IdeaFan for y510p y500 G500 Y580 y410p

[allstone]

I have been trying to find a way to control fans on my y510p. As everything else failed up to the Embedded Controller (EC), I have made myself a program to call dust removal function. With a help of this forum members, I have perfected calling and stopping timing which works on my computer.

Basically, running Furmark, if I do not run this program, I constantly run into 97C and thermal throttling for my GPU (see Mhz of core and memory):

And when I do run this program, it manages to keep temperatures at 88-93C and there is no more throttling in a long run.

Usage: run the program and it will enable dust function when you reach GPU/CPU temperature of 90C. Or Start/Stop it with keyboard shortcut Ctrl+Shift+Q or through menu on tray icon.

 

Download: IdeaFan v1.6 (includes OpenHardwareMonitor 0.7.1.0 beta)

So far, this has been tested on 2 computers (y510p Win 8.1/Win 8.1Pro/Win10). Both with Lenovo Energy driver v7.x and v8.x. I have tried it for both 2xSLI and 1x GPU, also for CPU stress tests or simply during high CPU usage, i.e. rendering videos. Feel free to use and leave some comments that it can be further improved if possible.

 

Danger! Do not use Furmark to stress your GPU unless you know what you are doing. I have used it here for testing purposes only.

Release Notes

v1.6

*change* Migrated to OpenHardwareMonitor 0.7.1.0 beta)

*fix* Temperatures are shown and working on Windows 10 (tested on build 1511)

v1.5

*fix* Supposedly startup error with OHM should be fixed

*new* Added another keyboard shortcut just for diversity

v1.4

*fix* When fan is running, nothing should disrupt the cycle (not tray menu nor high cpu/gpu usage or temperatures)

*change* Startup option changed from startup folder to windows task

*change* Updated default settings

*change* Removed sleeps, introduced timers/threads

*new* Lower cpu usage

v1.3

*change* Reverted default settings (to have several versions with all newest changes available)

v1.2

*change* Updated default settings (provided by octiceps)

v1.1

*new* Options are saved in ini file

*new* Automatic dust function control based on temperature (OpenHardwareMonitor 0.6.0 beta included)

*new* Startup option fixed

*new* Ability to change timeouts in ini file

v1.0 Initial Version to run dust function

And here are possible improvement requests for the fututre:

1. Make fans start hitting the 9s/0.5-1s cadence right off the bat. Need to make sure that the last time fans are stopped, its within 0.5-1s gap when they are stopped. This is due to the EC somehow remembering where it left off the last time and resuming from there. If you stopped it in the middle of a spin cycle, the next time you run it the fans will be inconsistent inside the first minute before finding their rhythm again. The length of the gap is also determined by last your stopping point. I've gotten it anywhere from basically none (fans spin down and spin up immediately) up to 1s, but it never goes above that. Edit: tested to stop at 0.5-1s gap, but that seems to have no effect on the initial position. Also tried variable timeouts - seems for EC its no different, with variable its unpredictable, constant value may be required afterall. For now its seems the only way to start off the bat for sure is to know internals of EC - or at least get a fan RPM sensor readings which we do not have.

2. When OHW 0.6 is not started, IdeaFan throws errors (need to check for them), only at COLD first start. See http://forum.techinferno.com/lenovo-ibm/7358-ideafan-y510p-6.html#post102492 (should be fixed in 1.5)

3. Introduce easy HotKey settings in INI file (like Ctrl=yes, Alt=yes, Key=q) and let change the hotkey. Also handle hotkey cannot bind error.

Edited September 4, 2016 by allstone
IdeaFan 1.6

[Jerenny]

Very interesting, downloading right now!

[octiceps]

Does this simply run Dust Removal in an endless loop like the script does? Or does it remove the 2 second gaps between when the fans spin, i.e. constant max fan speed?

Also, please do let us know if you ever find the BIOS settings that allow HWiNFO64 to access the EC for manual fan control.

[Jerenny]

Does this simply run Dust Removal in an endless loop like the script does? Or does it remove the 2 second gaps between when the fans spin, i.e. constant max fan speed?

Also, please do let us know if you ever find the BIOS settings that allow HWiNFO64 to access the EC for manual fan control.

I'm running it right now, it runs the fans for about 2 seconds, and then stops for half a second and repeats.

EDIT: It seems like its running at about double time? Normally it would run for about 7 seconds and then stop for 2? This seems about half that.

[octiceps]

I'm running it right now, it runs the fans for about 2 seconds, and then stops for half a second and repeats.

EDIT: It seems like its running at about double time? Normally it would run for about 7 seconds and then stop for 2? This seems about half that.

I timed it and normally it is 9 seconds on, 2 seconds off. So this is 4.5 s & 1 s? Hmmm. I'm gonna learn the scripting this weekend and see if this can be improved further. Constant max fans or reduce the gap even further if possible.

[Jerenny]

Results(Also changed thermal paste in Ultrabay GPU to Arctic MX-4 and no dust filter at the bottom):

I wouldn't use this mainly because I've always found these scripts to be too sound jarring around other people, to hear it go off and on, but to anyone that is thinking about trying this and you must absolutely have lower temps, definitely use allstone's script. This along with undervolting both GPUs at stock clocks and a laptop cooler(I use the NotePal X3), yields very epic results. Battlefield 4's Firing Range is a great way to test temperatures after you put it on some very nice settings, these settings even give my desktop's GTX 770 hell. These settings puts both GPUs to 99% stress the entire time, AS LONG AS you stand still right after you deploy onto the firing range.

After about 20 minutes, I was greeted with these pretty amazing results:

And that's with 99% GPU usage, actually playing games will make your usage fluctuate anywhere from 60-99, so expect even LOWER temps. Highly recommend this tool. You guys should definitely thank @allstone for making this.

EDIT: I was undervolted by -100mV.

[octiceps]

Yeah, that's why I'm trying to figure out a way to keep the fans at a constant speed. Even though it would be loud, it's nothing that noise-canceling headphones can't drown out. Hearing the fans turn on and off nonstop is much more annoying, not to mention it hurts cooling performance.

Nice temps BTW. I get about the same on GPU and higher on CPU (due to i7) in every game except PlanetSide 2 because it's a power virus. At constant 99% usage, it easily heats up my GPU 10-15C more than anything else I throw at it, which is especially problematic on the Ultrabay card since it's already hotter to begin with. Depending on ambient temperatures, I see close to 90C on it while playing PS2, and it being 100F every day here in the CA Bay Area isn't helping either.

Otherwise, I've already done a few other things to manage thermals, including using a cooling pad, repasting with IC Diamond and OCZ Freeze, removing the cloth mesh that's glued to the dust filter, and dremeling the bottom cover to increase the size of the intakes. Once I get fan speeds figured out, I should be golden.

[Jerenny]

Yeah, that's why I'm trying to figure out a way to keep the fans at a constant speed. Even though it would be loud, it's nothing that noise-canceling headphones can't drown out. Hearing the fans turn on and off nonstop is much more annoying, not to mention it hurts cooling performance.

Nice temps BTW. I get about the same on GPU and higher on CPU (due to i7) in every game except PlanetSide 2 because it's a power virus. At constant 99% usage, it easily heats up my GPU 10-15C more than anything else I throw at it, which is especially problematic on the Ultrabay card since it's already hotter to begin with. Depending on ambient temperatures, I see close to 90C on it while playing PS2, and it being 100F every day here in the CA Bay Area isn't helping either.

Otherwise, I've already done a few other things to manage thermals, including using a cooling pad, repasting with IC Diamond and OCZ Freeze, removing the cloth mesh that's glued to the dust filter, and dremeling the bottom cover to increase the size of the intakes. Once I get fan speeds figured out, I should be golden.

Regarding the fact that this script doubles the speed in the dust removal script, do you think it'd be just as easy to SUPER slow it down and have the 100% fan speed last for hours and then stop for hours(Which wouldn't be a problem, just restart the script yourself and have it run the fan for hours again)? More advanced to even just have it boost the speed of the UltraBay GPU...unless those two(which I hope to God they aren't) are linked to the exact same fan controller and operate symetrically, because honestly, the CPU and GPU0 temperatures are fine to me, even without the script as well as overclocked, it's just the Ultrabay temperatures that need serious work.

[ghoul]

Can you please run Unigine Valley, windowed, near 1080p?

MSAA not required, but let it heat up first with 3-5 rounds, then reset hwinfo, and then run 3 more.

This way we can have a reading of average and usage based fluctuation of temps.

Thx.

( At default speed, you can go lower with voltage, mine is ok @0.95V. @1V 1124mhz looks stable. Flashed into ultrabay, and merged into main bios, so no nvinspector needed )

Edit:: And if everything else fails, do some hardware tricks like this:

http://forum.techinferno.com/lenovo-ibm/4921-y500-y510p-spring-mod-spacer-guide.html

[octiceps]

Regarding the fact that this script doubles the speed in the dust removal script, do you think it'd be just as easy to SUPER slow it down and have the 100% fan speed last for hours and then stop for hours(Which wouldn't be a problem, just restart the script yourself and have it run the fan for hours again)? More advanced to even just have it boost the speed of the UltraBay GPU...unless those two(which I hope to God they aren't) are linked to the exact same fan controller and operate symetrically, because honestly, the CPU and GPU0 temperatures are fine to me, even without the script as well as overclocked, it's just the Ultrabay temperatures that need serious work.

I don't think it's possible to slow it down more than the default 9 seconds, not unless we get access to the EC to do real fan control. I believe what allstone's app does is start Dust Removal, cancels it after 4.5 seconds, starts it again immediately, cancels after 4.5 seconds, and so on and so forth. The 1 second gap is probably the minimum amount of time it takes to stop and start the Dust Removal command. If that's the case, maybe the best we can do is 8 or 8.5 seconds on, 1 second off. I won't know for sure until this weekend when I look at the AutoHotkey scripting documentation and play around with it. Modifying the script shouldn't be very hard, I just don't have time ATM.

[Jerenny]

Can you please run Unigine Valley, windowed, near 1080p?

MSAA not required, but let it heat up first with 3-5 rounds, then reset hwinfo, and then run 3 more.

This way we can have a reading of average and usage based fluctuation of temps.

Thx.

( At default speed, you can go lower with voltage, mine is ok @0.95V. @1V 1124mhz looks stable. Flashed into ultrabay, and merged into main bios, so no nvinspector needed )

Edit:: And if everything else fails, do some hardware tricks like this:

http://forum.techinferno.com/lenovo-ibm/4921-y500-y510p-spring-mod-spacer-guide.html

On it right now, I'll edit the post and leave a mention so you can tell when I edit this. As far as that spacer thing goes, I really need time to re-read that over and over because I didn't really understand the whole mod, I'll need a video or something...I'll read it after I get done with this bench.

EDIT: @ghoul , I'm done...Man Heaven Benchmark gets boring to watch after the second run. Anyways, I'm on stock clocks(1058/2500) and undervolted -100mV and these are my results. I ran no MSAA, 1920x1080, and the Extreme preset..I forgot to turn on tessellation, but GPU usage is GPU usage.

After 5 runs:

Reset HWiNFO64 and 3 more runs:

Can't leave out the fact that my laptop is still sitting on top of my CM NotePal X3. I haven't tested temperatures with this test without the cooler, but overall, I'm still impressed with this script, and can't wait for it to be improved if it ever does.

[octiceps]

LOL he asked for Unigine Valley not Heaven.

Your GPU usage is a bit on the low side. I always bench Heaven completely maxed out (Extreme Tessellation and 8x MSAA) and I never drop below 95% GPU usage, usually pegged at 99%.

[Jerenny]

LOL he asked for Unigine Valley not Heaven.

GOD DAMMIT. Dude this shit took like 45 minutes..Ugh...Guess i have to do this again...

[octiceps]

GOD DAMMIT. Dude this shit took like 45 minutes..Ugh...Guess i have to do this again...

Just completely max out Heaven per my post above. It's a more demanding test than Valley, especially on the GPU core, which is what you want to push to test temperatures.

[Jerenny]

I redid it, only this time it wasn't as boring because I fell asleep. Still stock clocks at 1.0V(-100mV undervolt), but this time Heaven is completely maxed out, 8x MSAA, Extreme Tessellation, Ultra quality, 1920x1080, and fullscreen. I only took about an hour nap. Here are the results, you can see the exact time at the bottom of the HWiNFO64 window.

EDIT: Heaven was indeed in fullscreen. I just Prt Scr'd the HWiNFO shot and pasted it onto the screenshot I took of heaven, which you can see from the temperatures at the top left correlating with the Max temperatures in HWiNFO. That's why the "Current" temperatures are much lower than the temperatures you see in the MSI afterburner overlay.

[octiceps]

So what are your actual GPU voltages under load now after the undervolt? Wish you didn't hide those sensors in HWiNFO64; the info would be useful.

[Jerenny]

I said 1.0V after the undervolt o-o...Is that not what you're looking for?

[octiceps]

I said 1.0V after the undervolt o-o...Is that not what you're looking for?

Sorry, missed that part. Is it 1 V on both cards under load?

1 V is what my 650M is at stock and I was able to push it to 1120 MHz on that voltage, so it should definitely be doable for most if not all 750M at 1058 MHz unless you got a bad chip or something. Interesting that default is 1.1 V, which seems needlessly high and no doubt contributes to the many reports of overheating 750M cards. Extra voltage really ramps up heat like crazy. I got a 5C increase on primary GPU and 10C increase on Ultrabay GPU from just +30 MHz +50 mV.

[Jerenny]

Yeah, both cards 1 V under load.

I'm attempting it right now, 1.0 V is the lowest stable undervolt I can get in heaven, I think I can go a little further to .987 before it gets a bit too unstable on everything that's not Heaven Benchmark. I'm trying out 1158(+100MHz) on 1.050 V now to see if that's stable. I believe my normal overclock of 1228/2700 is stable on stock voltage, 1.100V.

[octiceps]

Nothing beats Heaven for weeding out unstable overclocks. If my OC can pass Heaven at max settings overnight, it can pass anything. The only issue is that sometimes instability doesn't manifest itself until an hour or two (or five LOL) into the test. It's definitely not a quick-and-easy benchmark, especially for dialing in those last few MHz, but it's extremely effective.

[Jerenny]

Nothing beats Heaven for weeding out unstable overclocks. If my OC can pass Heaven at max settings overnight, it can pass anything.

True, and drivers just crashed at 1150/2500 @ 1.050V

[octiceps]

Just curious, what is your GPU ASIC quality? Mine is 78.0% and 84.5% on primary and Ultrabay respectively.

[Jerenny]

Just curious, what is your GPU ASIC quality? Mine is 78.0% and 84.5% on primary and Ultrabay respectively.

No clue what that is, where do I find it? And 1150/2500 seems stable on 1.075.

Never mind, and they're pretty bad, 73 and 68 respectively.

[octiceps]

No clue what that is, where do I find it? And 1150/2500 seems stable on 1.075.

In GPU-Z, right-click on the top of the window and select "read ASIC quality." You have to do it for each GPU.

EDIT: I see. It makes no sense why they would select leakier chips for the 750M and 755M if the manufacturing process is supposed to mature over time. That probably explains the high default voltage. I've been around the forums a while and the general trend seems to be that 750M/755M have equal but often lower ASIC quality than the 650M in the first-gen Y500/Y400. Maybe they're getting rid of the last of the GK107 wafers left, i.e. the lemons? Actually, it's not so far-fetched if we look at the generally underwhelming 880M compared to the 780M.

[Jerenny]

In GPU-Z, right-click on the top of the window and select "read ASIC quality." You have to do it for each GPU.

73.0% for first GPU, 68.3% for the Ultrabay card. So that's why I can't undervolt as much as you guys, huh? But in the end I can overclock/overvolt higher?