The ThrottleStop Guide

It's the new C7 Interrupt Response Limit feature. Increase it!

If the passive cooling fanless thing is anything like I've read from the MBP2015, the throttling is temperature based. Unfortunately, there isn't much that can be done, apart from reducing the voltages and hoping it takes a little longer to hit throttling speeds.

 

According to the core C states, it seems the CPU's working on the background. Do you see the Package C6 or C7 value changing?
Unclewebb said he set his slider at around 11 or 13. I've set mine at 31 when on battery and 15 on AC. Idk why, It just felt better

 

so what exactly C7 Interrupt Response Limit feature do?

 

Hi, My Vbios unlocked, i want to unlock my tdp of cpu,how can i do this? please help.

 

Hello, I am new here, and I was wondering, my computer has been overheating after playing SkyFactory. ( Modded Minecraft) Also my idling tempature is around 60c and im no expert, but that sounds a little high to me. I have a Dell XPS 15z with an Intel Core i7-2640M. So, What are the best settings for throttlestop? If throttlestop can help with this issue. I have also cleaned out the fan, and am not going to replace anything.

 

This is pure garbage... Uninstall Intels DPTF driver and change settings in the bios so the new errors in the device manager gets fixed. http://forum.notebookreview.com/thr...r-on-dell-drivers-downloads-confusion.681669/
http://superuser.com/questions/9091...al-framework-settings-what-exactly-do-they-do

 

It doesn't seem to really do much though... but it doesn't matter for the HQ chips either.

 

As I said earlier; Garbage. If you use cTDP settings in bios then you can actually forget proper Oc. If you have an Mx processor in an AW laptop then you can use this garbage with Alien fusion. Pointless.

 

Hi. I'd say this is due to how this voltage change are applied...
In real world this is just displacement voltage in absolute points meaning when your CPU at full load it normally use 1100mV (1.1V)
after -0.12 - it run 0.98 which is just 10% less than normal voltage ...
Now compare to low-power-mode ....
CPU can run in idle at 0.6V. After adjustment it run at 0.48 - which is 20% less than normal ....

IMHO - intel could change the way of tuning CPU voltage from absolute values to relative percentage ... then it could be more smart adjust. other way they could allow to set voltage for every possible multiplier as it was done for core 2 duo and older processors.

 

Hey guys, throttlestop prevents throttles, but is there no other option for preventing the same throttles by applying a great thermal paste? I am taking about the throttles that many people recieve while playing battlefield 4, they have also decided to use this software, but will reducing the temps not solve all the issues?

 

no.

Throttlestop prevents unwanted throttle and will not prevent heat-based throttle that is built into the chip (whatever your BDProchot is)

 

Okay thanks for the reple, i actually HAVE DISABLED BD- PROCHOT BUT MY PROCHOT IS STILL ACTIVE, my PHROCHOT is 97 DEGREES CELSIUS, is it safe? i have seen my temperatures reach 99 degrees celsius without throttling, is it dangerous for my CPU? what temperatures are actually dangerous? thanks in advance

 

Are you SURE you've hit 99c without throttling? Your throttle limit maximum is 100c and you WILL throttle at that point. If you go over it will thermal shutdown.

 

100c+ is dangerous for the CPU. Staying that hot for extended periods of times will degrade it. Replace your thermal paste.

 

You assume his system is capable of cooling his CPU even with replaced thermal paste.

 

Been there. My Y50 with an i7 4700HQ throttles at exactly 100 C. Thermal shutdowns at 101. Pretty stupid.

calm down hot-shot. I had my thermal place replaced, and it did only worse because it wasn't that good. Had to buy an Gelid GC Extreme (which is super expensive around here), and apply it several times to only get what I was getting previously with stock thermal paste. It ain't nowhere near as easy as to place it on a desktop CPU, with integrated heatspreader.

It turns out my problem was a severely clogged air filter. By throwing it on the trash bin, I've lowered my temps from 15 to 20 degrees.

 

...............................

You know, I tell people not to buy lenovo. And if only they would see this

 

Ha. Haswell. My Y500's i7-3630QM doesn't thermal shutdown until 105C. And yes I did test this back in the day with factory paste+FurMark+Prime95.

 

Hey, 1st of amazing program, once you understand most things, it's so easy to use!
been doing a lot of reading and watching videos on this.
I'm trying to OC my i7-4810mq. This is what i got so far.
Any tips or insight in the right direction would be great!
It's mostly with the turbo timer, i can't get haha.

 

i have lenovo y50 with 4710hq. could you help to unlock my bios to unlock tdp,multiplier of cpu?

 

i have lenovo y50 with 4710hq. I removed anti dust filter, cpu @ 3.5 Ghz on four core active 65mv undervolt on core and GPU 860m +250 Mhz On core @ 1.2 Volt and +490 MHz on Memroy.
On BF4 multiplayer Sige Map Max Cpu temp: 79 centigrade and Gpu Temp: 75 centigrade. before i remove anti dust filter my Cpu Temp reach 100 centigrade and GPU temp reached the max 93 and then GPU Core Clock slow down.
NOW i bought Liquid Ultra, soon i use it for both gpu and cpu. It seems my laptop to be very cooooool.I want to unlock my TDP and Multiplier of CPU. SomeOne Help MEEEEEEEEEEEEEEEEEEEE!!!!!!!

Solved!

 

Not their fault in my case, actually. I'll make a mea culpa here, I've used this laptop on pretty damn outrageously dusty places, like sitting it a table used for polishing woodwork, with a lot of wood dust, also on a small hanger, where my parents store their stuff, which is also dusty as hell. The air filter was clogged in a matter of two months. As I never had a laptop with a air filter, I didn't think about that, and just kept replacing the thermal paste, unsuccessfully.

Then I noticed that I could see the fan through the bottom of the laptop... but these weren't the fans, it was a dust-print of where the air is pulled. ( https://i.imgur.com/bP8VcFT.jpg)
I removed the air filter, and temperatures dropped around 20c. It still overheats to 95 when turbo boosts at 3GHz and the GPU is at maximum usage, with a 100MHz overclock. But it's a very thin laptop, so I don't really blame them. Originally, it disables turbo when playing games, so it sits at 70 C when plaing at 2.4GHz and the GPU at 65 C, maximum usage with the same overclock. It was really damn dirty ( https://i.imgur.com/7EK4idO.jpg)

If you want to complain, complain about the occasional build quality. They're pretty extreme: When they do it right, it is VERY GOOD (enterprise line), and when it's bad, it's VERY BAD (personal line). A bit of a plastic bit from my Y50 broke when I was putting it in my bag, and it caught the zipper of the bag. The screen also has a terrible issue, where its hinges suddenly breaks after a few months, also cracking the screen (because the hinge is partly screwed behind the screen, and it breaks there) (mine had this, but I fixed by loosening the hinge resistance). It also came with a batch of bad HDDs (mine had a bad one, too). Not to mention the horrendous TN panel display, which shifts colors at the slightest movement of the head (actually, if you fill the display with an color, say red, the top part will look different from the bottom part. That on the normal viewing angle) and has a terrible black (light-gray actually).They seem to have changed this on the 2015 line of Lenovos, by using IPS display. But my S400 and my Y50 are from 2014. Bummer. I managed to reproduce better colors by using a custom profile made by other users for this screen, increasing the "color vibrancy" to 50 and calibrating the RGB levels with an IPS display near, so I could mimic the color precision. Pretty rough, but you should see how was it previously.

"Fortunately", most of that was only present on the Y50 line. My S400 is pretty sturdy and I had an Y500, which from that list, only had the crappy display. Everything else was terrific, specially the build quality (funny to think that was why I had bought the Y50 -- good build quality..)

 

In the "Intel ® 64 and IA-32 Architectures Software Developer’s Manual" that can be found on the net there also are Package C3 Interrupt Response Limit (R/W) and Package C6 Interrupt Response Limit (R/W). Any benefit of playing with them? I haven't read the whole thing.

Also...



(http://i.imgur.com/EpyPmOG.png)

I guess it is calculated as IRL (bits 0:9 = the value we see in TS?) multiplied by the time unit. So the question would be does it always bring latency and if not, in which cases? When it jumps from distant C-states to distant C-states?

 

Actually, since my haswell thermal throttles at 96c, your machine should have been designed with a 95-96c thermal window. Some I've seen 97c on, and I think it might be a bit high since thermal shutdown is 101c but you know...

It's all lenovo.

As for your build quality issues, that's precisely WHY I tell people not to get it. I'll say the same thing I always say to everyone forever: if a machine's chassis cannot handle a processor, DO NOT PUT THE PROCESSOR IN IT. Wanted an i7 in the Y50? Should have made it cool better. The P650SE and the Y50 are similar in size and the former handles overclocked CPU/GPU workloads with a more powerful GPU without reaching thermal limits like the Y50 does. It's lenovo's bad design. If the 2015 models are good, then awesome. If not, well screw it.

Don't get me wrong: I'm basically a Clevo-recommending parrot right now. "Want a gaming laptop? K, you need a Clevo. No, I don't really need to know what you exactly want, the answer is a Clevo. Nobody else has anything better for the price/size points, and the ultrathin laptops suck anyway" is basically what I end up saying. It's true too. But I like competition. If Lenovo wants to bring out a machine that's essentially a P650RE or P650RA and price it $200 less than the competition? Things will get GOOD. But that's if Lenovo does that.

 

Hey! Next time you really tear your Y500 down, could you take a high def pic of a little component behind the MoBo? I've burnt a little SMD transistor on my Y500 and the display has gone berserk, flashing black and white, with occasional bursts of the current frame. I believe that if I change the transistor for the same model, it'll get back to life

http://i.imgur.com/GsBRUwr.jpg
It's that little three-legged component, which the arrow is pointing to. It's in the surrounding area behind the LVDS connector

When I was having temp issues with it (due to a clogged air filter), it was going full steam up to 99 C, but when it touches 100 C, it throttles for a while, which reduces the temps by a crapload, but soon brings the speed back to normal, which in turn takes the temp to around 92 C. This keeps going and going, until it just can't handle it anymore and just hits 101C and it turns off.

Without the air filter, it just keeps around 85~90C (worst case scenario). I've let it running a CPU burn test for 4 hours, and the maximum it reached was 93C (3rd core @ 3.0GHz), which is rather OK, given I'm not going to run the CPU full power for 4 straight hours anyway.
The table turns when the GPU comes in, and the shared heatsink ******** comes to show. A 15 minute stress test is enough to bring it to 95C (3rd core again, funny @ 2.8GHz, TDP throttle), with the card overclocked, running at 1187MHz @ 68 C, against stock 1020MHz. Following Lenovo "recommendation", I've made a profile to disable turbo (thus, run at 2.4GHz) when it reaches 65 C on the GPU (which is the most usual case during games) OR when it's at 93C.

This made it go through an 1 hour RealBench test at 82 C (goddamn 3rd core, @ 2.4GHz), 66C GPU @ 1187Mhz. And truth be said, while not perfect, I must give 'em the props for this feat. It's a slim ultrabook and reasonably lightweight, packing a helluva punch and keeping it cool enough, without the need of a cooling pad. Interestingly enough, even at prolonged periods at 80~90, it didn't feel too hot for the touch, on the keyboard.

If only the plastic and display quality were at least OK..


------------------
Now if I recall correctly, I saw a Clevo when I was about to buy the Y50. I remember that it had a fingerprint scanner between the touchpad buttons pretty well, as I wished the Y50 had too. This caught my eye, but when I looked at it, it was more expensive, didn't look as great as the Y50 and I still didn't know about Clevo, so I was like "what the **** is that Clevo brand anyway"
Can't honestly say I regret buying the Y50, though. It got a ****ty plastic, yeah, but it was much cheaper, and packs a lot of horsepower. Enough to make me get rid of my 4690k + 970.

 

Hi all,
first of all thank you very much unclewebb for this wonderful utility !!

I had similar issue with my Toshiba X70 (i7-4700MQ + GeForce 770M), it always had issues with overheating, it throttles when cpu temp reaches 100C and it reached that temp after few minutes of Prime95. During last weeks the issue become even worst and only 30 seconds of 100% cpu usage were enough to trigger thermal throttling. I was not sure whether the problems were in the design of the cooling system or simply a bad thermal paste, anyway I decided to try and repasted both cpu and gpu with Arctic MX4. It a was big success, now cpu never goes over 80C even when running Prime95 and only when gaming with heavy cpu and gpu usage it reaches 90C. Added bonus: when running on battery with low load fan run very slowly and saves battery.
I use Throttlestop to undevolt cpu to keep higher multiplier for longer time.

I could also overclock gpu but sometimes while gaming with cpu and gpu overclocked the pc turns off suddenly. I don't think it is due to temp since it never goes over 90C for the cpu and 80C for the gpu, however it is possible some other component is overheating or there is some other kind of protection and to avoid damage it turns off the pc. With a moderate gpu overclock it seems to work fine. Does anybody have an idea of what can cause this sudden power off ?

 

I posted these two examples to show what this feature does. First picture shows a C7 IRL setting of 11.



And a slight adjustment of this down to 3 disables the Package C States.



I included this new feature because on my laptop, the bios has locked the Package C State Limit register. Without this feature, the Package C States are always being used whether I like it or not. I know some people that create music absolutely hate the C States so I thought this new control might be useful to them. For people that want maximum runtime while on battery power, perhaps this setting will be useful to fine tune things. The CPU power consumption that ThrottleStop reports might not be 100% accurate but it shows a big difference in those 2 pics. Monitoring Battery Discharge using the Windows Performance Monitor or Battery Bar also shows a significant difference after a small change to C7 IRL.

In order to do some proper testing of this, the CPU has to be able to settle down when idle. When I upgraded from Windows 8.1 to Windows 10, I immediately noticed that some task was now running on my laptop and was keeping it busy when it used to be idle. I gave it a day or two to settle down but it never did. My CPU was running hotter, my fan was louder and I was starting to regret the upgrade to Windows 10. I found and disabled the Windows task that was causing this problem and now when idle at the desktop, my 4700MQ needs to spend less than 0.5% of the time in the C0 state to process all of the background tasks. As I have Windows 10 setup, it is just as efficient as Windows 8 or Windows 7 or even Windows Vista.

Some people are OK with Windows 10 spending almost 10% of the time in the C0 state when they are not doing anything. If ThrottleStop reports an average of 10%, on a 4 core CPU with hyper threading, that average of 10% is equivalent to 1 of the 8 threads spending 80% of the time in the C0 state. For me, that's a problem. That means you have some bloated or poorly written software running on your computer and your battery runtime is going to take a significant hit because of it. Bloated software causes extra heat and reduces performance but no one seems too concerned about efficiency anymore. That's one of my complaints about Intel XTU. Nice graphs but it is decreasing performance.

@GrofLuigi - That is the well documented register that ThrottleStop is adjusting. If I get some feedback about this new feature and it is useful to someone then I might investigate some of the other similar registers. It is useful to me as is but I am willing to add more control of this register if needed. To me it still seems like it could just be a CPU bug. It does not make sense that adjusting C7 IRL slightly would disable all of the Package C States.

@Gianpaolo64 - When overclocking, it is not uncommon for a CPU or GPU to randomly reboot before it reaches the official thermal throttling temperature. Intel Core i CPUs are much better at running fast and hot compared to their previous Core 2 Duo technology so your problem might be your GPU that does not like being overclocked. This is why the guys that are out setting world records use liquid Nitrogen for cooling. Cold CPUs have more thermal headroom. Hot CPUs are living on the edge and will lose stability sooner.

@MigaBoo - Some of the turbo timer and TDP adjustments that ThrottleStop lets you access might be completely ignored depending on the CPU model as well as various other registers within the chipset, etc. that might be limiting what your CPU can really do. For that reason, many of these adjustments are simply trial and error. My 4700MQ scores similar to your 4810MQ in the TS Bench test at least so it looks like you need to bump up those Turbo multipliers to overclock some more. Sadly, both CPUs have the same 47 Watt TDP limit so when flat out, there might not be any significant difference in performance between these two processors. They never mention stuff like that in the fine print.

 

Is it one task, or a couple of them? I'd like to disable them aswell =D

 

Yeah, I'll do it next time I re-paste

 

I seem to recall a couple of things in Windows 10 that didn't impress me. Maybe Windows 10 was not happy with the way I turned off the Auto Update feature and was fighting back by part of Windows getting stuck in an infinite loop trying to force me to comply. No thanks. I will decide how I choose to run my computer. Every installation of Windows is unique. Go into Services and start investigating. You are dragging a boat anchor behind you. Time to find out what it is and then cut it loose.

http://i.imgur.com/nzRt32N.png

 

xD I love you Unclewebb you have such a way with words by calmly saying you hate Windows 10 =D.

 

Thx for the response!
I got it to this so far, with turbo timer/watts, Way more stable than before. BF4 FPS doesn't fluctuate as often like before. CPU doesn't get that hot either. Stays in low 80's.
I'm gonna try for higher soon, but temps worry me for a laptop, but so far very good and impressed. I love the program, so easy to use haha. Was using XTU before and it was a nightmare.

When moving the Turbo Timer Limit, & PPO Turbo Time Limit and adding in the numbers for watts. How should they compare to each other? SHould one be higher than the other? Consistent? or just play around with it?

 

"Unlocked" TDP ftw! I managed to increase the short turbo TDP indefinitely =) No more 47 W for me

 

Nice, I'm assuming you did this after unlocking the hidden BIOS menus?

 

@GTMoraes - ThrottleStop shows that your CPU is still throttling during the TS Bench test. Run the Limit Reasons program to find out what is going on.

The interesting thing is that on my 4700MQ, the TS Bench test stays right at the 47W TDP limit when using an under volt of -50 mV on the CPU and Default VCCIN so there is zero throttling.



The power consumption data that controls turbo boost throttling is based on the VID voltage which seems to be significantly higher in the HQ chips compared to the MQ.

Higher VID = Higher reported power consumption = Less turbo boost

@Papusan is not a fan of the HQ chips for a reason.

 

As @unclewebb said, your window shows you are indeed TDP throttling. Can you check if adjusting the VCCIN back to default has you throttle to the same speed? If it does, then that means that adjusting the VCCIN simply skews the reading of TDP, and doesn't actually change the chip's capabilities. If you throttle further, then it means it can help, but not fix, the throttling of HQ users at the cost of potential stability/lifespan of the chip.

If you think Papusan is not a fan of the HQ chips you should see what I say about them when I'm not on NBR

 

I was playing around with my daughter's Core i7-3517U today and it works the same as your U CPU. The default multiplier for the 3517U is 24 and setting the Non Turbo Ratio to 24 disables Intel Turbo Boost. My 4700MQ does the opposite and uses full Turbo Boost when it is set to the default 24 multiplier.

While playing, I wanted to see if the Non Turbo Ratio setting had any practical value for the U CPU. I used the light weight Heaven 4.0 benchmark and to create some extra load I also ran 1 thread of the TS Bench test. It didn't take long before the U was throttling significantly. ThrottleStop was setup for monitoring purposes only.

http://i.imgur.com/wuhb56Z.png

Sick, sick, sick. The U in this Asus laptop was throttling along at only 997 MHz. Why? There is absolutely no reason for it to be throttling. Power consumption was well below the TDP limit. No worries. Changing the new Non Turbo Ratio feature from 23 to 7 smartened things up immediately. Now the 3517U was running at a more respectable 2500 MHz to 2600 MHz.

http://i.imgur.com/9nyWdtU.png

At this early point in the benchmark, performance was similar. For games that are GPU dependent, manufacturers can get away with some significant CPU throttling. For CPU dependent games or apps, most users that experience this big of a drop in MHz are not going to be happy with their purchase. I have heard that the the 4th Gen U is a big improvement over the previous 3rd Gen U.

I agree that the practical difference between those two settings seems to be so slight that there is no measurable difference between the two.

@MigaBoo - There is no need to adjust the PP0 Power Limit. It is not checked so I think you can just set it to zero and forget about it. Some ThrottleStop settings were intended to give users access to settings that some manufacturer might be using for throttling purposes. If your laptop is not using one of these features, you do not have to change it. There might be more fun in the FIVR window for you. Are you under volting yet to reduce heat and power consumption? There are a few other tweaks in the FIVR window that can improve performance.

 

That's why I said "unlocked" TDP. It still throttles, but it has two TDP limits: One for short runs, which throttles at 53W (the "30 seconds" one) and a long one, 47W.
I can change these values too, but it seems that the processor simply ignores it



And for some reason, the display brightness affects the TDP value... God knows why.

100% brightness:
http://i.imgur.com/dY8qUKt.png

0% brightness
http://i.imgur.com/jtaMOC0.png

Yep! and it takes some balls to change something there, as a bad configuration could lock me out, and there's no way to reset it, other than directly write to the BIOS through an external hardware

 

Yeah, those sliders are pointless because it's a 47xxHQ chip. If it were a 48xx or 49xx chip, your short power could be adjusted as high as the laptop would allow you to take it, but you'd still end up back at 47W after 2.5 minutes max.

Unless you somehow managed to bypass the 2.5 minute limit? I've never seen a HQ chip bypass the 2.5 minute limit without having free TDP usage limits like a MQ chip.

In these pics your throttlestop window looks like you're hovering over it. If you aren't, and this happens sometimes (mousing over anything in the taskbar causes it to look like you're hovering over it even after your mouse has gone) then shift + right click on anything in the taskbar will fix it.

Has nothing to do with throttlestop but it drives me insane whenever it happens on any machine so I might as well tell you =D

 

It completed the TS Bench 1024m in ~290s, always over the TDP limit.
Those sliders were unavailable before unlocking the Turbo Boost locks (it's a setting in the unlocked BIOS). Now I can set the maximum TDP on ThrottleStop and on the BIOS, but the processor simply ignores it.
I can set a huge short boost time frame, however, and that's what I'm showing

 

But you seem to be at a 53W or 51W TDP lock when adjusting your brightness. I wonder if your CPU is in fact locking to 47W, but your iGPU is not being counted in that? It would make sense with the brightness slider, as brightness has nothing to do with a CPU's TDP normally. Your machine has Optimus so it definitely does use the iGPU and that takes power.

 

Previously, when it throttled at 47W, I couldn't even think about getting past 3.1GHz. Now it went through the test at ~3.3GHz.

Also, before unlocking the BIOS, I used to see these TDP bursts for 30 seconds, and it would then fall back to the default 47W TDP for the rest of the benchmark

When the IGP is working, I used to get a maximum of 2.8GHz on the CPU. Now I can get up to 3.1GHz at the same games.
It can't handle the temps at that frequency, though. Not even with a basic cooling pad


The brightness thing is something to think about. The results are pretty consistent.
Perhaps the laptop, as a whole, has a power limiter?
The power brick's rated for 20V 6.75A, so up to 135W. It, the brick, should be able to handle it, but idk about the internal circuitry of the laptop.

 

I'm thinking that whatever you did unlocked the TDP limit to include the iGPU. I know that dGPU-only models like the GT72 will lock to 47W exact, but I've seen iGPU-using laptops at ~49W sometimes. Because it doesn't get to 57W (the short power limit) I know it's still the 47W lock.

The burst is doing what it's supposed to. It probably really is something you did in the BIOS.

Because this is why lenovo sucks =D

It's weird, because when you LOWER brightness the CPU seems to use more power. Unless you messed up which was 100% and which was 0% in your earlier post, because I can't see how 0% brightness uses more power and the CPU clocks slightly faster (I can assume that the CPU clock adjusts up and down though, as your original screenshot had 33.36 and your screenies have bounced between 33.00 and 33.49).

 

The pics are correct. It uses more power, but it also increases the multiplier. It's like if I reduced the brightness, I increased some power headroom, so the processor could draw more power. I don't know anything else I could shut off.. perhaps if I... use the key to completely turn the screen off?

http://i.imgur.com/5NIWhGD.png

Too damn weird.
I really think it's some sort of shared power pool, and the laptop can't get past this X value.

Or not, because it is able to work at high frequencies with the GTX at full power. Hell, I don't know

 

That screenshot doesn't have TDP throttle showing... I think there's indeed a shared power pool with your iGPU. That makes sense to me now. Lowering your screen brightness and such should automatically take load off your iGPU, and that gets transferred to your CPU, and throttlestop doesn't read the iGPU's power as part of the CPU package power, so it seems like the CPU is using more power, but in effect it's actually holding its max limit? That's my theory...

It probably is that there's a hard-coded limit for the CPU in that machine. The M17x R4 has the same issue; their CPUs won't draw over 68W at all.

But since I know for a fact that hacking the BIOS to unlock a CPU's overclocking limits and such can be done as Dufus has done it, it might make sense that you somehow bypassed your CPU's inborne TDP limit (though the board itself has a limit for that package).

 

Package power goes high, the multiplier follows. But it only goes so high until it's throttled, which happens for some reason. The TDP Throttle radio button isn't much precise, and it might light up on 53W, but not at 54W. Go figures.

Like unclewebb once said, don't trust that number. Idk what does this number means, for sure. What I'm sure is that it isn't at the previous 47W throttling, as the frequency has increased to 3.3GHz, but idk why the brightness affects how high can it go before throttling, and why doesn't it reach 58W. Heck, why does Unclewebb's 4700MQ can go up to 3.3GHz with 47W consumption, and mine sips 6W more?
Your IGP theory sounds right but... I don't think the IGP feeds the display backlight. That's another circuitry, the IGP only deals with signal data, I believe.

 

Dufus' Limit Reasons is accurate, though.

Maybe the screen and CPU share the same power? I don't know. All I can tell is that it seems you have a direct limit on power somewhere, and un-loading your iGPU and/or screen seems to benefit you. Here's an easy way to test: hook up an external screen and disable your laptop's screen. That way we'll see if the extra power still gets diverted to your CPU or not. If it does, then that means your iGPU has been pulling the same load the whole time and it IS a package limit (probably 57W) that is shared by your screen and your iGPU, and that your iGPU is basically drawing a constant of power. While on the external screen only, try loading your dGPU with say... unigine heaven on lowish settings (don't wanna kill your CPU) and then running TSBench and seeing if closing Unigine Heaven while TSBench is running grants extra power to the CPU to use (indicating that the iGPU being stressed as a passthrough for the dGPU took more power). If it does, then that means Lenovo had a RIDICULOUS design for that machine when it comes to power limits.

Simple. He's at 1.02v and you're at 1.1v for essentially the same speed. I'm at 0.9976v (TSBench runs at 0.9926v) and I stay UNDER 47W when I do TSBench on my 4800MQ at 3.5GHz. But my chip is one of the hottest I've ever seen so don't wish for it. I prefer my older one which was cooler but drew a bit extra power.

Yeah you're right; spoke about this theory a paragraph above.

 

I want to download that program called "maxmulti". please give me link...
thanks.

 

Nuu. You haven't even listed your specs.