Trying to understand EDP OTHER and BD PROCHOT warnings

[Toan]

Hi everybody,
I have an old Lenovo Ideapad 700 15-ISK (i7-6700HQ, intel HD graphics 530, GTX 950M, upgraded from 16 to 32 gb ram, 970evo nvme and 1tb sata hdd).
In these last months I decided to approach undervolting to get lower temps and probably also lower consumptions. I think I achieved quite good results even if I still have some doubts. BTW, this is all thanks to TS and this forum (unclewebb in particular) for all the "support" and explanations found.
Back to the question, I am really struggling to understand random warnings related to ring EDP OTHER and BD PROCHOT on CPU and GPU (I suppose iGPU since according to my understandings dGPU is not shown in TS limits windows).
I have really no clue about BD PROCHOT, but I found something related to EDP OTHER that I want to share.

Trying to recap quickly, hope this list with sub-point will be enough clear:
- Problems are almost similar with stock UV, power plans, etc., so not UV related
- Now it is quite cold, so it seems that I never triggered a really high temperature, so no thermal problems, often I barely hear the fans now
- BDPROCHOT warning is triggered only in idle, really often for both CPU and GPU (sometimes only CPU), but I haven't found a real reason
- EDP OTHER seems to be triggered mainly on idle and seems to be related to iGPU:

- iGPU benchmark without CPU usage triggers EDP OTHER warning but there are differences according to power plans and power overlay modes (Win10 slider):​

- balanced power plan with better performance power mode (set to central position of the slider) triggers the warning and the red flashing error giving 31 fps​

- balanced power plan with max performance power mode (set to right position of the slider) triggers the warning without the red error giving 29 fps​

- high performance power plan triggers EDP OTHER constantly​

- balanced power plan has balanced setting for graphics while high performance power plan has maximum performance setting for graphics (HWiNFO shows also GT fuse limit triggered) and not only Ring: Max VR Voltage, ICCMax, PL4)​

- iGPU benchmark with CPU usage does not trigger EDP OTHER warning and gives 1-2 fps boost (stays around 30-31 instead of 28-29):​

- no matter the plan or the settings, CPU usage stops the warning immediately.​

- the iGPU setting in power plans is somehow responsible of EDP OTHER and trigger also GT fuse limit in HWiNFO

Logging data with TS and HWiNFO was not useful, I wasn't able to figure out anything else. But rarely, HWiNFO reported also IA: Electrical Design Point/Other (ICCmax,PL4,SVID,DDR RAPL).

I think I read almost everything here about this topic, I know that the first suggestion should be to try increase IccMax value but not completely sure which one and moreover, I think VRM is somehow "fragile" in this laptop, and don't want to risk since it is not completely clear to me how Lenovo handles things here, but I would like to better understand the issue and, with the help of more experienced users, potentially find a solution (if there is something to solve).
Moreover, these are the specs that I found on intel documentation (mine in red) and the values are quite respected but I wasn't able to find values for other components:

While about ThrottleStop, these are the values (please, take a look also to the UV, if you have some suggestions, It seems stable but don't know exactly how to handle the SA)

TS main window settings - nothing changed from default

CPU (Core & Cache equal) - only offset voltage changed from default

iGPU - only offset voltage changed from default

SA - only offset voltage changed from default

TS TPL window - nothing changed from default

Thank you for the help and the patience to read all this

 

[unclewebb]

For the core and the cache I always recommend setting IccMax to the max which is 255.75. There is nothing else you can do if that does not fix EDP OTHER throttling or warnings in Limit Reasons.

For BD PROCHOT throttling, clear the BD PROCHOT box on the main screen. I have no idea why some throttling happens. I just know how to try and fix it.

I would also check the MMIO Lock box which is near the top right corner of the TPL window. Checking the Speed Shift box in this window is a good idea too.

Is your -189.5 mV offset undervolt for the core and the cache 100% stable? Try running the ThrottleStop TS Bench 960M test. Stop this test if it reports any errors. If you are stable, great. If not, add some voltage.

No need to read books and Intel manuals. Attach a ThrottleStop Log File if you need help.

 

[Toan]

Thank you for the reply.
I recently tried to dig deeper and stumbled upon that documentation by chance and I thought it makes sense.

About setting IccMax to the max value, I never fully understood, is it "safe" on a laptop? What components will be stressed more? I assume it could also lead to higher temperatures. Should this be adjusted only for core and cache even if the warning happens only on Ring? iGPU and SA values are even lower, are they fine as they are?

I think BD PROCHOT warning (probably it is even not throttling) happens very rarely and for fractions of a second. However, I’m not entirely sure since I haven’t been able to capture either BD PROCHOT or EDP OTHER in the log. The log file never shows when the error happens but when I check the limit windows I see the yellow boxes. That is the only reason why I didn't attached a log file.
BTW, if there is nothing else we can do, I will simply check the box instead of living with it.
About the other check boxes (MMIO Lock and SpeedShift box), until now, I left all as it is, could you briefly tell me why I should check them, what are the changes and the benefits that could get?

I better observed the parameters and I have a question: in TPL window, is normal to have speedshift min value set to 1 while minimum should be 8?

Core and cache seem to be completely stable. Surprisingly, they remain stable even at -205mV, At -210mV, I started to have first errors and some BSODs but I suspect those might also have been caused by a more aggressive undervolt on the System Agent (around -22.5mV). Could it be or this is a wrong assumption and it is related only to core and cache values?
I used TS Bench 960M test running single core (35x) and full (31x) without any noticeable problem. I also undervolted the iGPU just to further reduce power consumption, and it seems stable as well.
I initially thought that that undervolting might help address the warnings, it hasn’t made any difference so far, Same annoying and not easily addressable warnings.
For now, I will keep these stable and "conservative" values, maybe in future I will try to increase UV even more. I know that probably the benefits in going on would be minimal but just for the sake of knowledge and optimization. However, as I said I am not sure how to handle SA and in general IccMax, so I decided to stop.

I know I made a lot of additional questions, sorry, I got carried away. I hope these questions could also help someone else.
Thank you again for the help, I really appreciate and esteem your opinion

 

[unclewebb]

is it "safe"

If you are concerned about safety, do not use ThrottleStop. My only goal is to maximize performance by minimizing throttling. You can set the voltages and Speed Shift Min or Max to any values you like. Setting Speed Shift Min to 1 or to 8 accomplishes the same thing. Either value will allow the CPU to slow down to its minimum speed when required. Setting Speed Shift Max to 35 or to 255 accomplishes the same thing. Either value will allow the CPU to run at its maximum speed when required.

You can check or not check BD PROCHOT. I choose not to check the BD PROCHOT box on the main screen of ThrottleStop.

If a box in Limit Reasons under the CORE column is red, that indicates throttling is in progress. Any throttling should be investigated further. Yellow boxes in Limit Reasons can be ignored if they are not causing a throttling problem while you are using your computer.

I have never used or undervolted a 6700HQ. Only you can determine what undervolt values are best for your CPU. I prefer not to undervolt the System Agent. There is no point in adjusting anything if it can reduce stability without increasing performance in any noticeable or meaningful way.

 

[Toan]

Thanks for the clarifications!

Since the BD PROCHOT warning is always yellow (never red) and happens only rarely, I think I’ll keep it enabled for now and just monitor how it behaves under higher ambient temperatures—maybe in summer, when it could potentially turn into a real throttling issue.

I'm sorry if I was not completely clear. When I say “safe,” I mean that increasing IccMax could put more stress on some components and potentially create less benefits (higher temperatures first of all) and I just want to be sure and think twice about it. Instead, only reducing voltage (obviously step by step and not with insanely extreme values) at worst, can cause instability and BSODs if pushed too far.
I know that it is not dangerous itself but could be if done improperly, I know it is all under my responsibility, I am not too much concerned about safety, I just want to (almost) know what I am doing.
This can also apply to SA. I understand that SA is probably the parameter that affects multiple components, (first of all the memory controller), therefore undervolting could potentially mess with more things and causing instability that could be hard to address without gaining any benefit. I tried UV on SA just to check if this could somehow solve or reduce warnings but so far, it hasn’t made any difference.

My goal is mainly to mantain the actual decent performance, minimizing temperatures and consumption. From what I can tell, throttling hasn’t been an issue so far, and everything seems stable—but I’ll keep an eye on it. As said in previous messages, I’ve already identified a situation where TS shows the red flashing “throttling is in progress” message, so I’ll focus on that.

I’ve read a lot of posts about undervolting on the 6700HQ, and my results seem pretty much in line with others, so I’m happy with where I’m at. I might try disabling the iGPU to see if it helps me understand the warnings better. I still think the iGPU is the main cause of both the errors, but maybe this is just its normal behavior.

I will update this thread if I find anything new, meanwhile any other suggestion is really appreciated.

 

[Calenhad]

BD PROCHOT is a flag that enable other hardware in a computer to trigger the PROCHOT flag for the CPU (BD = Bi-Directional). It would seem some vendors even use this for power management, I guess it is a quick (and imho dirty) way to force a CPU to minimum frequency and keep it there. But it is also meant to reduce CPU performance if some other component is not working as intended. A bad battery, faulty charging circuit, malfunctioning fan, or overheating GPU are all potential sources for valid BD PROCHOT flags (and many other components depending on the computer)

In my view, the main issue with BD PROCHOT is that it is intended as a safety feature. But misused and buggy in many cases. If it is continually triggered in your computer and you disable it with Throttlestop, you have to take a calculated risk and assume it is triggered for some non-critical reason. Because worst case your computer is about to catch fire, best case the firmware developers are lazy ****** that use it to bruteforce power savings.

 

[Toan]

I have updates!! I kinda solved the EDP OTHER problem, and it’s indeed related to the iGPU.

Short version: I used XTU to tweak the iGPU frequency multiplier, and all the warnings, errors, and throttling disappeared under all conditions.

Detailed explanation (in case it helps someone else):
After some thinking and research, I decided to experiment a bit. I started by messing with Intel drivers, but nothing changed: the iGPU’s boost clock of 1050 MHz was never reached, no matter which driver I tried.

Next, I reinstalled the latest drivers, reset ThrottleStop to stock voltages, and switched to XTU to test if the iGPU's frequency multiplier slider could help me "unlock" the correct boost clock. My goal wasn’t overclocking—just small adjustments (+/- 50-100 MHz) to see what happens.
To my surprise, EDP OTHER didn’t appear anymore, and in HWiNFO, none of the usual alarms were triggered (“Ring: Max VR Voltage, ICCMax, PL4” or “GT fuse limit”). Even better, the iGPU frequencies were correct this time.

However, here’s where it got interesting:
- Resetting stock frequencies in XTU still didn’t trigger any errors.
- When I applied ThrottleStop’s undervolt settings again, the errors came back and the boost frequency dropped back to 1 GHz. Resetting TS undervolting didn’t fix it.

So I decided to apply UV before using XTU, carefully checking that XTU reads the correct values and it still worked without errors and with correct frequencies.

So the procedure I finally found is:
After restarting, apply TS undervolt, close TS, open XTU, made a small change on iGPU frequency multiplier,save, and then revert the changes back (I don't want to OC iGPU), close XTU (also its service).
Every time I change UV values, I need to open XTU and repeat the trick.
With this procedure, undervolting is applied, and there are no errors at all, plus the iGPU frequencies are correct. I don’t know if it’s placebo, but the iGPU also seems snappier when scaling frequencies. Before, it was locked to the base clock most of the time, and I never saw intermediate steps (e.g., 450-600-900 MHz). It felt more like an on/off situation (350-1000 MHz).

I also made other tests with previous drivers and observed the same behavior. Therefore, I am 100% sure nothing else was touched or changed, at least from the GUIs.
So it seems that XTU somehow unlock something, that TS somehow reset even applying only undervolting.
In my case, XTU seems essential as the final step to avoid EDP OTHER, even though I’m not practically doing anything with it.
I don't know if there are alternatives to XTU that can be used or if there is something that can help me permanently fix this without using XTU every time. Please let me know.
Hope this could be useful, fell free to ask if something else is needed.

By the way, this process solves the warnings, but it doesn’t improve performance much—aside from the 50 MHz higher iGPU clock.

Regarding BD PROCHOT:
Thanks for all the clarifications! For now, I’ll leave it enabled as planned, but I noticed that undervolting has affected its behavior. With stock voltages, BD PROCHOT was always triggered for both CPU and iGPU, but now it seems to trigger only for the CPU. I’ll keep investigating, especially under higher temperatures, to better understand when and why it happens.

 

[Valour549]

Straight from an Intel engineer

Do you know what exactly triggers BD PROCHOT?

BD Prochot is a sensor path intel leaves exposed for OEMs to use to throttle the CPU for whatever purposes they need. There is no standard for what they do with it after that other than what data it wants to see. I recently had a post up about a Lenovo P330 Tiny with their BD Prochot on the i7 8700T. Lenovo appears to be feeding it with VRM temps and current passage for a motherboard PCIe slot power circuit.

In a lot of laptops, the OEM will run a chassis temperature sensor somewhere on the motherboard to this point, as there are skin temp requirements for devices like this that they must stay under. Likely this sensor is on the keyboard daughterboard near the CPU or GPU, as those will generally create hotspots. The OEM needs to keep those in spec, and that means throttling the CPU.

I am not sure why the ability to unlock it has changed (many 13th/14th Gen can't unlock it), but it may be down to OEMs locking out the option to ensure a device keeps some safety mechanism in place.

 

[Toan]

Thank you all for the support and explanations. Hopefully this thread could be helpful to someone.

Lenovo appears to be feeding it with VRM temps and current passage for a motherboard PCIe slot power circuit.

This seems to be my case, or at least, the most plausible.

I tried to verify all these suggestions, I don't think it is related to keyboard sensors, the chassis is really cold sometimes
Could be VRM temps, don't know how to prove it.
Maybe sometimes the additional NVME disk requires too much power, maybe sometimes the dGPU.
I don't know exactly, but I think that this could be the way since I noticed that with wifi card enabled the warning seems to appear more often.

Now I am quite happy with UV, solved somehow the main problem (EDP OTHER warning), this is just out of curiosity.
BTW, I notice another little strange behavior: UV does not reset every time while rebooting laptop, but only if I disconnect the charger. Seems that the registers keep their values. Quite convenient

Thank you all again and have a nice Christmas