Overclock your AMD Epyc

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

ExecutableFix

Active Member
Nov 25, 2019
123
64
28
So has anyone reverse engineered it to figure out exactly what it writes, and where? and then tried the same thing on rome?
Well, me and Neo have been working together with infrared (who makes/maintains the zenstates app) and haven't been able to get it to work. No reverse engineering needed, the source is even on github
 

charliehorse55

New Member
Oct 3, 2019
16
2
3
Well, me and Neo have been working together with infrared (who makes/maintains the zenstates app) and haven't been able to get it to work. No reverse engineering needed, the source is even on github

Oh :/. That's kind of sad to hear. Is there a discord or something where you are sharing what you've tried? I'd love to learn more about this.

Those 1.4 GHz 64C ES chips on ebay are so tantalizing, but only useful for me if I can overclock it back close to stock.
 

Gordan

Member
Nov 18, 2019
39
8
8
The base clock doesn't seem to matter too much, my 7402P is 2.8 base 3.35 max boost, and it stays at about 3150-3200 under stress test load on all cores. There is a good chance that you will see a similar effect with the 64C, though inevitably it will boost less because heat dissipation will be a serious issue. If I'm doing 3200MHz at 200W TDP on 24 cores, you would be burning close to 500W to do it on 64 cores, which just doesn't seem plausible anyway, even with water cooling.
 

jerrytsao

MILAN X P5800X
Sep 11, 2016
39
31
18
Shanghai, CN
The base clock doesn't seem to matter too much, my 7402P is 2.8 base 3.35 max boost, and it stays at about 3150-3200 under stress test load on all cores. There is a good chance that you will see a similar effect with the 64C, though inevitably it will boost less because heat dissipation will be a serious issue. If I'm doing 3200MHz at 200W TDP on 24 cores, you would be burning close to 500W to do it on 64 cores, which just doesn't seem plausible anyway, even with water cooling.
Not quite high, 280W high-binned SKUs such as 7R32 48C (basically a 7H12 version of 7642) has an all-core turbo of 3.3G, 7H12 64C should probably be around 3G, that put 64C 3.2G ~350W.

 
Last edited:

Gordan

Member
Nov 18, 2019
39
8
8
Does anyone know if the cTDP limit is purely a function of the BIOS code? Or is the limit somehow enforced by the CPU?

If it is purely a BIOS enforced feature, are there any tools to modify this?

Are BIOS firmwares cryptographically signed? Would disassembling/decompiling, modifying and recompiling approach work as a vector for approaching raising the cTDP limit?
 

ExecutableFix

Active Member
Nov 25, 2019
123
64
28
Does anyone know if the cTDP limit is purely a function of the BIOS code? Or is the limit somehow enforced by the CPU?

If it is purely a BIOS enforced feature, are there any tools to modify this?

Are BIOS firmwares cryptographically signed? Would disassembling/decompiling, modifying and recompiling approach work as a vector for approaching raising the cTDP limit?
It depends. A bios isn't encrypted nor does it have a checksum. Some parts are just assembly, others PE32 and the rest some other languages. I don't belive the microcode is encrypted either, but there's as far as I'm aware no way to modify it.

The cTDP in the bios doesn't affect my ES CPU, maybe it does work on a retail Rome chip?
 

ExecutableFix

Active Member
Nov 25, 2019
123
64
28
That's actually exactly the case, underload freq dropped on that particular ES chip.
It doesn't indicate an overclock tho. The max boost frequency is 2.2Ghz while the all-core boost under heavy load is 1.2Ghz. So far I've managed to lock the clocks on 2.175Ghz across all cores which isn't an overclock by any means
 

charliehorse55

New Member
Oct 3, 2019
16
2
3
It doesn't indicate an overclock tho. The max boost frequency is 2.2Ghz while the all-core boost under heavy load is 1.2Ghz. So far I've managed to lock the clocks on 2.175Ghz across all cores which isn't an overclock by any means
Any progress on per-core controls?
 

Gordan

Member
Nov 18, 2019
39
8
8
What's the point of a WB on a CPU that runs lukewarm at stock settings and can't be OC-ed?
 

Gordan

Member
Nov 18, 2019
39
8
8
"Can't be oced" go check the thread title genius

The purpose of this block is mostly to cool the VRM
Right, I foolishly assumed that you were talking about Rome. Possibly because that's what all the recent posts on the thread talk about.
 

alex_stief

Well-Known Member
May 31, 2016
884
312
63
38
The point of this monoblock is mostly hassle-free VRM cooling. And it's even cheaper than two separate CPU waterblocks, and then you would still need to figure out something for the VRMs.
Good to hear that yours got shipped, I am still waiting for mine. Ordered it back in November. The french really know how to do a general strike :rolleyes:
 

TXAG26

Active Member
Aug 2, 2016
397
120
43
The point of this monoblock is mostly hassle-free VRM cooling. And it's even cheaper than two separate CPU waterblocks, and then you would still need to figure out something for the VRMs.
Good to hear that yours got shipped, I am still waiting for mine. Ordered it back in November. The french really know how to do a general strike :rolleyes:
I thought a lot of the EU countries went on “holiday” for a month or two around Christmas.
 
  • Like
Reactions: alex_stief

alex_stief

Well-Known Member
May 31, 2016
884
312
63
38
Nah, that's just between December 21st and January 6th for this holiday season. There really was a general strike in France.
 
  • Like
Reactions: TXAG26