ASUS KRPA-U16 not working with EPYC 7D13 (100-000000322)

RolloZ170

Well-Known Member
Apr 24, 2016
1,567
404
83
54
I have tried with different memory sticks (that work fine when the 7742 CPU is installed).
the 7D12 don't support hynix and no 2133mhz RAM.
maybe the 4 channels only applies to the 7D13 too.
but my chineese CPU supplier sells 7D13 too and says this CPU currently don't work on standard mainboards.
 

ExecutableFix

Active Member
Nov 25, 2019
123
60
28
So, amazingly, having access to actual unimpeachable and guaranteed correct documentation, I know what I'm interpreting.
Have you actually used a 7D12/7D13 before tho? Most likely not. I have had the exact same problems with my 7D12 at first and got it for cheap because "it wasn't working". Guess what? It did work just fine.

It's very simple: AMD does not create custom steppings or microcode for these SKUs, so all of them (yes, all of them) will work in regular EPYC boards as well (granted you have a board that supports Milan). You just have to find the right configuration because something about these SKUs is different from a normal variant (PCI-e lanes, memory channels etc.)

I'm quite busy at the moment, but I'll try to give you a rundown of things I've done to get my 7D12 to work in 2 weeks or so.
 

ExecutableFix

Active Member
Nov 25, 2019
123
60
28
the 7D12 don't support hynix and no 2133mhz RAM.
maybe the 4 channels only applies to the 7D13 too.
but my chineese CPU supplier sells 7D13 too and says this CPU currently don't work on standard mainboards.
I'm pretty positive the memory channel thing is exactly the problem here. ASUS doesn't make it any easier because of the 2 DIMM per channel situation
 

RolloZ170

Well-Known Member
Apr 24, 2016
1,567
404
83
54
Can your Chinese CPU supplier suggest a mainboard that can be purchased that will work?
asked the same: but still waiting for reply.
in your first message you wrote the 7D13 does not POST, later you gave some POST codes.....what have you changed ?
if your 7D13 is NOT vendor locked ther is a chance to get it to work, but as ExecutableFix wrote the ASUS is not the best choice for finding the issue.
if you have HYNIX memory buy samsung or micron. if you have still 2133Mhz RAM buy RAM with higher speed. then check the primary mem.slots one by one.
 

rootwyrm

Member
Mar 25, 2017
60
68
18
www.rootwyrm.com
Have you actually used a 7D12/7D13 before tho? Most likely not. I have had the exact same problems with my 7D12 at first and got it for cheap because "it wasn't working". Guess what? It did work just fine.

It's very simple: AMD does not create custom steppings or microcode for these SKUs, so all of them (yes, all of them) will work in regular EPYC boards as well (granted you have a board that supports Milan). You just have to find the right configuration because something about these SKUs is different from a normal variant (PCI-e lanes, memory channels etc.)

I'm quite busy at the moment, but I'll try to give you a rundown of things I've done to get my 7D12 to work in 2 weeks or so.
Perhaps you shouldn't spout pure guesswork and speculation contrary to known public information. There are two known variants, 7D13 and 7V13. Neither are 'OEM' parts. They are contract customer parts. Unless your name is Jason Zander, Mark Papermaster or Dr. Lisa ****ing Su, you know absolutely zero about what AMD can and does offer contract customers. Whereas I do know for a fact just from public information that 7V13's have some level of unique microcode.
Here, I'll give you a free hint.
And since you aren't the CEO or even an employee of AMD, you most definitely have no right much less the least capability to speak for them as to what they do and do not do.

Not to mention the complete absurdity of your delusional insistence that this is a Dell part or even a standard OEM part. If this is a Dell part, then please do explain why not a single Dell system or Dell customer or any customer has found a single one of these "Dell OEM" processors when they've been shipping a full year. Go on. I'm waiting to hear how out of more than 20 million systems shipped in just the first 6 months of 2021, not a single one of them received this part or ended up on eBay or some other reseller. Or anywhere at all within Dell's ordering system.
Seriously. Explain how it is that out of who knows how many hundreds of thousands of Dell EPYC servers that have shipped, not a single one in the entire world has a "Dell 7D13."

RolloZ170 said:
but my chineese CPU supplier sells 7D13 too and says this CPU currently don't work on standard mainboards.
Yeah, and I'd bet they aren't getting them from Quanta's loading docks either. I'd bet money these are coming from whoever's building Project Olympus boxes, which has been very hard to track down. The only thing known for sure is that the production parts use really, really brown FR4. Like, baby poop brown, not retro brown.
 

RolloZ170

Well-Known Member
Apr 24, 2016
1,567
404
83
54
Not to mention the complete absurdity of your delusional insistence that this is a Dell part or even a standard OEM part.
saw many offers for 7D12: works only on Dell (vendor locked to Dell-BIOS)
unlocked ones are very rare and working easy on supermicro boards.
for that reason it is not impossible to think they are out of Dell machines (out of ! not made for Dell)

Whereas I do know for a fact just from public information that 7V13's have some level of unique microcode.
Here, I'll give you a free hint.
can not find any information about microcodes there...
and no reason for special microcodes. why do they need them if there ???
 
Last edited:

ExecutableFix

Active Member
Nov 25, 2019
123
60
28
Not sure why you're suddenly angry and trying to bash me, but I'll happily correct you. Not trying to make this thread a fight because that won't help anyone.

Perhaps you shouldn't spout pure guesswork and speculation contrary to known public information.
Ouch, I don't think you realise who you're talking to. (not trying to brag here, just proving a point)

There are two known variants, 7D13 and 7V13.
Wrong. There are way more SKUs. Just scroll through my tweets and you'll find them ;)

Neither are 'OEM' parts.
Correct, didn't say they were.

They are contract customer parts. Unless your name is Jason Zander, Mark Papermaster or Dr. Lisa ****ing Su, you know absolutely zero about what AMD can and does offer contract customers.
You're right in both regards. I don't know the ins and outs of these custom SKUs, otherwise I would've pointed out the problem we're currently facing.

Whereas I do know for a fact just from public information that 7V13's have some level of unique microcode.
Here, I'll give you a free hint.
Wrong again. I own a 7V13 and I've owned 4 of them since September 2020. I can 100% confirm it does not need/use custom microcode. All the fluff Microsoft added is done via firmware, mostly bios and AGESA.

And since you aren't the CEO or even an employee of AMD, you most definitely have no right much less the least capability to speak for them as to what they do and do not do.
Correct. That doesn't mean I don't know what they do (ref. my twitter)

Not to mention the complete absurdity of your delusional insistence that this is a Dell part or even a standard OEM part.
Thanks for being rude about something I didn't even mention. They're not and I've never insisted that they were

If this is a Dell part, then please do explain why not a single Dell system or Dell customer or any customer has found a single one of these "Dell OEM" processors when they've been shipping a full year. Go on. I'm waiting to hear how out of more than 20 million systems shipped in just the first 6 months of 2021, not a single one of them received this part or ended up on eBay or some other reseller. Or anywhere at all within Dell's ordering system.
Seriously. Explain how it is that out of who knows how many hundreds of thousands of Dell EPYC servers that have shipped, not a single one in the entire world has a "Dell 7D13."
Don't need to since this isn't a Dell part. Not sure why you're so angry in this part of your comment?

Seriously, you really need to do a little bit more research before you start insulting people. You're of no help by just lashing out at everyone. Let's go back to a civil conversation now alright?
 

RageBone

Active Member
Jul 11, 2017
548
138
43
@zedei how much of the usual hardware swapping riggamarole have you done so far?
Do you know what i mean?
I again made the mistake of assuming, when will i learn my lesson?
 

zedei

New Member
Jul 3, 2021
11
1
3
@zedei how much of the usual hardware swapping riggamarole have you done so far?
Do you know what i mean?
I have tried 4 different types of RDIMM's (2133, 2400, 2666 and 3200). But the 3200 is SKHynix. These chips work fine with the 7742.
I have tried 1 RDIMM in each of the 8 slots one after the other with attempted power-on for each. (Motherboard manual says for 1 DIMM, put it in slot C2)
I have updated to the latest BIOSes (this board has 2 BIOS's for dual boot between ROME and MILAN).
 
  • Like
Reactions: RageBone

RageBone

Active Member
Jul 11, 2017
548
138
43
I have tried 4 different types of RDIMM's (2133, 2400, 2666 and 3200). But the 3200 is SKHynix. These chips work fine with the 7742.
I have tried 1 RDIMM in each of the 8 slots one after the other with attempted power-on for each. (Motherboard manual says for 1 DIMM, put it in slot C2)
I have updated to the latest BIOSes (this board has 2 BIOS's for dual boot between ROME and MILAN).
that seems pretty diligent and reasonable.
Right now i can't think of anything additional to do, if that changes i will mention it here.
 

RageBone

Active Member
Jul 11, 2017
548
138
43
An hour or so.
holy moly, that is way way way to long.
My perceived record was close to 10 resets before it worked.
An hour is many many more then that.

So yeah, something is very wrong in that setup.
 

LOPP

New Member
Nov 7, 2021
3
1
3
Also a person selling this model has told me this: “the boxes I have they are big qpack boxes and they are compatible with dell it says dell socket on the plastic frame housing”
Does this mean anything to anyone?
 

MistyMoon

New Member
Jun 18, 2021
2
3
3
Even more mysterious than the 7D12/7D13 is the motherboard, which I've hardly ever seen with only four channels of RAM except for ITX.
The 7D12 seems to be a purpose-built CPU for storage. Some boards may not light up the machine if the memory is plugged in full. Only the four outermost channels (CDGH) are available.

---------

Hahaha, the other day at OCP Facebook announced the project called North Dome, which is supposed to be the source of 7D12/7D13.
 
Last edited:

RolloZ170

Well-Known Member
Apr 24, 2016
1,567
404
83
54
700 MHz base o_O I wonder what kind of ridiculously low Vcore these run at?
if cores are parket(idle) this frq. is normal for all processors.
the 0,7Ghz is not confirmed but i think 2,6Ghz is too much...
to keep 36 cores at 2,6Ghz under 95 watts is not impossible with out turbo.
 
Last edited:

dbTH

Member
Apr 9, 2017
128
49
28
I owned a 7D12 for a while and ran it in my home system. It does NOT require a custom microcode (in fact, none of the custom SKUs do because they share the same stepping as retail parts) and it was not vendor locked. A SKU is only vendor locked if it has previously been installed in a Dell or HP system. These SKUs are not meant for HP or Dell, so there’s no factory lock applied to them.

The 7D12 only has 4 memory channels which indicates it was made for a specific board. It will still work on normal boards, but the memory has to be in the right slots, which is why you need to try to boot it with one DIMM first. My guess would be that the same applies to the 7D13.
Did you happen to run any CPU benchmarks on this 7D12 CPU? Many other system components added up the power usage, what was the actual idle and loaded power consumption?
 
  • Like
Reactions: Psycho_Robotico