Finally: Overclocking EPYC Rome ES

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

dealangel2020

New Member
Apr 8, 2020
18
2
3
H11DSI 2.0 BIOS,I can light up ZS1406E2VJUG5,cannot start the system, win10
H11DSI 2.1 BIOS,000000053-04 7742 QS,I can overclock to full core 3.0g
H11DSI 2.0 BIOS cannot Start win10,cannot Start win 2019,The installation system will also get stuck
 

dealangel2020

New Member
Apr 8, 2020
18
2
3
Oh yes that's right. I forgot to mention that there are a couple of bios settings that need to be set on the DSI for the ZS cpu's, it's finicky. So turn on IOMMU and disable Above 4G decoding and it should work
This bug belongs to bios, rome es cpu does not matter, I use the official version of epyc 7702 also to need this step
 
  • Like
Reactions: ExecutableFix

dealangel2020

New Member
Apr 8, 2020
18
2
3
Hi, can you please post your settings for oc? I have the same cpu. What is your Corona 1.3 bench?

Thanks,
Hrvoje
I use the default 64c scheme, and the frequency is changed to 3.2g,The time is 11.58 seconds
But I do n’t know why I ca n’t upload records。
6 00:00:12.28 39,568,700 QiFeiDian… AMD QiFeiQQqun 5167476 @ 2.0GHz (×4) Win
This is my record. I modified the model, I do n’t want to let more people know this cpu 000000053-04
 

dealangel2020

New Member
Apr 8, 2020
18
2
3
This is for dual system? So you used preset 64-core Best of both and that`s it?
That's right, the original R20 running score can be increased from 31000 to 33000.
If you increase the frequency, the performance will decrease, the power supply of the motherboard may be limited.
 

MrCake117

Member
Feb 28, 2019
32
24
8
25
Japan
Here's my result on a h11dsi rev 1.01 with High multi core preset (Unstable for me)

High multi core preset with max frequency of 3450 mhz seems to be more stable

31k.JPG
 

hrvoje.sedlic

New Member
Dec 23, 2019
21
3
3
Do you have a custom bios to make this corona result? We tried before on h11dsi-nt (bios 2.0b) and results were lower then 1 cpu system.
 

Brainbug

New Member
Jul 5, 2015
12
2
3
55
@I.nfraR.ed
Thank you very much for this, in a quick test I can confirm that it is working with a KRPA-U16 and ZS14 CPU. (Set --oc-frequency 1550 --oc-vid 64)

Impressed that the SMU Commands were found so quickly, considering how little public information there is about the SMU.

For others:
1. Don't forget to modprobe msr
2. All Core Boost doesn't show up in the pstate list, but is properly listed using lscpu

I don't have my thermocouple right now, but will connect it to the VRM along with the multimeter, in a few days.
 
Last edited:

I.nfraR.ed

New Member
Aug 20, 2019
19
27
3
Bulgaria
I wouldn't call it too fast. I've been working on the matter since almost an year now. Have been lucky to not kill anything, but had many "oops" moments. Retail parts seem to have more protection in place, while ES are easier to kill/damage.
SMUDebugTool sped up the development a lot. with all the reports users gave me.
First time I found Matisse SMU addresses was also by trial and error, but did not have the scanning tool yet, so I did it by hand.

Almost everything related to these projects is trial and error.
I am not even a C# or Python developer and had to get familiar with both at least on a basic level.

Good that other people joined, otherwise this wouldn't be possible.
 

dealangel2020

New Member
Apr 8, 2020
18
2
3
Oh yes that's right. I forgot to mention that there are a couple of bios settings that need to be set on the DSI for the ZS cpu's, it's finicky. So turn on IOMMU and disable Above 4G decoding and it should work
Does copying the second half of the bios into the bios of H11DSI0.221_D32 work?
 

Nabladel

Member
Jan 27, 2017
38
5
8
37
Hi, can you please post your settings for oc? I have the same cpu. What is your Corona 1.3 bench?

Thanks,
Hrvoje
This was the command I used for the overclock
Code:
sudo ./zenstates.py --no-gui --oc-frequency=3400
The system is running linux. I don't think Corona bench 1.3 is supported? However I did notice that performance is actually lower when overclocked (compiling some projects, bench-marking y-cruncher). I haven't had time to investigate this further yet though.
 

jpmomo

Active Member
Aug 12, 2018
531
192
43
I am trying to understand the specific order that I need to follow in order to at least get my cpu to post. The cpu that I am trying to get to post (I haven't been successful yet) is the 2S1404E2VJUG5 (64-core). The mb that is the guinea pig is the sm h11ssl-nc rev 1.01. I have a naples ES cpu that does boot with that mb. I am assuming that I would need to boot with the naples cpu to be able to make sure the current bios has an Agesa < 1.0.0.3. Not sure where to check that but when I put humpty dumpty back together (currently server parts are everywhere except where they should be!) I can boot into the bios and look around for the Agesa version. If the Ageas version is ok, I think I then need to load one of your custom bios. the bios that I think I need to load is Rome_H11SSLI_Rev1. You mention "The only thing you need to do is copy the second half of the bios (the Rome part) and duplicate it on Rev. 2 boards or just upload that on Rev. 1 boards." I am not clear as to what I need to do. What do you mean by copy the second half of the bios? I was thinking that I would load the whole file (Rome_H11SSLI_Rev1) just like I would a normal bios from sm. After I sort that out and load some new bios, I am assuming my naples cpu would no longer post on that mb. I would then swap the naples cpu with the ES rome cpu and see if the bad boy boots! After that, I would need to boot that server with windows and run your overclocking sw to try and optimize. If that is successful, could I run vmware on that server and still retain the overclocking? Thanks for any pointers and for all of the hard work you have done over the past year trying to get these lost soul cpus working again!
jp
 

dealangel2020

New Member
Apr 8, 2020
18
2
3
I am trying to understand the specific order that I need to follow in order to at least get my cpu to post. The cpu that I am trying to get to post (I haven't been successful yet) is the 2S1404E2VJUG5 (64-core). The mb that is the guinea pig is the sm h11ssl-nc rev 1.01. I have a naples ES cpu that does boot with that mb. I am assuming that I would need to boot with the naples cpu to be able to make sure the current bios has an Agesa < 1.0.0.3. Not sure where to check that but when I put humpty dumpty back together (currently server parts are everywhere except where they should be!) I can boot into the bios and look around for the Agesa version. If the Ageas version is ok, I think I then need to load one of your custom bios. the bios that I think I need to load is Rome_H11SSLI_Rev1. You mention "The only thing you need to do is copy the second half of the bios (the Rome part) and duplicate it on Rev. 2 boards or just upload that on Rev. 1 boards." I am not clear as to what I need to do. What do you mean by copy the second half of the bios? I was thinking that I would load the whole file (Rome_H11SSLI_Rev1) just like I would a normal bios from sm. After I sort that out and load some new bios, I am assuming my naples cpu would no longer post on that mb. I would then swap the naples cpu with the ES rome cpu and see if the bad boy boots! After that, I would need to boot that server with windows and run your overclocking sw to try and optimize. If that is successful, could I run vmware on that server and still retain the overclocking? Thanks for any pointers and for all of the hard work you have done over the past year trying to get these lost soul cpus working again!
jp
I can't change the bios myself, and I need the source code. I can contact ASRock manufacturers, they will help me modify the bios to support es.
 

jpmomo

Active Member
Aug 12, 2018
531
192
43
you can get agesa version from release notes of the latest bios. they mention always the changes made by each bios release
Thanks. I am assuming when they mention < 1.0.0.3, they are talking about the Rome agesa version? the release notes for my bios (rev 1.1 mb) is on a different track. so if my bios release notes shows:

1.1 (02/15/2019) 1. Added support for AMD PSP region update feature flag for BIOS update and DMI check of SUM. 2 Fixed bug of the website "AER/MCE ENABLE/DISABLE FOR XML FORMAT" being unable to find the setting of item "Correctable Error Threshold". 3 Added vendor and status info to the driver health warning message. 4 Added support for "wait for F1 if error" to driver health warning message. 5 Updated ASPEED VBIOS to version 1.08.00. 6 Updated AGESA NaplesPI to 1.0.0.B based on AMI 5.013_NaplesCrb_0ACIJ021.
Would that still be ok?
 

jpmomo

Active Member
Aug 12, 2018
531
192
43
I can't change the bios myself, and I need the source code. I can contact ASRock manufacturers, they will help me modify the bios to support es.

Not sure if you are replying to my questions as I am using a supermicro mb h11ssl-nc. I am wondering if I can just load the pre-modded bios from ExecutableFix. ex. just load the entire bios file Rome_H11SSL1_Rev1?
thanks,
jp
 

jpmomo

Active Member
Aug 12, 2018
531
192
43
this higher agesa version than 1.0.03
I was wondering if there was a different track for the naples cpu vs the rome cpu with regards to the agesa version. ex. for the rev 2 mb, they use a different bios that I thought was relevant to the requirement for agesa to be < 1.0.0.3 but for the rev 1.1 mb, they use a much older bios that only refers to naples. the bios for the rev 1.1 board starts with 1.0b but the bios for the rev 2 mb starts with 2.0a. their agesa seems to be related to the generation of the amd cpu (either naples or rome.) So when you reference the requirement for the agesa to be <1.0.0.3, is this referring to the rome agesa version or both the naples and rome?
Thanks for helping me to try and get some clarification,
jp