The ThrottleStop Guide

I think this is the post you're looking for: http://forum.notebookreview.com/7415484-post476.html

 

My understanding was that there is an additional bit lock for the turbo ratio limits. I thought MSR 0x610 only contained power limits and turbo time limits, which can only be used to make turbo run longer at the max value.

 

I have read that the turbo ratio limits can only be adjusted higher by the bios. With the testing I've seen, that's probably true but it took a few years to unlock the Core 2 Extremes so who knows.

After values are written to the turbo ratio limit register, the CPU needs to be restarted so I don't think this is going to be possible with any Windows based software like ThrottleStop.

 

Hi UncleWebb,

I hope you remember me from a long time ago and my questions regarding the nature of the ES CPU I had?

Well, I got another ES on ebay, which the guy claims to be an i5 580M. Can you please help me verify if this is true? Here are screens of the CPU running under minimum load and full load (Matlab).

 

Is it possible to restart the CPU using some pre-boot DOS script? I tried setting the turbo ratio limits and rebooting, but the values reset on reboot.

 

The pure DOS-only CPUID can set the CPU MSR registers BUT applies them only to the single core that is running the OS. There is no thread affinity under DOS. So I was able to change the VID for one core under DOS but it's impossible to enable dual-IDA.

Perhaps that utility is helpful?

 

Tried it. It would set the MSR, but when I booted windows it would set back to default. I then used MSR tool to set it back higher again, but no luck.

Interesting little utility though.

 

Septimus_DSX: The CPU-Z Specifications box always reads the name string directly from the CPU. This box shows Core i5 M 580 and (ES) at the end of that string so it is definitely a Core i5-580M ES CPU.

The maximum multiplier when 1 core is active is 25 and the maximum when both cores are active is 23 so averaging a multiplier of 23.67 when loaded seems about right.

Khenglish: I have no hardware to play with so I won't be able to help you out. You have a good idea but I think the bios will always reset this register to default values.

 

I posted on the Intel forums here about the max turbo ratio limits MSR for Sandy Bridge. It's a longshot, but I didn't know what to do before bed so I made a post there.

 

I'm unable to download 3.40 from the download site. Have tried both IE and Firefox, both just loop back to the same download page when I hit the download button.

Any chance of mirroring the file on a proper download site?

 

Taomyn: Thanks for the feedback. I added a MediaFire link to my sig for the new version.

 

Hey Unclewebb, I'm having a couple of problems with turning throttlestop off, I might need to end up posting a video here, but I'll try and explain it....

I'm testing temps with throttlestop and windows power management so I'm switching between both of them, before I go any further her's my specs;

dell n5010
i5-460m
5650m
windows 7

My problem is basically, I can't get my turbo back. I've got 4 throttlestop profiles: full turbo, 2.5ghz, 2.3ghz and 2.1ghz.

The problem comes when I've turned off throttlestop - windows is on 100% high performance profile and it only goes to 2.5ghz. Now this is the weird part, when throttlestop is turned off and on windows power management, my cpu won't go turbo (2.8ghz) unless I change the throttlestop profile to turbo even though its turned off, but the lower clocked throttlestop profiles don't take effect (2.1, 2.3ghz profiles)

So effectively unless I use throttlestop I can't use turbo, I've lost my windows power management turbo profile.

Is there any way to stop it controlling the cpu all together?


regards Alex

 

Hey Unclewebb,

I'm trying to use throttlestop to make my XPS to save as much power as possible.
I'm already using power saver mode with 7x multi. Would cutting CMOD and CHIP clock mod reduce power consumption? Which one do you recommend changing? I know it kills performance...

 

unclewebb,

A nice plug for your ThrottleStop software over at Anandtech.

See:
AnandTech - Mobile GPU Faceoff: AMD Dynamic Switchable Graphics vs. NVIDIA Optimus Technology

Glad to see you getting a little wider recognition, which you rightfully deserve too!

 

unclewebb

Great to see others taking notice of your great work! Thanks for being there for me and everyone else.

 

@unclewebb

As you know i am currently working on bios hacks for the latest Clevo Notebooks.
During our testing we found that the TurboLimit(s) value in the TPL section doesn`t reflect the value set in bios:

Turbo Limit(s):

bios value <> Throttlestop reports
108 <> 96
63 <> 56
45 <> 40
28 <> 16

 

Have you checked to see which is right? Normally I'd believe BIOS over a program, but in this case I believe throttlestop is directly reading from the CPU MSR. Check MSR 0x610 with MSR tool or RWeverything or whatever you prefer, to see if the BIOS is actually setting what it says it is. Are programs reporting the CPU speed consistent with what the BIOS says or throttlestop? It'd probably be hard to tell and require some testing.

See pg 647 here on how to read the MSR values.

 

where can i download the older TS 3.30? 3.4 didnt work for me

 

I would love a P150HM bios that allowed overclocking... For the 2920xm that is...!

 

many thanks l, Unclewebb!

any ways of a small process to stop windows 7 x64 going idle? speedstep ruins audio!

would prefer to run 4 cores at 2ghz+ ready for audio stream on the i7 920xm N61JQ

 

alex2009: Make sure this option is NOT checked:

Do Not Reset FID / VID on Exit

When that box in the Options window is cleared, it tries to reset your CPU to its default specs so Windows can take over full control of it.

If you reboot your computer and don't use ThrottleStop and Turbo Boost is not working correctly, that's a problem with your bios.

DuranXL: I don't recommend using either type of clock modulation for any reason. It can make a CPU run sluggish but doesn't seem to save any power. Even using a lower multiplier might not save you any power. A slow CPU is not very efficient. It's better to run a CPU fast so it can quickly get its work done. This allows it to spend a bigger percentage of time in one of the low power sleep states like C6 where the real power savings can be found.

tilleroftheearth & Bobmitch: I was shocked to see ThrottleStop mentioned in a mainstream media source. TS is not widely known outside NBR members. AnandTech got some nice FPS increases when using TS with a reduced multiplier. If they would have found a freezer and properly cooled the Acer 3830TG, they could have used ThrottleStop to run that CPU with full turbo boost, the way Intel designs their CPUs to run.

Prema: None of the Alienware users that I have helped have reported a difference between the bios TDP and what ThrottleStop is showing. The only thing you can do is take Khenglish's advice and read that MSR directly to make sure that ThrottleStop is working correctly. I have seen other situations with SpeedStep where what you set in the bios might not be what you end up with when you get to Windows but I have never heard of an issue with the TDP setting. Let me know what you find out.

notroxcity: When you say, TS 3.40 didn't work for me, it would help if you explained what didn't work. One problem I discovered is that on Sandy Bridge processors, you can not use a reduced multiplier and the Disable Turbo option at the same time. In TS 3.43, I have blocked this so you can't use these two options at the same time. If that is the problem you were having, you might be able to find TS 3.43 hiding in my sig.

AudioManiac1234: There is a registry modification you can do to force a 920XM to run at full speed all the time.

Analysis of slow SSD performance Intel HM55/PM55 chipset | Tech|Inferno

Hopefully the cooling system in your N61JQ is up to the job.

 

TQ!! downloadeded TS 3.5 n it works flawlessly when i used 3.4, changing multipliers or profiles did nothing to my cpu speed. great work unclewebb!! appreciate it!

 

Thanks! So lets say I would do one superpi run on 7x multiplier and one on 12x. Let's say that on 7x it takes 40s while on 12x it takes 20s. You say that while on 12x it's twice as fast, it does not in fact consume twice the power?

Your 'power saver' feature forces 7x multiplier when idle. I assume 7x multi does consume less power in idle state than 12x?

edit: Btw i have this bug when I switch from profile power saver to my profile "disable turbo". It stays in 7x multiplier even though I force 12x multiplier

 

THX, i found the culprit...sort of a higher level boot override set in place by Clevo...now values are matching the bios.

There is still a similar override in place to keep the multis from being changed...but that one is hidden...grrr

 

Are you trying to change the multi from the bios?

 

Yes, in our tests we have TDP, TDC, RAM etc under our control now, but the XM core multis are still overwritten from a hidden override each time the bios boots...

 

Here's a picture of the bug/problem:



edit: I'm assuming the disable turbo mode overrides the multiplier setting, causing the multiplier to stay at the current value.
If I turn off disable turbo it engages 12x multiplier. Perhaps it's better to grey-out the multiplier setting when engaging disable turbo

 

Prema: From what I have read, Intel claims that it is not possible to change the multiplier from within Windows on the new Sandy Bridge Extreme processors. The only way this is possible is the CPU needs to be reset by rebooting it. Most bios versions ignore this and then set the CPU to the default bios values so the overclock is ignored. I would think that the CPU must have some sort of lock bit to protect the Turbo multipliers from being reset when you reboot but Intel has left any mention of this out of their public documentation so ThrottleStop is presently screwed.

Maybe someday.

DuranXL: That bug you found was fixed in version 3.50. It is no longer possible to select Disable Turbo when you are using a reduced multiplier. Version 3.50 disables and blocks the Disable Turbo option since that isn't needed when using a reduced multiplier. TS 3.50 handles Disable Turbo more efficiently so it is worth upgrading to.

That is exactly right. Power consumption does not go up linearly with speed. That's why you are better off running a CPU fast so it can get its work done as fast as possible and then go back into one of the sleep states. The C6 sleep state in the Core i processors turns off voltage to that core so power consumption drops to virtually zero.

For a crude example, let's say you have a 3D rendering project that will take 30 minutes if you run your CPU at 2000 MHz and the same project will take 60 minutes if you run your CPU at 1000 MHz. If your CPU consumes 20W at 1000 MHz and 25W at 2000 MHz then the total energy consumed would look something like this.

CPU @ 2000 MHz
30 mins @ 25 W = 12.5 W
30 mins @ 0 W = 0.0 W
Total = 12.5 W

CPU @ 1000 MHz
60 mins @ 20W = 20.0 W
Total = 20.0 W

Most people have been brain washed into believing that a slow processor is the best way to save power but not only is this not true, it can actually increase power consumption. In the early days of ThrottleStop development I was having a hard time convincing people of this so I created the "Power Saver" feature to keep users happy. On a lot of CPUs, the ThrottleStop Power Saver feature isn't going to save you much, if anything. Here's a paper that discusses power consumption in some modern Intel CPUs.

Power Optimization – a Reality Check
http://www.cs.berkeley.edu/~krioukov/realityCheck.pdf

"...the best possible approach is the so-called 'race to sleep' model of computation."

The above wattage numbers are just random numbers off the top of my head. The 45nm Core 2 mobile processors also have some reduced power states so running them fast is a good thing for performance without any power consumption penalties.

ThrottleStop reports power consumption of the Sandy Bridge CPUs so you can obtain some real world values. For the original Core i mobile processors, I think HWiNFO reports CPU power consumption. When testing, use a consistent load like Prime 95 Small FFTs and also try to use a consistent CPU temperature. As a CPU heats up, power consumption will go up so that can interfere with accurate and repeatable results. Once you have some data points, you can see if going faster is worth the small increase in power consumption.

 

Sorry, i wasn`t very clear. We are testing a modded bios with XM (Extreme) CPUs only and still the bios is overwriting the core ratios...

There are different places in the Clevo bios that allow for the multis to be set.
First is the "CPU configuration", the second (in XTU bios only) is the "Performance Tuning" section, the third, and this is our problem, are hidden values in the "Boot override" section. Those are forcing even the Extreme CPUs back to default...

 

One thing I would suggest trying is doing sleep/resume while in windows. This may skip the "boot override" section of your BIOS since this is a quicker restart than normal. This trick does not work for me, but maybe it will for you since your bios has the higher multiplier set in it.

 

Hi i ve been playing crysis 2 and it seems to slow down sometime.. I want to check if it is throttling or not, how do i cheack this using throttle stop?

 

On the top right corner it shows the currently used multiplier...

 

How woud I know if it throttle or not?

 

Hey man, I'd just like to say thanks, I was getting these weird drops in crysis 2, it was going from 50-60 and then drop to 20 for 2-3 seconds but it would do it so frequently I'd get frustrated. Not anymore, thanks to throttlestop. I have yet to try Deus Ex, but I don't think I like that game anyways, so I'm debating whether I care... But yea.

Now I can play crysis 2 on 720 in extreme with a consistant 40-50. Thanks bud!



edit- and more importantly, BF3 is now running extremely smooth, which is kind of what I bought this laptop for, until diablo 3 comes out.

 

thrashassassin: That's good to hear. ThrottleStop can make the difference between unusable and quite playable. Too bad laptop manufacturers that sell gaming hardware don't put more effort into their designs so the fancy hardware could be used to play some games without throttling to death.

trinox: Run ThrottleStop in monitoring mode and put a check mark in the Log File option. When you are finished gaming, have a look in the ThrottleStopLog.txt file and see if the multiplier stays nice and high or if it sinks down and throttles to 8.0. Also make sure both types of clock modulation are always running at 100.0% in the log file.

Try playing the same game with ThrottleStop actively running. Click on the Turn On button and then click on Set Multiplier with that set to Turbo and see how game play is then. Run a second log file so you can have a look at your CPU multiplier and CPU temperature. Higher temperatures are usually a good sign that your CPU is working harder and not being throttled. If you are not sure, upload your log files to www.mediafire.com and post a link so I can have a look.

 

Hmm not sure if 3.50 works better. Now I can't force 12 multi. It will go back to turbo. So now when I want to disable turbo I have to set multi to turbo then enable 'disable turbo'. Not that logical.
Also sometimes it's buggy
http://img823.imageshack.us/img823/8162/24255259.png

Ok thanks, clear

 

So i just downloaded 3.5 TS on my g74sx i read a write up chasity did on the ROG forum. WHat set up should i be using for my system? "Unlock Bidirectional PROCHOT" is checked off in the options and "BD PROCHOT" is not checked on the main screen is this correct? ALso can i just X out of the program after doing that or should i be hitting "TURN ON" THanks

 

Bro, I'm new on this, but my m14x seems to have overheating, so I want to turn off the turbo mode of my i7 2630 in order to get low temps on my cores, how can I do this with ThrottleStop? is setting clock modulation and chipset clockmode under 100%? or how can I doit? thanks dude.

 

Are you using the Windows High Performance profile with the Minimum and Maximum processor state set to 100%?

You might have to use a different method compared to previous versions of ThrottleStop but if it works correctly to disable Turbo Boost then that's a good thing. To test this, put a load on your CPU like a single thread of Super PI and then toggle Disable Turbo on and off while the benchmark is running and see if this feature is working correctly on your CPU.

I agree but you will have to mention that to the Intel engineers. Turbo Boost was an add on to the original Core 2 design so they have at least 3 registers all playing a part in controlling Turbo Boost. I am slowly finding out the differences between how Core 2, Core i and the newer Sandy Bridge Core i processors handle these registers. The public documentation leaves out some details about changes between CPU generations so I will need some help from the user community to get this working correctly on as many different processors as possible.

I am not surprised. I need access to some newer Core i hardware and lots of time but at the moment I don't have either. If you have the time, show me what works and what doesn't and try to tell me how I can recreate the problem you found. Only use a single profile when initially testing to eliminate that as a possible problem.

TS 3.50 Disable Turbo seems to work better on the newer Sandy Bridge Core i processors. If the previous method works better for your CPU then you will have to use an older version of ThrottleStop until I have the time to find out what's wrong and come up with an improved method that works on both the old and new processors.

ZACK02: Start ThrottleStop, if it is in Monitoring mode, click on the Turn On button, click on Set Multiplier and set that to Turbo and then click on the Disable Turbo box. That should work correctly on your CPU to disable Turbo Boost. The ThrottleStop Clock Modulation feature might also be a good way to control heat on the newer Sandy Bridge CPUs like you have but I would avoid using Chipset Clock Modulation. You will have to do some trial and error testing to see what works best.

 

Hm okai, I might try to recreate the problem.
What I meant with 'not logical' is that the disable turbo button is obsolete, as putting the multiplier on anything but "turbo" would logically disable the turbo.
Now when it's on disable turbo it just disables the turbo, but ignores the mulitplier setting...meaning that switching to a disable turbo profile is useless because it keeps multiplier from the last profile.

I now use x12 multiplier, and it works this time. Before it would set itself to x11 (just like when you go 1 down from turbo, it goes to 11 not 12)

 

Dear Unclewebb,

I want to ask for help.
And so this post is not completely off topic, here are just a link to my "ThrottleStop" settings and the corresponding voltage values. (Penryn CPU)

I deliberately use that old "ThrottleStop beta" version. This has several advantages. Most importantly, IDA's start capability in all profiles.
Moreover, this version retains its profile even after the energy saving mode. (despite starting profile 1 setting)

The "290 beta 6" version is the last of these benefits. By using SetPll for me it is absolutely necessary. My Windows startup profile is "1".
After finish the boot process I manually change to profile "2".

Maybe it helps someone.

Now to my request for help. I am trying to date a "QX9300 / Q9200" to persuade to work on a PM965 (HDX Dragon). Many "NBR-VIP" are already on board.

I also think you could still have a good idea. If you have 30 minutes time, then please just read this one page (Q9200) and then the pages for my "QX9300" test.
The test of the "QX9300" starts on this page and has not been completed.

If you have no time, my bad luck. I have two processors, when not successful, return in eight days. Otherwise, this test is even more expensive.

To all others, sorry for off topic and sorry for bad English. That's not me, that's "Google Translator" with my correction. I can unfortunately only German.

 

Unclewebb, I present to you my incredibly interesting case of throttling on a 2nd-gen i CPU:

I own a 2920XM ES on a MSi MS-16F2 (with unlocked BIOS).

Here's a picture of my situation:





My log is attached at the bottom of this post. As you can see I have managed to find a way to modify my TDP max values and such through the modified BIOS. Here's the problem though -- the multiplier oscillates so much it drives me crazy. It fluctuates between x28-x31 for no reason. The TDP will stay at a constant 63W-67W, yet the CPU will find some reason to downclock. I have tried to see where the downclocking begins and I notice that it is a case of C0 downclocking. Could there be ANY explanation why the ratio won't stick. I have even bumped up the "time" for the turbos to their max values in the BIOS and I'm afraid it didn't do much. This happens with the BCLK at 99.8mhz too, so it is not an overclocking problem. Temps hit ~94*C max, which should not trigger any throttling. What exactly is this C0, I might have an option someone hidden in my BIOS.

Excellent work on this program, I hope one day we can overclock our XM CPU's without the need for BIOS support. Thanks.

 

DuranXL: When you are using a single profile, does ThrottleStop 3.50 work correctly? The base multiplier for your CPU is 12 so I think the TS Set Multiplier adjuster should go from 7, 8, 9, 10, 11, 12 and then to Turbo.

With the CPU partially or fully loaded, can you step up and down through your multipliers using ThrottleStop without any problems? The monitoring area should show any multiplier from 7 to 12 and when you go to Turbo, if your CPU is not over loaded, you should see some Turbo Boost with the average multiplier randomly varying and higher than 12.00. If this works correctly then the problem you posted is probably in the profile switching code. My Core i laptop is down at the moment so it will be a week or two before I can do some further testing.

Disable Turbo used to work independently of Set Multiplier. On the Core i5-520M I tested, this seemed to cause conflicts between the two settings. In TS 3.50, I combined these two features so Disable Turbo would work immediately and more reliably on the newer Sandy Bridge CPUs but this is still a work in progress. I'll post a link to TS 3.30.1 in my sig in case anyone needs an older version.

I think I understand your point that Disable Turbo is not needed anymore because a reduced multiplier should accomplish the same thing.

RickiBerlin: I can not help you with your problem. BSEL mods and getting Quad Core processors to work on older motherboards is beyond my skill level. Sorry.

Xonar: My best guess is that your CPU is being throttled internally, likely due to a lack of power. The maximum amount of power that can go to your CPU on that motherboard might be limited so throttling is the result. C0% is a good indicator of your CPU being throttled internally, especially if Task Manager is showing 100% and C0% is showing far less. Try testing with Prime95 Small FFTs. What is the maximum multiplier you can run and still maintain 100% C0% when fully loaded? Your log file shows immediate throttling when the load tries to go up. If your motherboard can not deliver enough juice to your CPU, there probably isn't anything that ThrottleStop is going to be able to do about that. Are there any other users having success with these ES processors on your motherboard? The Alienware M18x guys are having excellent success with these EBay specials so you might want to ask them.

 

Well I'm sure you know, the M18x is the only one able to OC from the BIOS (can set multiplier through the BIOS) AFAIK. My notebook is the same as any other, no hard multiplier options exist in the BIOS only a turbo ratio LIMIT option. The notebook is designed to accommodate 55W TDP XM-CPU's so I see no reason why I should be throttled at stock turbo ratios.

To make it clear, I'm trying to prevent throttling at STOCK TURBO speeds. No overclocking, voltage mods or anything.

Here's what information I've learned so far:

I originally was 100% sure it was a TDP issue. When running wPrime, or the like, my CPU would hit it's max turbo at x31 for ~60 seconds before it throttled down to x28. I noticed the TDP would hit around 55-65W and then just drop to 55W and hug it. It wouldn't budge past 55W. So, I unlocked my BIOS and managed to change the TDP limits as you can see in the second screenshot above. This did actually prevent the CPU from dropping to 55W TDP after ~1 minute. If you look at the logs very closely, you will see that TDP does not seem to be indicative of multiplier ratio. Sometimes at 63W I'll be at x28 and a minute later at 64W I'll be at x31 and then the next minute ill be at 60W pushing x30. It just oscillates freely, completely random! On one thread, the multiplier fluctuates between x32 and x33 for some reason, I figured it was just a blip -- TDP is ONLY at ~26-30W. Therefore, I believe the TDP limit WAS an issue, until I raised the limit. Now, I'm starting to think there's another limiting factor, that's why I came to ask you .

Attached is my new log on TS with P95 running as you asked. Now this is where it REALLY starts to get confusing for me. C0, Clock Mod, Chipset Mod = 99.9-100, that's exactly what I'm shooting for. But notice the dreaded multipliers. . . They oscillate so much! Is 95*C a throttle-threshold for SB CPUs? Or is 100*C where throttling begins? Could this behavior be based on how Turbo works? Are my TDP/Temps too high to maintain such a multiplier? Also, if you notice, towards the end the multiplier actually RISES x1 more than previously before and maintains it there, even though before at the same TDP it was x1 lower. Because the scale on the graph is to accommodate all the values, seeing a 1-3x multiplier change is very hard to actually see. But, if you look closely it's not a straight line at all, lots and lots of jaggedness to it. Windows Task Manager reports the same C0 that TS does. So weird, why does it work for Prime95 and not Intel Burn.

The reason why these paltry 300mhz are very important to me is, in SC2 on Ultra for as long as the CPU is locked at x31, I get a very fluid 60 FPS down to 50 FPS with max units on screen. When the CPU throttles down to x28, my FPS dips dramatically to 45 FPS normally and 30-35 FPS during a fight. My GPU temps were 77*C maximum. Is turbo really this unreliable? I have read so much on it and I thought that as long as headroom exists, your CPU will always be turboing.

I accidentally pulled the plug for my AC, that might explain that big dip towards the end. Thanks for taking your time and help.





PS, I have set my Turbo Ratio Limits for (1-4) cores in BIOS to 40. I have also changed in TS all the limits to 40 and no dice on that either, still only turbos up to x31 in 4-core.

 

Yes TS works fine when I use 1 profile.
Anyway I now can't make a profile that disables the turbo properly. (so 12 multi). I put 12x multi on profile #2, press save, go to #1 and back to #2, however the multi reverted back to turbo for some reason.

 

Some things to mention about internal CPU throttling:

As far as I remember the TPL value you set is an *average*! So it's normal to see higher wattage from time to time and have it settle towards the average when CPU is constantly loaded.

Also 94°C is quite close to the 100°C where throttling should definitively happen. Keep in mind that a single CPU core might already trigger temperature related throttling and that the 94°C reading may be archived just *because* of throttling (=would be higher without).

Right after starting Prime95 I already see my CPU downclocked to x28 (mobile i7 Quad) and then crawling towards x26 after some time. So I assume this behavior to be normal.

 

Concerning "Power Saver" + "Multiplier". I'm still on v3.3 and did a quick test of combining "Power Saver" with a lower multiplier. Seems to work here (mobile SNB i7 Quad).

What issues did you experience exactly?

 

Xonar I think it's temp related too. Remember that TS is not polling the temperature sensors as fast as they are actually being updated. I see that TS recorded a temperature of 97C, which is close to the throttle point. This core may have popped up to 100C, then back down again in between updates. You can check "more data", which will update the temperature data much more quickly. Try this and see if it records 100C.

I'm surprised with such high temps that your CPU will run at the max multiplier. On my 2920xm ES, prime95 will start out between x27 and x28, then drop down to x25 as the CPU heats up. I found that unless my room is cool, I will hit 100C, and the CPU will throttle to a x20 multiplier for a split second, then go back to x25 or so.

You also may be able to disable the 100C throttling limit. It probably isn't a very good idea, but I'll look into it tonight.

 

There may possibly be an option hidden in the BIOS. Mine has one, I can raise it like up to 120*C I think. I'm not going to do this though, it's a bad risk.

 

It's not dangerous if you didn't change the critical temperature threshold. But on second thought still risky because some notebook have very high critical temperature threshold. By looking at your CPU temperatures, look like the passive cooling trip point is set to 95C on your notebook.