AMD ES ZS1406E2VJUG5 microcode version?

Notice: Page may contain affiliate links for which we may earn a small commission through services like Amazon Affiliates or Skimlinks.

Gabber

New Member
Jun 19, 2020
13
6
3
Italy
Please note that there is some kind of bug during IPMI bios updating on ASUS KRPA when you have ES epyc installed (checked on ZS1406E2VJUG5 and ZS1407 ) that at the end it bricks motherboard and the only option recover is to flash chip with 3rd party programmer. Flashing through IPMI is working fine with retail CPU. For Supermicro H11SSL motherboards this problem do not exist and downgrading works perfect trough IPMI (just downgraded from 2.1 bios to 2.0a to boot ZS ES cpu)
I tried flashing (downgrading) bios via IPMI on Asus KRPA with both retail and ES cpus and it never worked. Looks like it worked, but when you restart the system notingh change. I never had a brick condition, maybe it can depend by the version of starting bios and bmc firmaware... I don't know for sure.

The only way that really worked for me was the directly flash with a programmer.
 

boomheadshot

Member
Mar 20, 2021
64
3
8
Good morning everybody, I would like to ask to @ExecutableFix and @yesoos if they fix the Asus KRPA bios problem?

I'm in a similar situation, I have:

- AMD ES ZS1406E2VJUG5
- 8 x 32GB Samsung M393A4K40DB3-CWE (3200MHz) or 1 x Kingston KSM32RD8/16MEI (3200MHz)
- ASUS KRPA-U16 shipped with 0501 bios


I try to flash the 0302 bios via ipmi (with /bios_update in maintenance tab) and it shows correct completition, but if i check again it show up the 0501 version again.
Then I bought a OEM cpu (AMD Epyc 7262) and the system boot fine (also windows works) so I decide to flash the 0302 bios via AZ Flasher and it looks there.
Now the real strange thing: Bios info shows 0302 version, AZ Flasher shows 0302 version, ipmi shows 0501 and I get D1 post error with the ES cpu; the oem one works great.

I'm looking for a solution because i'm becoming crazy!! Was someone of you able to resolve this bios problem? Can you please explain me how to resolve it?

thank you a lot and sorry for my english!
I'm thinking of putting together a similar system, but only with an 2s1404e2vjug5. Does it say the BIOS version on the motherboard box? I just want to see if it's possible to check this before buying the motherboard.
 

Gabber

New Member
Jun 19, 2020
13
6
3
Italy
I'm thinking of putting together a similar system, but only with an 2s1404e2vjug5. Does it say the BIOS version on the motherboard box? I just want to see if it's possible to check this before buying the motherboard.
Sorry, I don't really know. If you power-up the motherboard, you can connect to it by the ipmi RJ45 port and check the bios version; try to ask it to the seller... maybe you will be lucky.

PS: Buy an item with a good return policy and give it a try. It is not impossible to flash the bios.
 
  • Like
Reactions: boomheadshot

boomheadshot

Member
Mar 20, 2021
64
3
8
Sorry, I don't really know. If you power-up the motherboard, you can connect to it by the ipmi RJ45 port and check the bios version; try to ask it to the seller... maybe you will be lucky.

PS: Buy an item with a good return policy and give it a try. It is not impossible to flash the bios.
Thank you for your comments and that link about the SPI header pinout from the ROG forum, everything really is correct (I checked with a multimeter).

I only had a CH341A programmer and I didn't have a 16 pin clip, and the only solution was to use was to use jumper wires. Guess what? In my city I could only find 2.54mm wires, so I had to jerry rig it. On the motherboard (Asus KRPA-U16) the SPI header pins are in fact 2mm apart, so I couldn't squish more than 2 wires beside each other (the plastic end is just too wide). Took me a while to hook it up, but I had to hook up a naked wire on either side (they were surrounded by the plastic insulation of the neighbouring wires. Checked them with a multimeter just in case


spi1.jpg

here, the beige cable is exposed and wrapped around the pin


sp2.jpg
Here I had to expose the blue one.


Anyway, the SPI header pinouts from
were absolutely correct, but I checked everything with a multimeter just in case.

sp3.jpg

Here's how I arranged the jumper wires for my CH341A:
SPI pin --> CH341A pin
5 --> 1
6 --> 6
8 --> 5
4 --> 4
7 --> 2
2 --> 3
9 --> 7
3 --> 8

I think that if you're also gonna use a CH341A, a usb extension cable will be essential because I couldn't hook up the programmer without having the exposed wires fall off, so be prepared ahead of time and make sure you have one.

As stated in the source, pin 1 is not used.

I'm a complete noob so I'm pretty proud of doing something like this for the first time ever, so YAY.
I used AsProgrammer on Windows, and I had to gamble because the exact chip on this board isn't on the supported list, but I chose another similar one which only has 2 or 3 letters that are different at the end, but the size was also 32768 kB so I just YOLO'd it.

Took 7 minutes to read the chip, and it took a whopping 20 minutes to flash version 0302.
I tried checking the version of the BIOS that I originally had, but it was hard to read (I think I found it, it was like 4102 or something, but I could have read it wrong). If it's important, I've got the dump saved on another drive and I'll upload it if anyone needs it.

No onto my next problem - I've got a 2s1404e2vjug5, and with the overclocking tool it overheats even on 1.8GHz with a IceGiant Prosiphon Elite. I first tried Deepcool Fryzen, but as soon as you load the cores (I did it on Cinebench R15), the CPU die temp shoots >110C on HWinfo64.

Then I got the IceGiant Prosiphon Elite, but the problem still persist. WhenI start the benchmark, the temperature shoots up to 110/120C, and as soon as it finishes, it drops back down to 59 C
bench1.png
Am I screwed? Or is this an inaccurate temp or something?
Or is it that the 2S is that much worse than the ZS?
 

RageBone

Active Member
Jul 11, 2017
617
159
43
that header is right beside the 32MB BMC EEPROM, are you sure you have actually flashed a bios to the bios EEPROM?
 

boomheadshot

Member
Mar 20, 2021
64
3
8
that header is right beside the 32MB BMC EEPROM, are you sure you have actually flashed a bios to the bios EEPROM?
1621184388582.png

The BIOS chip is circled in red. It's a Winbond 25Q256JVFQ

To be safe, I performed a continuity test with a multimeter.

W25Q256JV Datasheet by Winbond Electronics | Digi-Key Electronics
the only difference is that the chip on the ASUS ends in -JVFQ and not just -JV
On page 7, you have this.
1621185240551.png

I used a multimeter to perform a continuity test. I connected the SPI header pins to the CH341A programmer in the correct pins (without connecting it to USB, everything powered off), then I put 1 probe of the multimeter on one of the legs of the BIOS chip, and the second probe on the SPI pin just under the jumper cable, and it would give a beep if it was correct. If you're curious, these are the corresponsing legs/pins:

I was baffled by the terminology (I guess in the data sheet they're using more politically correct terminology or what, but luckily there was a person that helped me). Some of the terms are in fact the same, but not all.
CHIP leg#(datasheet name) ---> SPI header pin
7 (chip select) --> 5 (chip select)
16 (CLK) --> 6 (CLK)
15 (Data Input) --> 5 (Master out slave in)
10 GND --> 4 GND
8 (Data Output) --> 7 (Master in slave out)
9 (Write Protect input) --> 2 (WP#)
1 (Hold or Reset input) --> 1 (Hold#)
2 VCC --> 3 VCC

Maybe this is overly simplified, but Gabber's noob-friendly explanation really saved me, so I hope my instructions also help a layman like me.
 
Last edited:

RageBone

Active Member
Jul 11, 2017
617
159
43
The Winbond 25Q256JVFQ with 256M Bit = 32M Byte is in most cases the EEPROM for the BMC, an AST2500 in your case.
Of cause it could be used as the actual BIOS EEPROM, but i believe that to be unlikely.

Do you still have a working ipmi remote management?
I assume you don't.
Is the Heartbeat-LED still blinking?

I'm failing to actually identify a second SPI EEPROM on the board from pictures on the web.
So i can't tell you right now where another one would be.
Ausus moved away from the usual SOP8 package to something without leads on all their nice consumer AM4 boards as far as i know.
Also AM4 seems to be 1.8V which would be bad since your CHA is using 3.3 or 5V.

I know my Supermicro H11s still are using 3.3V based SOP8 chips for the Bios.
 

boomheadshot

Member
Mar 20, 2021
64
3
8
The Winbond 25Q256JVFQ with 256M Bit = 32M Byte is in most cases the EEPROM for the BMC, an AST2500 in your case.
Of cause it could be used as the actual BIOS EEPROM, but i believe that to be unlikely.

Do you still have a working ipmi remote management?
I assume you don't.
Is the Heartbeat-LED still blinking?

I'm failing to actually identify a second SPI EEPROM on the board from pictures on the web.
So i can't tell you right now where another one would be.
Ausus moved away from the usual SOP8 package to something without leads on all their nice consumer AM4 boards as far as i know.
Also AM4 seems to be 1.8V which would be bad since your CHA is using 3.3 or 5V.

I know my Supermicro H11s still are using 3.3V based SOP8 chips for the Bios.
1621264265535.png

I'm pretty sure it's actually the BIOS header because it worked and I got a dump of my BIOS and was able to flash version 0302 from the other thread and everything actually booted after this

Here's the dump of my BIOS in case anyone needs it (SHA256: 18b4b6ced46514eb9aa9b38e193d5d224962c676aed2eb94e6957461d3a0cf55). It actually does seem to be version 4102, I guess it's a more recent version.

The heartbeat LED still blinks. I haven't tried using IPMI because I don't know how to use it/why I need it :S, but I checked the IPMI HW monitor in bios and it only showed the CPU temp and the 12v/5v/3.3v and some other voltages, without any irregularities. IDK if that's what you m

HWinfo64 does show that the mobo sensors is a nuvoton nct6793d, and after some Googling those sensors really do seem to be riddled with such problems, and the general consensus has been to just update the chipset drivers. Never knew those could be essential, the more you learn. Only last night did I realize that there are actually Linux drivers on the official webpage. I'll try another version of Windows server first (for some reason couldn't load the USB, it just freezes with the blue windows logo. I've tried several UEFI/Legacy combinations, making a Rufus GPT or MBR partition scheme, but no luck. The Win 10 pro version actually works fine for some reason. Because I guess if I manage to install the chipset drivers, my problems will be remedied (at least partly).
 

RageBone

Active Member
Jul 11, 2017
617
159
43
Welp, you are obviously right in that you flashed a BIOS to a chip previously containing a BIOS, so, everything is fine i guess?
The Backup is 32MB as well so it could be the actual EEPROM you marked but then, where is the BMC EEPROM?
Eh, doesn't really matter at this point.

The ipmi is usually used for remote management. Power on and off, resets, boot from virtual media, iKVM, and Bios and FW Updates.
Many more small features that can be useful.

So you might have been able to downgrade the BIOS through that. Might because some vendors do licensing stuff to make you pay for that feature.
Or don't allow downgrades. Or don't allow modded Bioses.
So in any case, good experience to know how to recover Bioses with a CHA341A.

Congratz!
 

boomheadshot

Member
Mar 20, 2021
64
3
8
Well I saw previous posters say that flashing via IPMI/with a retail CPU doesn't work (not like I even had the opportunity to try that), so I was prepared for the flashing route (kinda).

So I installed windows server 2019, installed the chipset drivers, no luck. Same thing, constant temperatures on the motherboard.

I remember running a windows update on win 10 pro when I just hoped it would automatically update whatever is needed, and then after that I started getting BSOD's related to the NVIDIA driver. It's either during those updates or my first stress test (my only guesses).

Can the BMC firmware have such an effect? What if I try updating it? Do I need to back it up, is it possible for me to lose CPU support if I happen to update it?

Here's what the IPMI HW outputs (idk if this matters, but just to give you an idea):
1621280233963.png

What's interesting is that I've tried running monero mining for a couple of minutes (in case the CPU Die temp is actually correct and I don't wanna dry it).
When I set the CPU freq at 1.8 GHz, my hashrate goes up to like 25k/s, so approximately 200 hashes per thread. When I set the frequency up to 2.6 GHz for a short while, this doesn't help the hashrate at all. But when I go from 1.8 GHz --> 2.6 GHz on Cinebench R15, it gives me something like a ~20-25% increase (from 4300 @ 1.8 to >5000 on 2.6 GHz).
But the thing is that I've only got 4 sticks of ram (they were kind of expensive, ~$100 for each M393A2K43DB3-CWE 16 GB module, only 1 place to buy from in my country). I reckoned that if threadripper can pull off 50+ kh/s, then this CPU should be able to do it on 4 channels.
So I performed another test. I took out 2 sticks of RAM (left only 2 as per the manual in the correct slots), and my hashrate only dropped from 25kh/s to 22.5 kh/s, so each channel can do ~11kh/s. Am I making sense with this line of thinking or am I missing something? I can splurge on another 4 x $100 sticks, but only if it's guaranteed to work.

I was really hoping to get around 40-45 kh/s with this CPU, because there are people that get >40 kh/s on a ZS cpu.

Am I really that outta luck with the 2S, or do I need to sort out the thermal reading and then fine-tune the frequency/voltage.

When I checked the specs on cpu-world.com, the only thing is that the ZS only has a 200 MHz higher turbo clock than the 2S. Can there be such a huge performance difference between these 2 steppings? I remember reading about someone saying that the ES's are hella inefficient and put out much more heat.
 

Gabber

New Member
Jun 19, 2020
13
6
3
Italy
@boomheadshot I reply to all without quote.

If you extract the metal pins from the plastic housing of the jumper cables and you wrap theyself with same electrical tape you can use they in the 2 mm pin array.

That flashing time seams long to me, but if the flashing process works great that is not a problem (and it seams to work!).

That temperature is crazy!!! But it was a strange number... I can't really say if it is true or not. Maybe not.
With my ZS I got a mean of (real on pool) 30-31kH/s with no more than 63°C and a mean over a day of 60°C.
You should look for kH/s / W not kH/s alone ;)

When I really overclock it (3.4-3.8 GHz all core) I pass 80°C but get nealry 10k in cinebench 15.
Also check out CPU VRM temperature: I have a custom watercooling loop and a fan blowing directly on VRM.

Can you check your wall power consumption? I don't really know the difference between the 2S and ZS in terms of heat, but remember that all are samples and and not all equil each other.
 

RageBone

Active Member
Jul 11, 2017
617
159
43
i forgot to mention. BMC usually also handle some sensors and Fans and sum such additional stuff.
So a BMC FW Update could fix those issues you are having. Could!

It is unusual for such an update to affect cpu compatibility.
It probably technically could, but it would be pretty unreasonable to do so from the OEM.

You would be safe with a full flash BMC backup.
Though, where is that eeprom?
 

boomheadshot

Member
Mar 20, 2021
64
3
8
@boomheadshot I reply to all without quote.

If you extract the metal pins from the plastic housing of the jumper cables and you wrap theyself with same electrical tape you can use they in the 2 mm pin array.

That flashing time seams long to me, but if the flashing process works great that is not a problem (and it seams to work!).

That temperature is crazy!!! But it was a strange number... I can't really say if it is true or not. Maybe not.
With my ZS I got a mean of (real on pool) 30-31kH/s with no more than 63°C and a mean over a day of 60°C.
You should look for kH/s / W not kH/s alone ;)

When I really overclock it (3.4-3.8 GHz all core) I pass 80°C but get nealry 10k in cinebench 15.
Also check out CPU VRM temperature: I have a custom watercooling loop and a fan blowing directly on VRM.

Can you check your wall power consumption? I don't really know the difference between the 2S and ZS in terms of heat, but remember that all are samples and and not all equil each other.
The ZS can do 45k, apparently.

My electricity is kind of cheap ($0.04 kW/h), so that's not that big of a deal. I was still hoping to get at least 40kh or something.

Can you please test with 4 RAM DIMMs to see the difference? I would really appreciate this.

There is definintely a problem with the sensors, I'll try to backup the BMC firmware and update it.
I'll also look around for BMC EEPROM
 

sungurvid

New Member
Mar 10, 2022
1
0
1
I noticed both retail and es/qs Rome CPUs use the same microcodes,my friend tested them es ones supported by Asus KRPA-U16 motherboards manufactured before April 2019.I guess it is cpld chip controls what models CPUs supported. vidmate app download insta save
 
Last edited:

RageBone

Active Member
Jul 11, 2017
617
159
43
So far it seems that those cpus work without any mircocode inside the Bios.
At least as far as i and other did test it.

Asus had issues with CPUid on lga2011-3 boards.
Haswell CPUs worked but Broadwell did not unless you had a specific Revision that had updated FW.
That update was not provided publicly and only done in Asus service centers.
So their CPLDs, SIOs, BMCs and MCus doing some CPUid checking has precedent.