help with ox129 bios update

[webworm4156802d9]

Hi! Hope someone can help me with novice questions. According to what I have read Intel has identified my I7-14700F processor as needing the 0x129 Bios update to address
excessive voltage issue. I have downloaded Bios update, extracted files, and copied Bios folder to a normal (non-bootable) USB drive. Question 1. Do I insert USB drive in port when
machine is off, start machine, press delete to enter bios setup then select MFLASH and proceed from there per instructions? OR do I insert USB drive at some other point in
the procedure? Question 2. Is the ME FW update required as part of the Bios update? If so should ME FW update be done before or after the Bios update? or can I skip the ME FW update
entirely? sorry for the stupid questions but I'm a beginner. also new to forums!

 

[sherpa25]

A Noctua NH-D15S chromax.black? Wow, it should do better than this! What is going on. Can you show a photo of your system?

Yes. Built by our vendor near my location where I've been getting PC stuff. Photo of internals (case opened)? Pretty basic though...will try take a photo, as its behind my monitors and tight to access.

 

[sherpa25]

Here it is. Also included latest CB run...result 18125. Let me know if there's anything else. Thanks.

Yes. Built by our vendor near my location where I've been getting PC stuff. Photo of internals (case opened)? Pretty basic though...will try take a photo, as its behind my monitors and tight to access.

 

[citay]

citay, CPU Lite load mode 8, CEP disabled. Ran both OCCT and Linpack for 20 min. Same settings as detailed in previous post. No Errors detected. OCCT CPU Package and Core Max temp 88C. Linpack CPU Package max temp 80C Core max temp 83C. Cinebench=32223 see Hwinfo attached. does everything look ok?

Yes, looks ok. If you don't want to do additional testing, you could keep it like this. If you're in the mood for more testing, reduce the CPU Lite Load mode by two steps at a time and test for stability (so 8->6 now). As soon as you see instability or weird behaviour, let's say at Mode 4, you'd raise by one step (4->5) and re-test. Then, once you find the first step that seems fully stable (let's say 5), you'd go up by another step (5->6) for added stability headroom.

 

[citay]

Yes. Built by our vendor near my location where I've been getting PC stuff.

They overdid it a bit with the fans. The three fans on top of the case don't really do much good, i mean, we see it from the cooling performance. I still don't know what cooler the OP has (webworm), but presumably a smaller one than your mighty Noctua (i have the same, but with two fans), yet his system is doing fine with 170W power limits.

So i would disconnect, if not take out the top mounted fans completely, or maybe just leave the rearmost one. You want to have a defined airflow from front to rear, with a slight bias from front bottom to top rear. Air always takes the shortest path, the path of least resistance. So what do they think is happening when they plaster the case with fans like this? I'll tell you what. You have an underpressure inside the case, from there being more exhaust than intake fans. Furthermore, the close proximity of intake and exhaust fans means the air will be pulled in and pushed right out again, on the whole front upper area. The airflow isn't well defined at all. Only a fraction of the cold air makes it through the CPU cooler.

Now let's look at it with the two top front fans removed/disabled.

Suddenly there's a clearly defined airflow without any cold air being blown out again right away! It can first pass through the CPU cooler and then the rear fan can exhaust the warmed up air. It's even debatable if you need the rearmost top fan, this would have to be checked with a CB run for CPU temperature. And like i said before, check the fan curves too. This cooler, in a roomy case like this, should be able to deal with 170W while staying below the 90°C range.

 

[webworm4156802d9]

will try testing at mode 6 later tonight. Please give me your opinion on my fan curve questions below. When I went to adjust system fan curve the page in bios showed that DC mode was selected (no smart fan mode). I grabbed the pink ball and moved it up from 7.2 volts and saw that the fan increased speed. I switched to PWM and grabbed the ball and moved it up and saw that the speed increased so I concluded that both methods controlled the fans. I left the PWM on and selected smart fan mode. I adjusted the temp trigger point and percent of fan by grabbing each of the four balls on the fan curve plotting graph and moving them accordingly. The color-coded summary to the left of the graph window showed the values for each of the 4 (levels) points on the graph. I exited the bios. The confirm changes window appeared but not all the changes I made to each of the levels was listed so I clicked no and returned to the graph window and redid the settings. The same issue appeared upon attempting to confirm the changes. I tried to redo the settings a few times by holding the mouse on the balls and clicking enter, also entered the values manually in the summary to the right and clicked enter. Each time the confirm/save settings changes window did not reflect all of the changes to each level that I had made. I finally confirmed the changes that were being displayed and exited. TWO CONCERNS: I noticed that the 4 balls plotting the graph did not exactly match the color-coded temp trigger points/ percent of fan values that were displayed in the summary to the right of the plotted curve in the graph. But MORE IMPORTANT, the changes that were confirmed did not show all the temp triggers and percent of fan values. See the attached screenshot. The bios changes that were presented for confirmation before exit were: Level 1: both temp and fan percent were confirmed. Level 2: (no temp confirmed) but the fan percent was confirmed. Level 3: (no temp confirmed) but the fan percent was confirmed. Level 4: temp confirmed but no fan percent was confirmed. Similar thing happened when adjusting Pump fan, not all changes were confirmed. WHAT IS THE EXPLANATION FOR THIS? DID I MAKE THE CHANGES CORRECTLY

? I'm left wondering if all the values are in place to control my fans.

 

[citay]

I grabbed the pink ball and moved it up from 7.2 volts and saw that the fan increased speed. I switched to PWM and grabbed the ball and moved it up and saw that the speed increased so I concluded that both methods controlled the fans.

Yes, any 4-pin (PWM control) fan can also be DC-Voltage-controlled, it's just that the PWM control is superior.

I left the PWM on and selected smart fan mode.

Good.

I'm left wondering if all the values are in place to control my fans.

Might just be a display issue in the BIOS, but it sure looks like a decent fan curve. With your powerful cooling, you can leave the 100% PWM signal at 90°C instead of 80°C. And try altogether lower points for the other three, i see that you still have 900-1100 RPM there for idle, this is a bit too much. Ideal would be 500 RPM or less, if your fans allow for it. Just a tiny bit of air movement in idle, but minimal noise and no unnecessary dust buildup. For example, i will give you my values, for one system fan i have: 35°C: 24%, 50°C: 32%, 75°C: 60%, 90°C: 100%. You can even try the lowest point at 20%, with a lot of fans this results in some of their lowest RPM.

Then you can see if the noise is more to your liking, and then we can tackle the rest.

 

[webworm4156802d9]

ok. so, I guess I'll just assume all the settings are in there as displayed in the color-coded summary regardless of what the "confirm changes" dialog box says upon exiting. I'll post the hwinfo after OCCT/Linpack testing of the mode 6 change in while, before I mess with the fans. I will say that I find myself constantly checking the system fans visually to make sure they are working because I don't really hear them at idle.

 

[sherpa25]

They overdid it a bit with the fans. ..........

Thanks for the valuable info. Seems to have improved after I removed the 2 fans at the top (front and mid), and still have the rear one, for testing.
Attached shots of HWInfo after CB run (22180 on 2nd run). Can't I get this to perform any faster?
Also attached my fan curves, for reference. Not sure what voltages to be optimum. Thanks.

UPDATE: To try improve, I tried lowering CPU SA to 1.22 and CLL to Mode 5 (which I had before the BIOS update to 1J). Result is last HWinfo pic with CB=22929.
Will try later increasing to Mode 6 or 7 to test.

 

[webworm4156802d9]

ok. No errors at mode 6. Cinebench=32579 please see attached HWinfo. THREE Questions: 1. Everything ok in HWinfo? 2. I never did anything with the CPU Current Limit (A) mentioned in your guide. Should I set any limits there? If so, would setting limit now require redoing anything I've already done? 3. I noticed something about core power limit exceeded, package ring power exceeded and ring limit reasons exceeded on HWinfo (I think the only yeses on the charts). What is that about?

 

[citay]

Man, you guys have me feeling like a juggler here, having to deal with two different setups in the same thread

Also attached my fan curves, for reference. Not sure what voltages to be optimum. Thanks.

You can see that the CPU fan behaves quite differently from the system fan, if you look at the RPM. So for the CPU fan, this curve is pretty good at 550 RPM in idle, maybe set the 100% PWM point at 85°C or 90°C, to limit the noise output slightly. For the SYS_FAN2 you can go right to 20% at 35°C, or even try lower PWM values, see if it shuts off below 20% or if it goes to lower RPM (depends on the implementation they went with).

For SYS_FAN3, the only voltage-controlled one (pointing to a 3-pin fan), you will have to check its minimum turn-on voltage. Drop the voltage all the way down, then try something like 3.5V and see if it starts to spin at that. Because to overcome the fan motor's resistance for the spin-up, the voltage-controlled fans need a certain minimum voltage. Once it's spinning, you could theoretically go lower with the voltage again, since the resistance doesn't have to be overcome again, but then on the next cold boot it won't start spinning until the voltage exceeds the minimum voltage (requiring higher CPU temperatures to trigger that). So that's the downside of 3-pin fans, they sometimes can't go to such low RPM as PWM-controlled fans.

hanks for the valuable info. Seems to have improved after I removed the 2 fans at the top (front and mid), and still have the rear one, for testing.

Yes, things notably improved! Not so much in the absolute temperatures, but the CPU is boosting much higher now because it sees additional headroom, leading to improved scores. Then once you lowered the mode for CPU Lite Load, additional improvements ensued.

Result is last HWinfo pic with CB=22929.
Will try later increasing to Mode 6 or 7 to test.

If mode 5 is stable, why try a higher mode next? You would try a lower mode to see if it's also stable. If it is, only good things can happen.

 

[citay]

ok. No errors at mode 6. Cinebench=32579 please see attached HWinfo. THREE Questions: 1. Everything ok in HWinfo?

Looks good. Voltages further reduced, temperatures further lowered, performance further improved. What's not to like.

2. I never did anything with the CPU Current Limit (A) mentioned in your guide. Should I set any limits there? If so, would setting limit now require redoing anything I've already done?

For you, you can set 307A there. It's just an additional "insurance". The maximum allowed current ever should always be 400A, but that's more for people who have set 253W power limits. For you, 307A should be good in that it's high enough not to intervene, but low enough to prevent trouble in the worst case. Nothing special that it requires. You would just check if it limits things more than the power limits (which then you'd raise the current limit a bit). But here it should stay out of the way, with 170W power limits.

3. I noticed something about core power limit exceeded, package ring power exceeded and ring limit reasons exceeded on HWinfo (I think the only yeses on the charts). What is that about?

That's all pretty standard stuff if you set limits in the BIOS. Once the CPU hits the limits, it tries to observe them, and the "yes" bit is triggered, obviously. The one you want to avoid is the "thermal throttling" bit.

 

[sherpa25]

Man, you guys have me feeling like a juggler here, having to deal with two different setups in the same thread

LOL . Why I asked in my initial post if I could chime in and solicit here, or just create a new thread. Really wanted help from an expert/experienced engineer before I messed things up.

I'll try out the fan settings suggestions.

For my suggestion to try increasing the Mode, this is what I understood from your other guide ('How to set power limits' ) where you said "Once you find the lowest stable mode (say if Mode 4 is stable), i recommend to actually set it one mode higher than that (Mode 5 in this example), to have stability headroom and not be on the edge of stability.....". So if all stable and better now at Mode 5, I thought I'd try with Mode 6 or 7. Did I understand it wrong? Oh, let me know if it's better to make a separate thread. Again, sorry too @webworm4156802d9 for hijacking the thread. Better for me if here though, as I learn more from the additional setup

 

[citay]

LOL . Why I asked in my initial post if I could chime in and solicit here, or just create a new thread. Really wanted help from an expert/experienced engineer before I messed things up.

Well, it's alright, we have managed to handle it so far. Let's keep going now until everything is optimized.

So if all stable and better now, I thought I'd try with Mode 6 or 7.

That's after finding the lowest stable mode first. In other words, yes, if you found that Mode 5 is the lowest stable one (meaning, at mode 4 you got the first instability), then yes, you'd set to Mode 6, and that would be it, done. But if "below mode 5" is an untested unknown, you don't just assume it becomes unstable somewhere below, you'd keep testing. Who knows, maybe Mode 3 is even stable. If you don't test, you don't know, and then there is surely no need to raise the mode. Raising by one can only be done when you actually found the lowest stable setting.

 

[webworm4156802d9]

citay, CPU Lite Load on Mode 4. No Errors after 20 min of OCCT and Linpack. Cinebench=33012. Please see HWinfo attached. THREE QUESTIONS: 1. everything ok on HWinfo? 2. I checked the Current Limit. It's on 280A. (I never touched it or checked it before). Can I leave it on 280A? Is it negatively affecting anything we have been doing? 3. Is the OCCT and Linpack testing adequate? Is 20 min. long enough? Are the default settings ok? (IE. should I select "extreme" for mode on CPU instead of normal? Should Linpack memory be set higher than 2048?) I guess thats more than 3 questions!

 

[citay]

THREE QUESTIONS: 1. everything ok on HWinfo?

Yep, still lower voltages and still gaining performance. You should be starting to anticipate what answer you will be getting, because as long as the lowered mode is stable, it will improve just about everything, without any downsides. The only concern to have from undervolting is about stability.

2. I checked the Current Limit. It's on 280A. (I never touched it or checked it before). Can I leave it on 280A? Is it negatively affecting anything we have been doing?

I don't know where they pulled 280A from (it's not in any Intel recommendations), but it's high enough not to interfere, so yeah, leave it at that, same difference at these power limits. Only with higher power limits (if you had more powerful cooling) would it eventually kick into action.

3. Is the OCCT and Linpack testing adequate? Is 20 min. long enough? Are the default settings ok? (IE. should I select "extreme" for mode on CPU instead of normal? Should Linpack memory be set higher than 2048?) I guess thats more than 3 questions!

Well, you can also run a bit of Prime95 Small FFTs torture test, moving the mouse every once in a while to cause a transient load. 20 minutes of that should be plenty. You can also run this old Cinebench R15 version, you will notice that it will finish very fast with a modern CPU, so run it at least a handful of times in quick succession. Keep HWinfo open and see if you get any WHEA errors (the very bottom entry on the very right). You can just play games or do other demanding stuff. Anytime you notice an odd behaviour you never had before, like a crash-to-desktop from a game, let alone any error in any stress test, and you should raise the CPU Lite Load mode by one and re-test.

 

[sherpa25]

Hi @citay. Just an update after trying lower Modes 4, 3, and the latest now, at 2 (pics attached). Performance and temps bit better from 4 to 3, but then went back/slightly worse at 2. What do you suggest next? And should I just go back to Mode 3? BTW, was my changing of CPU SA to 1.25V ok (from Auto)?

Update (4th pic): Put it back to Mode 3. Now tried running OCCT 20mins, and with max temp reaching 98deg, should I be worried (since this is more of a stability test)? What could I do next?

 

[webworm4156802d9]

Thanks. Should I move down to CPU Lite Load mode 2 and continue to use the current test procedure (20 min. OCCT and Linpack) until I find the first unstable mode then move up until I find the first stable mode, finally settling on the mode that provides one mode of headroom? AND THEN double check for stability using Prime95 or CinebenchR15? OR.... should I stop now at mode 4 and run Prime95 and CinebenchR15 before proceeding further?

 

[citay]

Hi @citay. Just an update after trying lower Modes 4, 3, and the latest now, at 2 (pics attached). Performance and temps bit better from 4 to 3, but then went back/slightly worse at 2. What do you suggest next? And should I just go back to Mode 3? BTW, was my changing of CPU SA to 1.25V ok (from Auto)?

At around 23K points, 200 points difference is nothing, not even 1% difference. If you do another run with Mode 2, you could get the same difference as an increase over Mode 3 instead of a decrease. This is completely within the margin of error (2%), depending on what Windows was doing in the background etc.

This <1% performance difference (therefore deemed identical) happens because we have now reached a point where the lowered CPU Lite Load mode resulted in low enough voltage to make the "CPU Package Power" (power draw of the CPU) fall below the power limits that we set for you. Even mode 5 was already slighly below 170W. If you check Mode 4, it's at 165W, so the power limits don't even come into play. Whatever you do from here will not improve performance within the power limits anymore (as it doesn't reach them anymore), but it will still improve the efficiency further, as you now simply get a lower power draw for the same performance. So we see Mode 2 has another 9 Watts less and 3° less.

The fan curves still have some room for improvement. Lowering CPU SA voltage is good, i tend to do that as well, usually the BIOS auto values are a bit excessive. If your memory system is stable at a lower voltage there, it's good. You can run the OCCT RAM test.

 

[citay]

Thanks. Should I move down to CPU Lite Load mode 2 and continue to use the current test procedure (20 min. OCCT and Linpack) until I find the first unstable mode then move up until I find the first stable mode, finally settling on the mode that provides one mode of headroom? AND THEN double check for stability using Prime95 or CinebenchR15? OR.... should I stop now at mode 4 and run Prime95 and CinebenchR15 before proceeding further?

Whatever is more to your liking. I would probably be brave and try Mode 2 straight away. This way you don't waste time, because if Mode 4 is stable even in the other tests, you'd lower the mode anyway and repeat the testing. So give it Mode 2 right away and see how stable it is.

 

[sherpa25]

At around 23K points, 200 points difference is nothing, not even 1% difference. If you do another run with Mode 2, you could get the same difference as an increase over Mode 3 instead of a decrease. This is completely within the margin of error (2%), depending on what Windows was doing in the background etc.

This <1% performance difference (therefore deemed identical) happens because we have now reached a point where the lowered CPU Lite Load mode resulted in low enough voltage to make the "CPU Package Power" (power draw of the CPU) fall below the power limits that we set for you. Even mode 5 was already slighly below 170W. If you check Mode 4, it's at 165W, so the power limits don't even come into play. Whatever you do from here will not improve performance within the power limits anymore (as it doesn't reach them anymore), but it will still improve the efficiency further, as you now simply get a lower power draw for the same performance. So we see Mode 2 has another 9 Watts less and 3° less.

The fan curves still have some room for improvement. Lowering CPU SA voltage is good, i tend to do that as well, usually the BIOS auto values are a bit excessive. If your memory system is stable at a lower voltage there, it's good. You can run the OCCT RAM test.

So it appears Modes 2-5 are within the power limits.

In what ways can the fan curves improve, lowering? So I can try adjust them.