Really NEED Undervolt guide for 14700k+MSI z790 mobo

A

AndyTheGreat

Member
Joined
Jan 31, 2023
Messages
148
Hi, community. Just bought 14700K for my MSI z790-P Wifi DDR5. It burns like hell. I get 99-101 in Cinebench 24 and 102 in AIDA, around 99 in Premiere Pro and After Effects. Did not even try other software since it's obvious my DeepCool LS720 (promised by Tom's Hardware to cope with 13900k's over 300 watts) cannot withstand the 14700k 278 without thermal throttling.
So I either have to get another AIO - and I have no idea which one because there are aio owners saying their particular aio beats the temps of 14900k getting not more than 87 in tetsts and there are ones saying even the best 420 mm aios cannot cope with their 13700k - or undervolt my CPU.
I followed the instructions from some post right here on the MSI forums (cannot find it) and lowered the long term and short term power to 253 watts and also set the Lite Load from the default Mode 9 to Mode 3 and the temps are now ten degrees less. Is it enough? Is there any other method or is this one correct even?
I'm completely dumb in undervolting - those are just numbers in BIOS for me, so I really need some sort of an MSI mobo guide to correct and safe undervolting - step by step with screensots or a video. Please, help.
 

Attachments

  • Снимок экрана 2024-04-03 104026.png
    Снимок экрана 2024-04-03 104026.png
    459.3 KB · Views: 1,815
Sort by date Sort by votes
AndyTheGreat said:
Actually I didn't change anything except for the offset... The rest was Auto. I'mm gonna take all the settings screenshots right now... Oh, yeah, about the custom loop. There are no companies that can give you the service of setting up a custom loop in your PC and later on give the support for it in my city, so it's a dead end. I don't have experience in this either
Click to expand...
Is your PC a pre-built?
 
Upvote 0 Downvote
VCore max. 1.124V, CPU Package temp max. 86°C

VCore max. 1.126V, CPU Package temp max. 86°C

VCore max. 1.406V, CPU Package temp max. 96°C

VCore max. 1.298V, CPU Package temp max. 87°C

What we can also observe, with Cinebench, the load seems to be shifting more to the P-cores compared to OCCT Power test, since we suddenly see a couple P-cores in the 90s in Cinebench, while the E-cores stay in their usual high 70s range. So while the overall power draw remained somewhat similar (same power limits), the P-cores draw perhaps a higher proportion of it during the CB run, getting hotter and causing the throttling. So the CB run is actually more useful for power-limit-adjustments to your cooling capabilities (and that's part of the reason i always use it when looking at HWinfo sensors from people's systems, instead of something like OCCT or Prime95).


AndyTheGreat said:
Btw look in the reasons to limit theres something called PL3 - what is it?
Click to expand...

Package Power Control - 010 - ID:655258 | 12th Generation Intel® Core™ Processors

edc.intel.com edc.intel.com

It's another mechanism to ensure the power limits are not exceeded. Slightly more aggressive than PL2, it will limit how often there are spikes above the power limits (overshoot) by limiting the frequencies "reactively", and then PL4 can't even be exceeded because the frequencies are limited "preemptively".


Arctucas said:
My radiators are external, using a small (5000 BTU) window A/C unit to supply cool air.
Click to expand...

Hehe, Arctucas (no doubt derived from "Arctic" somehow?) again making sure that there's some "cooling performance envy", with - which i mentioned before in another thread - the coolest-running CPU i have ever seen at such a power draw. Even though it's delidded which is an "unfair advantage" if you will. Yeah people, this is not something you can achieve easily. If we take this as an example, we can all pack in and go home. No chance.
 
Upvote 0 Downvote
citay said:
What we can also observe, with Cinebench, the load seems to be shifting more to the P-cores compared to OCCT Power test, since we suddenly see a couple P-cores in the 90s in Cinebench, while the E-cores stay in their usual high 70s range. So while the overall power draw remained somewhat similar (same power limits), the P-cores draw perhaps a higher proportion of it during the CB run, getting hotter and causing the throttling. So the CB run is actually more useful for power-limit-adjustments to your cooling capabilities (and that's part of the reason i always use it when looking at HWinfo sensors from people's systems, instead of something like OCCT or Prime95).
Click to expand...

That's a great observation, CiTay. I think you noticed this behavior before on another thread, i.e. when we saw R23 throttling but Prime95 wasn't. Hmmm...good to know.
I personally like to use R23 for just about everything once I know I have a stable system. On a side note, I'm currently working with R23 to develop a synthetic game stress test for thermal tuning my two PC cases. I have found that you can combine R23 (with specifically restricted thread count) and Unigine Heaven to simulate the way a modern game hits your particular CPU and GPU combo. While they're both running, you can then tune (or plan to tune) your fan behavior - or observe problems with case airflow using incense sticks or other means. That's my plan, anyway. It frees me up from having to actually play the game. And as we noted above, most, if not, all stress tests do a poor job of simulating real-world workloads. Incidentally, one thing about that OCCT Power test is that it's mental in the way it hits both the CPU and GPU. Other than testing the PSU, I don't see it very useful. No other regular usage application maxes out both CPU and GPU at the same time.
 
Upvote 0 Downvote
Unigine Heaven is another useful tool for the GPU for sure, that one i use when i curve-optimize / undervolt my GPU. And it's even used professionally for GPU testing (alongside with stress tests), see this video. The main goal of "OCCT Power" is likely, as the name would imply, to cause the maximum system-wide power draw in order to maximally stress the PSU (if done correctly, this would actually also include transient testing, spikes of power draw usually during load changes, something like an ALT-TAB in and out of a game would also cause). This is then indeed different from a pure CPU-based stress test like Prime95.
 
Upvote 0 Downvote
citay said:
Unigine Heaven is another useful tool for the GPU for sure, that one i use when i curve-optimize / undervolt my GPU. And it's even used professionally for GPU testing (alongside with stress tests), see this video. The main goal of "OCCT Power" is likely, as the name would imply, to cause the maximum system-wide power draw in order to maximally stress the PSU (if done correctly, this would actually also include transient testing, spikes of power draw usually during load changes, something like an ALT-TAB in and out of a game would also cause). This is then indeed different from a pure CPU-based stress test like Prime95.
Click to expand...
Interesting...
 
Upvote 0 Downvote
I should add that the reason I'm building a custom synthetic game simulator is because of this pesky issue we have with not being able to control one or more system fans using the GPU temp. You would think that by now Nvidia and AMD would have worked with the BIOS guys to make this temp available. In today's systems the bloody graphics card is the major source of heat - and heat rises. And by design, the GPU tries to max itself out (wattage wise) in every game and synthetic test - my RTX3080 runs at 350 Watts!! Yikes! Adding to the problem, most modern cases are not well designed to get air to the bottom of the case, which in turn causes all that heat to rise and hit the Chipset, the DRAM sticks, the M.2 drives, and eventually the VRM. Compounding this issue (and the main reason for my homemade game simulator) is that any efforts to reduce the CPU temperature has an immediate impact on the front fan airflow. In other words, the cooler your CPU runs, the hotter your GPU runs. It's quite the balancing act. And making the fan curve more aggressive opens the door to more noise when doing normal daily tasks. That's not a perfect solution, either. I'm sure everyone is dealing with this issue in one way or another. I have lots of ideas, so it will be fun to find out what works and what doesn't. After all, every PC case is different.

MSI are you listening???? Be the first to add GPU temp to your BIOS fan app and I guarantee you that your motherboard sales will triple!! BTW, I just copyrighted patented this suggestion so I expect a small commission with each motherboard sold. Just a wee fee. Nothing outrageous... :beerchug:
 
Upvote 0 Downvote
Yeah, ideally GPU temp should be selectable in the BIOS, but i'm pretty sure the BIOS only reads out the sensors of the Super I/O chip on the board, which naturally only encompasses the board's sensors, not ones on add-on cards. But a GPU-temp-driven fan curve can be set with the help of the program Fan Control in Windows.
 
Upvote 0 Downvote
citay said:
Yeah, ideally GPU temp should be selectable in the BIOS, but i'm pretty sure the BIOS only reads out the sensors of the Super I/O chip on the board, which naturally only encompasses the board's sensors, not ones on add-on cards. But a GPU-temp-driven fan curve can be set with the help of the program Fan Control in Windows.
Click to expand...
Well, they bloody well need to hurry up and add it to that Super I/O chip!!! I mean, don't you think, CiTay? It's not like we're trying to put man on the moon or anything.
Yep. I got the Fan Control exe ready to go as me backup plan. Incidentally, does anyone know if there are any downsides to using this app? I do prefer to do everything in BIOS if I can...as I've only got a humble i5 in working order. The i7 is still in its box.
 
Upvote 0 Downvote
Don't know if today's Super I/O chips offer that possibility already, but you are right, where there's a will, there's a way. So there's no will, evidently.

FlyingScot said:
Incidentally, does anyone know if there are any downsides to using this app? I do prefer to do everything in BIOS if I can...
Click to expand...

As far as i know, once it's loaded as autostart, it takes over the fan control, and once you exit Windows, the BIOS takes back over. No downsides, other than that your BIOS curves get superseded. You can of course set the same curves as in the BIOS as your starting point in Fan Control, and then work from there.
 
Upvote 0 Downvote
citay said:
Don't know if today's Super I/O chips offer that possibility already, but you are right, where there's a will, there's a way. So there's no will, evidently.

As far as i know, once it's loaded as autostart, it takes over the fan control, and once you exit Windows, the BIOS takes back over. No downsides, other than that your BIOS curves get superseded. You can of course set the same curves as in the BIOS as your starting point in Fan Control, and then work from there.
Click to expand...
Okay. That sounds good. I will install it on one of my PCs and give it a go soon. I know JayZ and many other youtubers really like it. And apparently it beats the pants of the motherboard software solutions, including _____ Center. ;)
 
Upvote 0 Downvote
FlyingScot said:
That's it...I quit!!!
If you need a window AC unit to play games on a modern Intel system then I'm buying a Playstation!!
Click to expand...
Heh.

Of course, one does not need it.

However, taking advantage of extra cooling is not unheard of. I see several people using water chillers ($400-$700) in addition to their multiple radiators. It is all about gaining performance.

Also, it has been suggested that at least some of the issue with degradation of the Raptor Lake series is due to excessive heat. So, it is sort of a preventative measure to help prolong the life of the processor.
 
Upvote 0 Downvote
AndyTheGreat said:
Lets just say I there's no way I can buy, install and service it.
Click to expand...
OK, just making a suggestion.

I understand the initial investment can be daunting.

I used air cooling for years before AIOs were available. However, they were always sub-par when compared to custom loops, and eventually will quit, necessitating complete replacement, whereas if a component in a custom loop fails, it is only one part to repair/replace.

Good luck with getting your rig straightened out.
 
Upvote 0 Downvote
AndyTheGreat said:
Guys, nobody said anything about my bios settings I posted on the previous page. Are they ok or do I have to change anything
Click to expand...
Andy, I’m not trying to be difficult, but don’t we need to start over from scratch? You have to ask yourself what the most important objective is to you. Do you want the highest R23 scores or do you want lower temps? Do you want the best possible single threaded performance (perhaps for gaming) or do you want the best possible multi-threaded performance for video editing, etc.? Are you content to let Intel algorithms manage all the overclocking or do you want to wrestle that control away from Intel so you know everything that’s going on at any given moment?

For example, you now have Lite Load Control set to “Intel Defaults”. I was going to suggest that you change it back to “Normal” and start tweaking CPU Lite Load, but then I remembered that you have taken manual control of voltage and frequency. So, that negates messing with CPU Lite Load. At least it does with my generation of motherboard.

You see, the advice that we give you will need to be predicated on what your objectives are. Now, if you want to try both avenues, I.e. see what the best Intel’s algorithms can do (with a little help from you) versus your best manual efforts then that’s okay, too. But that’s going to take many many hours of testing and benchmarking. There are no shortcuts.

The advice that is offered after you decide on your approach is very standardized across all the various MSI forum threads, especially those that CiTay has participated in. He follows a very methodical and repeatable approach, but always with the caveat that any advice given regarding settings that impact frequency or voltage must be thoroughly stress tested by the user…because every CPU+Memory+Motherboard+Case cooling combination is unique. People, including myself, have taken months to find the best possible combination of settings given our goals. In my case, I usually want the best possible all-core performance while keeping temps under 75C (for normal daily activities - not stress tests or R23, etc.). And I‘m willing to sacrifice a little single-threaded performance to achieve that goal.
 
Last edited:
Upvote 0 Downvote
Back
Top