Is it safe to disable IA CEP and IA CEP 14 GEN and use lite load with 0x129 ? any consensus ?

Sort by date Sort by votes
Are you crazy? That’s the debate of the century around these parts. We had just got everybody settled down and back in their corners after the fighting broke out, and you have to go and stir things up again. I mean, it was like a scene from Saturday Night Fever around here. What are you, a deliberate instigator or something? ;)

No, seriously. It comes down to personal preference. You can read all about it over here. https://forum-en.msi.com/index.php?...led-and-how-they-compare-to-lite-load.400984/
 
Upvote 0 Downvote
LOL I tried to keep up but it got very technical and way over my head to be honest ,so I thought I would defer to the wisdom of those more technically minded than myself .Toe tipping and backing away slowly …
 
Upvote 0 Downvote
The best that I can say on the subject, especially in under 1000 words, is that you probably want to decide on CEP enabled/disabled as your very first decision. I say this because I have a slight suspicion that it can alter certain presets, etc. Given that statement, which I’m still studying from afar, if I were in your shoes, a little confused about what to do, then I would start out with it enabled and then tune from there. Just keep an eye on your Cinebench R23 scores to make sure that they are in line with what you would expect for your particular CPU, plus or minus a few percent points. If your scores drop precipitously then back off the last thing you changed. If you just are not happy with the results you are getting then save what you have done to a BIOS OC Profile and then start from scratch with CEP disabled. Then compare the two and see which one nets you the best scores and temps.

One common tuning principle with CEP is that you might not be able to reach your undervolting goals by just lowering AC_LL. You might need to augment that approach with a manual undervolt, too.

Does that give you enough info or do you have another question about the general approach of undervolting? Feel free to ask me or anyone on the thread I sent you to.
 
Last edited:
Upvote 0 Downvote
So far I'm not aware of any evidence for unsafe effects of turning off CEP, but some people suspect it might not be ideal in the long term. I also don't understand what's a realistic situation the CEP should be protecting against, so far we've seen it get triggered by wrong settings that by themselves don't appear to be immediatelly "dangerous"

Also I'm starting to suspect AC_LL shouldn't be used for undervolting and should instead be set according to the electrical characteristics of the motherboard at a chosen LLC and undervolting should be done with a VID offset setting instead. So far from my tests it seems this way it gives better results with CEP on (but we still don't know why or how). However I still haven't tested any newer (2024) bioses or microcodes, so things might still change
 
Upvote 0 Downvote
charonme said:
So far I'm not aware of any evidence for unsafe effects of turning off CEP, but some people suspect it might not be ideal in the long term. I also don't understand what's a realistic situation the CEP should be protecting against, so far we've seen it get triggered by wrong settings that by themselves don't appear to be immediatelly "dangerous"

Also I'm starting to suspect AC_LL shouldn't be used for undervolting and should instead be set according to the electrical characteristics of the motherboard at a chosen LLC and undervolting should be done with a VID offset setting instead. So far from my tests it seems this way it gives better results with CEP on (but we still don't know why or how). However I still haven't tested any newer (2024) bioses or microcodes, so things might still change
Click to expand...
We should probably add the one caveat that if we were ever to ditch AC_LL as a mechanism for lowering voltages and temps, we’d need to think hard about how we advise people to apply their negative offsets, i.e. more off the top, less off the bottom, and a health chunk off the upper-mid frequencies - where the all-core workloads exist. I guess every action has a reaction; and complexity becomes hard to avoid no matter how we undervolt.

Let’s hope that Arrow Lake runs cooler (and with lower voltages) out of the box so we don’t have to keep trying to fix what Intel broke.
 
Last edited:
Upvote 0 Downvote
To the OP: If you don't want to spend a lot of time on it, go with the lowest stable AC_LL (--and CEP disabled) method. Follow this simple guide:

Bottom line: I have yet to see anything from Intel (or any plausible account or evidence elsewhere) that would indicate that CEP is intended for safety (i.e. to prevent degradation). The Intel technical description is that it throttles the chip when the voltage falls BELOW a certain threshold, i.e. it is a stability feature:
"CEP, or Current Excursion Protection, is a function designed to prevent system instability caused by insufficient CPU voltage. "

Stay Cool and Perform Better: MSI BIOS Update Allows to Disable CEP for Intel 14th Gen Processors

MSI introduces BIOS updates for Intel 700 and 600 series motherboards, allowing to disable the CEP (Current Excursion Protection) function. This adjustment allows processors to operate at lower voltages and temperatures without compromising performance.
www.msi.com www.msi.com

In contrast, whatever lowers your max temps, or lowers peak currents/voltages at higher temps, greatly reduces degradation. Aside from setting sensible iccMax (current) and PL1/PL2 limits, strong undervolts that kick in especially when the system is under high temps/load would be best for safety. And AC_LL reduction (which reducing CPU Lite Load does) is one of the best and surely the simplest way to do just that.
 
Last edited:
Upvote 0 Downvote
Listen partner, don't you go trying to talk me out of CEP when I was just beginning to think it was the greatest thing since sliced bread. Well, maybe just slice bread and ABBA. But that's IT...my final word on the subject... Except...well...maybe microwave dinners... But absolutely NO MORE than that...really!
 
Upvote 0 Downvote
lopez.pos155702d9 said:
I have yet to see anything from Intel
Click to expand...
Current Excursion Protection (CEP)
This power management is a Processor integrated detector that senses when the Processor load current exceeds a preset threshold by monitoring for a Processor power domain voltage droop at the Processor power domain IMVPVR sense point. The Processor compares the IMVPVR output voltage with a preset threshold voltage (VTRIP) and when the IMVPVR output voltage is equal to or less than VTRIP, the Processor internally throttles itself to reduce the Processor load current and the power
Click to expand...
(emphasis mine)
I don't see any official intel indication it's supposed to be a stability feature, nor any practical evidence that it helps stability in case you undervolt too much with a VID offset setting. I've seen plenty of crashes when I set my offset too low.
However it's not clear against what exactly "preset threshold" is it comparing and what is the situation when such a thing can happen (outside of the false positive when the board sends an unexpectedly low voltage to the cpu as a result of incorrectly set LLC and AC_LL or VRM offset) in which the CPU needs to be protected from excessive current (there is already the IccMAX limit mechanism) or why such an excessive current would happen (perhaps physical damage resulting in short inside the cpu?).
See also this interesting discussion
 
Last edited:
Upvote 0 Downvote
So, its still not clear if there are any downsides to disabling it?

as temperatures go, I think CEP turned off, Lite Load Mode to the lowest with LLC to mode 8 (most vdroop) keeps the CPU the coolest under heavy workloads, so there is an UPSIDE to disabling it.
 
Upvote 0 Downvote
disabling it only helps when your AC_LL is set incorrectly lower than it's supposed to be according to the LLC. Also buildzoid says CEP reduces undershoots so maybe there is some stability benefit after all, but I don't know about any actual stability measurements and comparisons. Maybe I should try it myself.
The downside would be the unknown mysterious hypothetical situation it should protect against would cause excessive current and possibly damage the cpu.
What are your exact settings where you see better results with CEP disabled? I'd like to try to reproduce it
 
Upvote 0 Downvote
Lite Load Mode 9, PL253, ICCMAX 307, the rest on auto, IA CEP disabled.

Right now, Ive tried to lower PL to 200, to see if it doesnt thermally throttle, but it STILL DOES ON PCORE 7! even though the rest is max 83 degrees.

I, Absolutely, Hate, The, i9 14900k. Worst purchase Ive done, too much headache

This is the result of (A 36k score) R23 run

DumbStupidCPUIHateYou.jpg
 
Upvote 0 Downvote
MigraineFilms said:
Lite Load Mode 9, PL253, ICCMAX 307, the rest on auto, IA CEP disabled.

Right now, Ive tried to lower PL to 200, to see if it doesnt thermally throttle, but it STILL DOES ON PCORE 7! even though the rest is max 83 degrees.

I, Absolutely, Hate, The, i9 14900k. Worst purchase Ive done, too much headache

This is the result of (A 36k score) R23 run

View attachment 192458
Click to expand...

Your VIDs seem high for 57/44/49. Although, temperatures and power look good (other than that one P-Core).
 
Upvote 0 Downvote
Arctucas said:
Your VIDs seem high for 57/44/49. Although, temperatures and power look good (other than that one P-Core).
Click to expand...
AC60 DC80, LLC Mode 7. If I lower it more, and raise the LLC mode, the VID might lower but the VCORE increases under heavy load, thus increasing temperature again.
I think it's the thermal velocity boost that is doing its job here. Whenever the temperatures hover around 80 degrees, it takes every opportunity to boost the preferred core (in my case, Pcore 7, 4 or 5). Although, even limiting all cores to 56, it still throttles pcore 7. I dont understand jack doodoo of the 14th gen i9.
 
Upvote 0 Downvote
MigraineFilms said:
Whenever the temperatures hover around 80 degrees, it takes every opportunity to boost the preferred core (in my case, Pcore 7, 4 or 5).
Click to expand...
I thought you had TVB and eTVB set to enabled? The Intel literature says that your CPU must be below 70C for boosting to occur. The other part of your observation is even stranger, I.e. boosting with a manual core ratio in place. I did’t think that could happen, either.
 
Upvote 0 Downvote
MigraineFilms said:
AC60 DC80, LLC Mode 7. If I lower it more, and raise the LLC mode, the VID might lower but the VCORE increases under heavy load, thus increasing temperature again.
I think it's the thermal velocity boost that is doing its job here. Whenever the temperatures hover around 80 degrees, it takes every opportunity to boost the preferred core (in my case, Pcore 7, 4 or 5). Although, even limiting all cores to 56, it still throttles pcore 7. I dont understand jack doodoo of the 14th gen i9.
Click to expand...

OK. As you may know, I have had some success with setting manual VCore while still using an offset, which keeps my VIDs, VCore, and VR OUT under some level of control. All the while, running LLC 3 and Lite Load Mode 1. AC_LL = DC_LL = 0.010mΩ. However, my board has arguably the best (or second best) VRM on MSI Z790 boards.

Anyway, you might be able to get Intel to RMA that 14900K, since it appears to have an issue with that one P-Core.
 
Upvote 0 Downvote
Arctucas said:
OK. As you may know, I have had some success with setting manual VCore while still using an offset, which keeps my VIDs, VCore, and VR OUT under some level of control. All the while, running LLC 3 and Lite Load Mode 1. AC_LL = DC_LL = 0.010mΩ.
Click to expand...
Yes, yes and yes. I have always been curious about your setup. IMO, you, Vassil_V and Charonme all went off in unique directions. I’ve been watching to see if there’s an obvious winner. Still watching… That’s Raptor Lake for you.
 
Upvote 0 Downvote
Back
Top