Finally: Overclocking EPYC Rome ES

[RolloZ170]

Nothing about IPMI chip here https://www.supermicro.com/manuals/motherboard/EPYC7000/MNL-2085.pdf

use glasses. Figure 1-1.
i see two SOP in a socket, one was BIOS you already indentified, other in socket SOP16 is BMC FW chip.

 

[mafon]

use glasses. Figure 1-1.
i see two SOP in a socket, one was BIOS you already indentified, other in socket SOP16 is BMC FW chip.

I found MXIC MX25L25635FMI-10G but have no idea how to flash it.

 

[RolloZ170]

I found MXIC MX25L25635FMI-10G but have no idea how to flash it.

buy a clamp SOP16 for the ch341a (because can only flash 8pin's)
the outer 8 pins only are required.

 

[Frk]

Which tool/utility do you guys recommend to monitor the set values of PPT, EDC and TDC? I'm trying to verify whether the values set by my port of ZenStates are applied properly. ryzen_smu is not compatible with EPYC ES.


Thank you.

 

[mafon]

I have bought SOP16 clamp CH341a, but there is no instruction how to connect wires.

 

[RolloZ170]

I have bought SOP16 clamp CH341a, but there is no instruction how to connect wires.

i bought a SOP16 test clamp/clip with a 8Pin header on it, plugNplay for ch341a

the SOP16 has 8 unused pins(reset included)
pinout above

https://forums.servethehome.com/index.php?threads/finally-overclocking-epyc-rome-es.28111/post-479604

 

[JUGS]

Does anyone know how we can stop ES epycs from going to very low frequency while idling?

 

[RolloZ170]

Does anyone know how we can stop ES epycs from going to very low frequency while idling?

they run at base clock. nothing you can do.

 

[JUGS]

they run at base clock. nothing you can do.

Nothing from ZenStates or BIOS itself? it goes to 900mhz which is very low and slows down the programs in general when load is not constant

 

[RolloZ170]

it goes to 900mhz which is very low and slows down the programs in general when load is not constant

that is a P / C state, e.g. cores are parked by OS.
at Intel EIST, at AMD it is called Cool'n'Quiet ??? not sure.
check BIOS and if there disable.

 

[JUGS]

that is a P / C state, e.g. cores are parked by OS.
at Intel EIST, at AMD it is called Cool'n'Quiet ??? not sure.
check BIOS and if there disable.

P0 is 1.4Ghz, by default it is 900mhz. it does not let me set higher than 1.4GHZ

 

[RolloZ170]

P0 is 1.4Ghz, by default it is 900mhz. it does not let me set higher than 1.4GHZ

more that 1.4Ghz only with load, run seti@home ?

 

[Frk]

Is the improvement only for dual-socket?
Can you post a link to the binary so I can test it on mine? I have a single 100-000000053-04.

Any update on this? I would also love to test this. I have 2x systems one w/ (H11SSL and single 100-000000054-04) and another (H11SSL and single ZS1711E3VIVG5)

It's not done yet. I need to verify if the voltages it sets matches the original version, but I don't know which tooling can reliably monitor the values of PPT, EDC and TDC on Ubuntu Server. Which tool/utility do you guys recommend to monitor the set values of PPT, EDC and TDC?

Thank you.

 

[kyocooro]

Any zenstates linux version that supports Milan ES around?

CPUs: 1
CPUID: 00A00F10
Package Type: 4
CPU not supported!

Hi all,

Following up on my overclocking issues with my dual ZS1406E2VJUG5 setup:

After further research and considering the issues some have noted with newer Proxmox versions, my current hypothesis is that the problem lies with how recent Linux kernels handle direct MSR (Model-Specific Register) calls. It seems that due to ongoing security hardening efforts, newer kernels have become more restrictive about allowing direct userspace writes to certain MSRs. Even if the /dev/cpu/*/msr interface appears loaded and writable, the kernel might be silently ignoring or blocking the specific MSR writes that ZenStates uses to apply overclocks. This would explain why the settings aren't taking effect and would also align with reports of the tool breaking between Proxmox 7.1 and 7.2, likely due to kernel updates.

Given this, I've concluded that the most promising path forward is to adapt the ZenStates script (or develop a new approach) to use SMU (System Management Unit) calls instead of direct MSR manipulation. The good news is that the smu-test-message utility appears to be functioning correctly on my system, which is an encouraging sign for this approach.

I'm now in the process of trying to convert the relevant parts of the ZenStates logic to utilize SMU calls. I understand this is a non-trivial task, as it involves a different set of commands and a deeper understanding of the SMU's interface for these EPYC CPUs.

I'll be sure to post any progress I make here. If anyone has experience or resources related to SMU interaction on EPYC platforms, any advice would be greatly appreciated!

Best regards.

Hi, can anyone share the ZenStates version for Milan?

@RolloZ170


i update code for Milan es, use with your own risk

GitHub - kyocooro/ZenStates-Linux-Milan-ES: Dynamically edit AMD Ryzen processor P-States, Milan ES support

Dynamically edit AMD Ryzen processor P-States, Milan ES support - kyocooro/ZenStates-Linux-Milan-ES

github.com

zenstates.py --oc-frequency 3400 --oc-vid 56 --ppt 1500 --tdc 700 --edc 700 --oc-enable --no-gui --lock-freq (MilanUnlockOverclock high-single core profile, 64cores milan es)


add lock freq feature for better perf ( get ideal from MilanUnlockOverclock)

 

[pnp_PioNer]

Hi all.
I have 2 custom bios for asus krpa-u16 for Milan and ROME based on Version 0701(Milan) and Version 4501(ROME). They have been working properly on my boards for over a year, dividing SNLOT No. 1 pci-e x16 into 4x4x4x4.
But I recently received a board with firmware Version 1002(Milan) and was unable to downgrade the bios version. The firmware message says that "the version I'm trying to flash is older than the one that has already been flashed".
Question: does anyone know how to downgrade the bios version?
bmc, flash from bios and dos works without problems, but everywhere there is a failure for the reason mentioned above.

 

[shpitz461]

Question: does anyone know how to downgrade the bios version?

If software is giving you issues flashing, you can always flash directly using a ch341A USB flasher.

 

[pnp_PioNer]

OK, it's obvious. then the next question is how to get out of the file.cap to make a dump image that will not ruin the board and will save unique data such as mac, sn ......

 

[RolloZ170]

OK, it's obvious. then the next question is how to get out of the file.cap to make a dump image that will not ruin the board and will save unique data such as mac, sn ......

remove the 1000h / 4096 bytes capsule header. mac / serial can be set if stored in BIOS (i doubt that)

 

[EdwardTwiss]

remove the 1000h / 4096 bytes capsule header. mac / serial can be set if stored in BIOS (i doubt that)

@pnp_PioNer I have had a similar issue with KNPA-U16 board in the past when trying to downgrade to use an EPYC ES on it.
You need to pop out the BIOS ROM and flash it with a flasher to the version you want to use. It was five or so years ago I did it but ASUS support wasn't much use apart from telling me you can't downgrade the BIOS to an older version.
Another option is to get hold of a BIOS chip and swap it out for that as they'll more than likely have an older BIOS on them: Krpa-U16 Bios for sale | eBay
You want to go back as far you can then flash up to the version you want to use. Setting up a BIOS flasher will allow you to roll back again if you go too far up in version like I did previously.

 

[RolloZ170]

You need to pop out the BIOS ROM and flash it with a flasher to the version you want to use.

right, another way without HexEditor.
load the cap with UEFITool and select UEFI section, extract as is...