, the BIOS still complains that because I have two (2) or more PCIe cards, I need a supplemental PCIe power input connected – which seems dumb. Not sure if there is a way to ignore the warning in the BIOS, but worst case I’ll order a matching cable from CableMod.
I hope this helps someone else!
Richard
10 Likes
it’s great to hear you got your system up and running, helps me feel better sitting on the Asus board for a couple months while I get everything together for my build.
a thought on the no boot with the 4090, maybe the power draw of the 4090 on startup without the supplemental pcie power plugged in tripped some kind of fault safety built into the board that isn’t explicitly explained, and possibly only triggers on the first boot for some reason. obviously, that’s just a guess but figured I’d share my thoughts on it.
anyway thank you again Richard for the update.
2 Likes
Sure thing – thought about the power draw part for the 4090, but as it is now, the 4090 is working and there are no supplemental PCIe power connectors plugged in – which should be fine for a single GPU, in theory. But, who knows. I really can’t explain why putting in a different GPU for a single POST and then putting back the original GPU magically fixed it. Definitely some teething issues with this platform.
agree 100% on the teething issues, edited my above post to include the idea that it could be a first boot issue only with power draw and once it starts in a none fault state once it then doesn’t set the fault off a second time. anyway, just a thought, doesn’t matter much since it works now!
1 Like