Optimal settings Z790 Tomahawk Wifi Max i9 14900k

M

MigraineFilms

Astral Fridge Magnet
Joined
Jul 24, 2024
Messages
317
Hi all, as I was filling other threads with questions, here is a dedicated thread to my question.

Can you help me getting my i9 under control, it keeps thermally throttling even under very low power settings, and I dont fully grasp the undervolt settings.

My system:
i9 14900k
MAG Z790 Tomahawk Wifi max
Deepcool LT720 AIO
Contact Frame
DDR5 Trident Z 6400
Samsung 990 Pro
RTX 4080 Super
Corsair Shift RM1000x

So far I've done this:

Lowered PL1 and PL2 to 175 watts
Lite Load 9 (AUTO)
Voltage adaptive/offset -0.125 (screenshots still say -0.110, changed this, still throttles)

It can run for hours perfectly fine, benchmarks all doing great, cinebench R23 multiple runs, Prime95 30 minutes heat/stress test and then when I take a screenshot, HWiNFO says its thermally throttling without showing any high temperatures. HWiNFO also refreshes slowly, even on 1000ms, since the latest update.

Any advice is much appreciated

P.S: I do not experience any instability in the past 6 months I owned this computer, besides a sporadic lag in the audio, there are no crashes, app hangs, game failures, BSOD's or any of that kind.

Edit: All the information is a bit overwhelming and I'm feeling I'm getting lost in all the numbers a bit
 

Attachments

  • BIOS1.jpg
    BIOS1.jpg
    218.1 KB · Views: 2,198
  • BIOS3.jpg
    BIOS3.jpg
    208.8 KB · Views: 2,139
  • BIOS4.jpg
    BIOS4.jpg
    229.2 KB · Views: 1,705
  • System1.jpg
    System1.jpg
    353.6 KB · Views: 1,552
  • System2.jpg
    System2.jpg
    337.5 KB · Views: 1,516
  • LOADThrottle.jpg
    LOADThrottle.jpg
    2 MB · Views: 1,661
  • IDLE.jpg
    IDLE.jpg
    1.4 MB · Views: 2,096
Sort by date Sort by votes
MigraineFilms said:
Un LLC più alto causerà più vdroop, ma poi diventerà rapidamente instabile e mi darà tensioni di carico medio elevate (in questo momento, non supereranno al massimo 1,36)
Click to expand...
seeing your screenshot confused me even more...
Actually the temperatures are now closer to those of my test (I had understood instead, from your previous post, that you had reached 89°).
But why the hell does HWINFO detect more sensors on your MOBO than on mine?!?!?
On mine I don't see 3VSB, AVSB, VTT, VBAT and VREF... And yet it is exactly the same MOBO
 
Last edited:
Upvote 0 Downvote
giggettoma15af02e3 said:
seeing your screenshot confused me even more...
Actually the temperatures are now closer to those of my test (I had understood instead, from your previous post, that you had reached 89°).
But why the hell does HWINFO detect more sensors on your MOBO than on mine?!?!?
On mine I don't see 3VSB, AVSB, VTT, VBAT and VREF... And yet it is exactly the same MOBO
Click to expand...
The temperatures are 89 when PL is set to 253, it is 75 when set to 230, without any noticable performance loss.

I dont know about the sensors, in all honesty: I never look at those because I do not know what they mean. Just look at vid, vcore, power and amps
 
Upvote 0 Downvote
Well, the new core ultra 285 is slower in premiere, faster in 3d rendering, slower in games. Seeing as I use my pc for premiere / after effects and games, the core ultra does not add any benefits to my workloads. Glad I spend my time finetuning my rig
 
Upvote 0 Downvote
MigraineFilms said:
Well, the new core ultra 285 is slower in premiere, faster in 3d rendering, slower in games. Seeing as I use my pc for premiere / after effects and games, the core ultra does not add any benefits to my workloads. Glad I spend my time fine tuning my rig
Click to expand...
Holy hand grenade, Batman! I just spotted TechPowerUp's CPU ranking when using a RTX4090 at 1440P. The i5 struggles to match the 13600K in gaming and the i7 and i9 both get beaten by the 13700K. I guess the rumors I heard earlier this week were true. Gaming performance is likely impacted by higher than anticipated latency, as well as an "odd" decision to run the Ring a lot slower than Raptor Lake. It's not a great day for "Team Blue".
 
Upvote 0 Downvote
FlyingScot said:
Holy hand grenade, Batman! I just spotted TechPowerUp's CPU ranking when using a RTX4090 at 1440P. The i5 struggles to match the 13600K in gaming and the i7 and i9 both get beaten by the 13700K. I guess the rumors I heard earlier this week were true. Gaming performance is likely impacted by higher than anticipated latency, as well as an "odd" decision to run the Ring a lot slower than Raptor Lake. It's not a great day for "Team Blue".
Click to expand...
Its a killer in 3d rendering and encoding, but 3d will often be GPU bound and rendering is only a fraction if the workflow after hours and hours of editing. AMD beats the Core Ultra on every level and the 13th/14th gens are considerably faster in gaming, albeit at 30 watt more power usage.
During 3d rendering and encoding, the core ultra sucks up 285 watt to 370 watt. Jesus
 
Upvote 0 Downvote
MigraineFilms said:
During 3d rendering and encoding, the core ultra sucks up 285 watt to 370 watt. Jesus
Click to expand...
That's interesting because I just listened to Steve at GamersNexus say that some of the power for Arrow Lake CPUs comes from the 24-pin motherboard connection, which makes accurate power readings very difficult to determine. It could also impact the accuracy of HWInfo power readings.
 
Upvote 0 Downvote
FlyingScot said:
That's interesting because I just listened to Steve at GamersNexus say that some of the power for Arrow Lake CPUs comes from the 24-pin motherboard connection, which makes accurate power readings very difficult to determine. It could also impact the accuracy of HWInfo power readings.
Click to expand...
Toms Hardware has some power draw graphs, they are not as hungry as raptor lake, but the extreme profile in the BIOS has a pl2 of 285 watts, and it will use it if neccesary.
 
Upvote 0 Downvote
FlyingScot said:
Hey, sign me up, mate!
It’s a 14700K all the way!
Click to expand...
As more reviews come in, they all draw the same conclusion:
Arrow lake is only faster in actual rendering workloads. This is cinebench, blender, handbrake en media encoder. These are applications that already work insanely faster on GPU's (albeit in slightly lower quality).

When editing video, power point presentations, office applications and games, its vastly slower than raptor lake (let alone AMD).

Also, with my undervolting and tuning, I reach the same power draw and temperatures as the ultra 285k out of the box has. Even in cinebench renders, the 285k draws 250 watt
 
Upvote 0 Downvote
MigraineFilms said:
As more reviews come in, they all draw the same conclusion:
Arrow lake is only faster in actual rendering workloads. This is cinebench, blender, handbrake en media encoder. These are applications that already work insanely faster on GPU's (albeit in slightly lower quality).

When editing video, power point presentations, office applications and games, its vastly slower than raptor lake (let alone AMD).

Also, with my undervolting and tuning, I reach the same power draw and temperatures as the ultra 285k out of the box has. Even in cinebench renders, the 285k draws 250 watt
Click to expand...
I think I‘m starting to get a better idea of the upsides to this radical new design by Intel. But it looks to me like it‘s going to take at least one more generation (the 3rd gen) to see this approach pay off for gamers and developers alike. Unless, of course, they get distracted by chasing better and better A.I. performance (the current Gold Rush) at the expesnse of everything else. Hopefully, we can have both.
 
Upvote 0 Downvote
Here I am again. In the meantime I have done other tests thanks to your suggestions. Currently I have the latest bios release on the MOBO and I have made changes to the following items:
aggressive fan curves
PCI_E1 Gen mode GEN5
CPU PCI Lanes configuration x16+x0
CPU LITE LOAD 7
CPU over temp 95
CPU Fast throttle 95
IE CEP Support disabled
IE CEP Support fo 14 disabled
CPU Load Line Calibration 7
CPU core voltage mode Adaptive+ offset
CPU core voltage offset mode -
CPU core voltage OFFSET 0.060

All other options are set to default.
With this set the system is stable, during the stress tests I reach the max temperature of 70 °, I have no thermal limitation, but only voltage and the benchmark scores are good. The max voltage, read on the Core VIDs voice, is 1.355 V (I don't know if it's the correct voice to read the voltage for the CPU, but I took the voice with the highest value).

Now in an attempt to improve things I tried to increase the offset from 0.060 to 0.070 and I got a very strange result. The cinebench scores for both r23 and r24 increased, but, unlike what I expected, the temperature also increased to 82° (especially on core 7). I have no thermal limitation, but as before only voltage. The even stranger thing is that when I tested with CPUZ in multicore I got a higher score than before, but in single core instead the score dropped.
Core VIDs voice, is 1.331 V

Can you explain to me why this anomalous behavior occurs? Why does decreasing the voltage with the offset increase the temperature? Why does it go worse in single core?
In daily use is it better to have more efficiency in multicore or single core?
I hope I don't bore you too much with these questions
 
Last edited:
Upvote 0 Downvote
MigraineFilm, is there a reason you went with “By PWM” vs. “By CPU” for your negative offset? Most people in the survey went with the latter. I’m asking because I’m getting asked why to select one over the other. Also, does HWInfo64 show the IA Offset as 0.000 when using the PWM option?

Cheers!
 
Upvote 0 Downvote
MigraineFilms said:
Hi all, as I was filling other threads with questions, here is a dedicated thread to my question.

Can you help me getting my i9 under control, it keeps thermally throttling even under very low power settings, and I dont fully grasp the undervolt settings.

My system:
i9 14900k
MAG Z790 Tomahawk Wifi max
Deepcool LT720 AIO
Contact Frame
DDR5 Trident Z 6400
Samsung 990 Pro
RTX 4080 Super
Corsair Shift RM1000x

So far I've done this:

Lowered PL1 and PL2 to 175 watts
Lite Load 9 (AUTO)
Voltage adaptive/offset -0.125 (screenshots still say -0.110, changed this, still throttles)

It can run for hours perfectly fine, benchmarks all doing great, cinebench R23 multiple runs, Prime95 30 minutes heat/stress test and then when I take a screenshot, HWiNFO says its thermally throttling without showing any high temperatures. HWiNFO also refreshes slowly, even on 1000ms, since the latest update.

Any advice is much appreciated

P.S: I do not experience any instability in the past 6 months I owned this computer, besides a sporadic lag in the audio, there are no crashes, app hangs, game failures, BSOD's or any of that kind.

Edit: All the information is a bit overwhelming and I'm feeling I'm getting lost in all the numbers a bit
Click to expand...
The latest 0x12b microcode BIOS updates MSI released lately have really helped keep the voltages down and has really made overclocking simple.
My BIOS settings were simply to lock all cores at 5.5 (55 as per how you’d enter it in the BIOS), and x5600 speeds for my RAM using “Memory try it” but that because I’m running 4 sticks of RAM on a dual channel board. I also locked my long and short duration limits at 253w with AMPS locked in at 307. No voltage spikes and my core voltage as well as my VIDs stay well under the 1.25v range under load.

DayZ and DCS are some of the most demanding and fickle games that use the CPU much more than they need to so those games would be a good platform to test them on. If you can run those with no issues you should be able to run almost any game.

DM me if you want help making sense of all this.
 
Upvote 0 Downvote
necky4157702d7 said:
The latest 0x12b microcode BIOS updates MSI released lately have really helped keep the voltages down and has really made overclocking simple.
My BIOS settings were simply to lock all cores at 5.5 (55 as per how you’d enter it in the BIOS), and x5600 speeds for my RAM using “Memory try it” but that because I’m running 4 sticks of RAM on a dual channel board. I also locked my long and short duration limits at 253w with AMPS locked in at 307. No voltage spikes and my core voltage as well as my VIDs stay well under the 1.25v range under load.

DayZ and DCS are some of the most demanding and fickle games that use the CPU much more than they need to so those games would be a good platform to test them on. If you can run those with no issues you should be able to run almost any game.

DM me if you want help making sense of all this.
Click to expand...
That MigraineFilm post you replied to was back in early August when he created this thread (at my suggestion) as a placeholder for all things to come. Since then he has tested at least 278 different combinations of BIOS settings - more than any human should ever subject himself to. The poor guy needs a wee resty, me thinks. ;)
 
Upvote 0 Downvote
FlyingScot said:
That MigraineFilm post you replied to was back in early August when he created this thread (at my suggestion) as a placeholder for all things to come. Since then he has tested at least 278 different combinations of BIOS settings - more than any human should ever subject himself to. The poor guy needs a wee resty, me thinks. ;)
Click to expand...
😂 I hear you, I need one too. I’ve been troubleshooting this MFer since Feb 2024. I’m done for now and with hyperthreading disabled 99.99999% of my problems are gone. Hope I didn’t jinx it.
 
Upvote 0 Downvote
Back
Top