[H-GEN] I need help with a new Threadripper system.

[David Seikel]

On 2020-10-05 17:52:58, Walter Seikel wrote:
>    Some of you might remember me from long ago.  I'm hoping someone can
>    help
>    me diagnose which bit of hardware is broken on my new AMD Ryzen
>    Threadripper system.
>    TL;DR - It's likely that either the motherboard or the CPU on my
>    expensive new system is DOA, having a compatible working system to swap
>    bits with might help track down which.
>    I have been building computers since the early '80s, sometimes
>    professionally.  So it is unlikely that I broke anything.
>    This is what I ordered from Newegg -
>    CPU: AMD Threadripper 3990X 2.9 GHz 64-Core Processor
>    Motherboard: MSI Creator TRX40
>    RAM: G.Skill Trident Z Neo 256 GB (8 x 32 GB) DDR4-3600 CL18 Memory
>    Video: Sapphire Radeon RX 5600 XT BE 6 GB PULSE
>    Storage: Seagate FireCuda 520 2 TB M.2-2280 NVME SSD
>    PSU: be quiet! Dark Power Pro 12 1500W, 80 PLUS Titanium efficiency
>    CPU cooler: Noctua NH-U14S TR4-SP3 82.52 CFM CPU Cooler
>    Extra fan: Noctua A15 HS-PWM 140 mm black fan
>    Case: be quiet! Dark Base 700 ATX Mid Tower Case

As discussed in IRC, here's an update.  Hopefully using my new list
account instead of Gmail.

Just a reminder, in the last email I had tried a single stick of cheap
locally purchased RAM, so I could test things while I waited for the
proper RAM to arrive.

The proper RAM arrived, and it fired up first time after installing it. 
However, there was a bit of a problem though the BIOS said I have 8
sticks of 32 GB RAM, it said I only have 64 GB of RAM!

After a lot of juggling of RAM sticks, I got it to recognise 192 GB of
RAM, which is what it tells the OS it has.

There's a message that flashes onto the screen for a fraction of a second
before it shows the first BIOS booting screen, using my phone I made
a video so I could actually read it, it says -

"Memory PMU Training error at Socket 0 Channel 2 DIMM 0 & DIMM 1"

Which doesn't match anything written on the motherboard or it's manual,
which lists the sockets as A1, A2, B1, B2, C1, C2, D1, and D2.  The
motherboard is also fussy about where to put the sticks for the supported
2 / 4 / 8 stick configurations.

I had tried each stick by itself, and then all the combinations of 2
sticks, then some combinations of 4, but then my fingers where getting
very sore.  The heat spreader on top of these sticks is rather sharp.

Then I just said to hell with it, time to play with my new toy. 
Installed operating system, did lots of speed tests, transferred
everything across from my old desktop, actually use the thing.

I still want to actually be able to ue all the RAM I paid for.

After some discussions with MSI, the makers of the motherboard, and
having already tried everything they suggested, they told me to contact
the seller, Newegg, which I did.  Newegg's 30 day return policy starts
from the Invoice date.  They don't have any local presence to quickly
test stuff.  So I'd have to send things back to USA, before Thursday.

The reason the RAM took so long to get to me is it started in Hong Kong,
flew to USA, then to east coast USA, then to west coast USA, then Sydney
(actually passing me), then finally to Brisbane, where it sat for a couple
of days.  UPS told me three times that they had some sort of hardware
malfunciton that would delay it's arrival.  Not sure if that happened
three times, or they just kept reminding me why it's late.

I'm really scared that I'll spend another month or two sending parts all
the way around the world several times, get well beyond Neweggs
replacement date, spend the rest of my savings, and end up with a useless
bunch of very expensive parts collecting dust.

At this point I'm not sure if it's the CPU, motherboard, or RAM that is
at fault.  What would REALLY help is if I can find some one local with
enough compatible parts that we can try swapping things to determine
which part the fault actually is.  I'm almost certain it's the
motherboard, but I could be wrong.  After that, I'd know which one te get
Newegg to replace early next week.

After the IRC discussion I called Umart to see if they could help.  Umart
have a diagnostic service, but wont do it on parts they didn't supply. 
They referred me to "My Computer Support" in Milton, who answered the
phone as "Nerds to go".  They don't have any Threadripper parts for
swapsies, but he said he'll pass my request onto their boss, see if they
can find someone to help me.  The boss might not be able to help.

Can any one help?

-- 
A big old stinking pile of genius that no one wants
coz there are too many silver coated monkeys in the world.