Does anyone actually have Link working?!

[Geth]

Guys, does anyone actually have Corsair Link working properly on Window 7 64-bit and using the CL Commander and Cooling Node such that their fans automatically adhere to a custom curve on system startup without any user intervention such as manually starting the CL console?

 

I have spent the last several hours documenting across several pages tests and results, and the closest I can get is CL v26.5214 with Cooling Node v1.2.5 with the fans only adhering to the my custom curve once I've manually started the CL console.

 

Someone, please tell me that you have it working completely on it's own and how you did it.

[wytnyt]

yes 4x as shown in my sig but those without issues have no need to come here asking for help...

[thenosbod]

I have no problems with mine.

[Sunwindion]

Nope. The previous version works (barely), the latest version just crashes whenever I run it. I've formatted my PC twice (not for Link) and the latest version never works properly for me. Corsair need to take their finger out and update it.

[Fauxbeard]

[Geth]

I have no problems with mine.

 

Nope. The previous version works (barely), the latest version just crashes whenever I run it. I've formatted my PC twice (not for Link) and the latest version never works properly for me. Corsair need to take their finger out and update it.

 

(attached images)

 

You guys have the fans automatically adhering to a custom curve as soon as Windows starts without having to open the console? How do you make it do that? As I said, the best I can make it do is to adhere to the curve only after I have started the console. Do you have it tasked for startup? I tried that but it still didn't adhere on startup, and when I opened the console afterwards to get it to adhere, it created duplicate processes for both CorsairLink.exe and CorsairLINK.exe with, I believe, those being the underlying service and the frontend/console respectively.

 

I get the impression that the current settings are supposed to be saved on the Cooling Node so CL can get everything running without the console being opened. That's how my H100i (v2.4.5110) seemed to work when I used it as the controller instead of the Commander. So why won't CL do that now with either v2.4.5110 or v2.6.5214? Since installing the Commander and Cooling Node, v2.4.5110, on the curve graph, only the H100i temp axis changes but the RPM axis remains where it is. However, with v2.6.5214, both axis correctly respond in unison, adjusting the fans to the temp to follow the curve, but only after I have manually started the console.

 

Is there something I'm doing wrong? Should I downgrade my Commander firmware? If so, where can I find an older version? I'm currently using Cooling Node 1.2.5 but I have tested 1.2.11 and 1.2.9 as well.

 

 

 

Edit: Attached the handwritten notes/tests/results from last night for:

• CL 2.4.5110 - Node 1.2.9(8) & 1.2.5
  
• CL 2.6.5214 - Node 1.2.5
  

I tested Node 1.2.11 before I started taking notes.

 

Key:

• "Reboot" = computer cold-boot.
  
• "Stress" = manually increased H100i temp by running Prime95 to heat up the CPU.
  
• "Console" = manually started CL console.
  
• "adhere" = fans adhered to the custom curve.
  
• "not adhere" = fans did not adhere to the custom curve.
  
• "firmware" = updated Cooling Node firmware to the version shown at the top of the page.
  
• "Reflash" = reflashed Cooling Node firmware to the version shown at the top of the page (usually to reacquire unresponsive nodes/devices).
  
• "- X -" = undesired result.
  
• "- ✔ -" = desired result.
   
  
• Items/events immediately after a Reboot = observed results after the reboot and before manual action is taken.
  
• Indented items/events = sequential actions and observations as the test is performed.
  

CLink testing notes.pdf

[Geth]

Can someone please tell me how to get CL to make the fans follow the custom curve on startup without manually starting CL? I've spent over 24 hours working on this and any help regarding how to fix the problem would be greatly appreciated.

[Tech_Greek]

Can someone please tell me how to get CL to make the fans follow the custom curve on startup without manually starting CL? I've spent over 24 hours working on this and any help regarding how to fix the problem would be greatly appreciated.

 

 

I ended up changing the task scheduler to point to the corsair link exe and then run as a hidden task (Windows 8.1).

 

Now I don't have to see the annoying "corsair link is still running" every...single...time...I start windows and minimize link.

 

Set my profiles for temp monitoring on the LED and then saved the profile to never open link again (at least until they fix it).

[Geth]

After countless hours this week (ie. all my free time after work up to 3am every night), I have finally gotten CL to work such that the fans adhere to the custom curve on startup without having to manually start the console. But I don't know how and the only thing that I can think of that differs from previous attempts that have failed is that the registry key "UIFirstLoaded" is now set to True, though I don't know what this key does.

 

Anyway, I'll attach a zip containing screenshots of my settings and the scheduled startup task (I modified the task that CL created when "Run at Windows Startup" was selected in the CL console Options tab) along with my own .bat files (outlined below), and I'll try to give details of the settings that ended up working in the hopes that it may help others.

 

 

The software and firmware that worked for me are Corsair Link v2.6.5214 and Cooling Node v1.2.5 (attached), and I'm using Windows 7.

 

After startup, there are two CL processes running that can be seen in the Taskmanager:

• CorsairLink.exe - I think this is the CL service/manager and must be running for CL to monitor and adjust the devices. Located in "[install directory]\Corsair\Corsair Link".
  
• CorsairLINK.exe - The CL console window. I found that this must be started before the fans will adhere to the custom curve, apparently even if the window isn't actually opened on screen and the icon does not appear in the tray. Located in "[install directory]\Corsair\Corsair Link\Frontend".
  

 

However, since there is no icon in the tray at this point, the only way to reach the console and see the settings or even the devices and node themselves is to open the console via CorsairLink.exe or CorsairLINK.exe (see above for the difference), right? Well, yes and no. The problem is that, while it will open the console, is will also create duplicates of the above processes, which, for me, led to .NET Framework errors and made the devices completely unresponsive to CL. Therefore, the existing processes in must first be ended in Taskmanager, which will probably cause the devices to stop adhering to your setting, as the governing CorsairLink.exe process is no longer running.

 

So, how to get around this? My solution was to create a batch (.bat) file (below) to end the processes and then restart them immediately, and I changed my Corsair Link shortcuts to run this instead of CorsairLink.exe. The downside is that this sometimes causes my fans to stop responding and to appear at 0 RPM in the console. Don't fret yet; there's an easy solution: go to the Devices tab in the console and click the "Check for Updates" button at the top, which will force the Firmware Updater window to appear next time you start the console. Now, run the .bat again to reload CL and, when the Firmware Updater window appears and asks you to update your Cooling Node firmware, click Cancel (do not update to the same or different Cooling node version). For me, this seems to make the Cooling Node reasses what's connected to it and the fans become responsive in the console again. Keep in mind that an additional Corsair Link icon tends to appear in the tray every time you use the reload .bat file but you can roll your cursor over them to make them disappear.

 

Important: using the reload .bat file instantly ends both processes. You WILL NOT be asked if you want to save you current setting to your profile. It's probably a good idea to always close the console window manually first before you use the .bat file to reload CL.

 

Reload_Corsair_Link.bat

 

Taskkill /f /im "CorsairLink.exe"

Start "" "C:\Program Files (x86)\Corsair\Corsair Link\CorsairLink.exe"

 

Remember that a scheduled task is created in your Windows Task Scheduler when you select "Run at Windows Startup" in the CL console's Options tab. Take a look at this task and make sure that it's configured for your version of Windows (General tab in the task properties). See my attached screens for the settings that I used.

 

I also use another .bat file that ends only the CorsairLINK.exe (CL console) process, if it's running, while leaving the CorsairLink.exe process (CL service/manager) running. I don't think this is absolutely necessary to use but I have it set up in another task to run at startup with a 30 second delay so CorsairLINK.exe won't be using my system memory unnecessarily after getting the fans to initially adhere to my settings. See it below:

 

Stop_CL_Console.bat

 

@echo OFF

SETLOCAL

FOR /f "tokens=2" %%i IN (

' tasklist ^| find "CorsairLINK.exe" '

) DO SET pid=%%i

IF DEFINED pid TASKKILL /f /pid %pid%

 

 

Good luck fellow traveller.

My working Corsair Link files and settings.zip

[Tech_Greek]

Here's my working task scheduler fix (for booting CL without an icon, but still having all devices, functions and fans working) on Windows 8.1.

 

You will NOT have a task icon by the clock but the service will be running (hidden).

 

Basically, I set my settings and then just stopped checking the program since it's so haggard right now with icon issues, stability, and overall problems. I know the service is running because my H100i LED adjusts based on the CPU Core Avg temp as it's supposed to (and if it's not running, it won't).

 

<SNIP>

 

Replace the YOURCOMPUTERNAME\YOURUSERNAMEHERE with your obvious settings and then import the task (disable the old one, verify this works before deleting the old one).

 

Geth, you are spot on in your assumption of the different EXE. The front end folder is strictly for the UI while the base folder EXE is for the service (hardware monitoring). Not sure why Corsair took this out of an actual Windows Service like it was in the Stable RC...

 

I've got BOTH exe on run as admin and windows 8 compatibility mode since it seems to be the most stable as such (even though the Task should call it as an admin it doesn't seem too happy to do that).

 

I'm not even sure why we are having to figure this out. Corsair, can you just go ahead and put us on payroll? :sunglasse

 

EDIT: I've upgraded to the latest version, I highly encourage everyone to uninstall the prior version and delete the scheduled task before installing the new one. Much more stable. I snipped the prior code so no one messes up.

[Geth]

I'm not even sure why we are having to figure this out. Corsair, can you just go ahead and put us on payroll? :sunglasse

 

Amen to that. I already did my time as a hardware and software tester for credit card swipe terminals and I hated it. Why must we settle for the beta versions of software for hardware that we have already dropped $100+ for? And then why are we expected to test it for Corsair and post the bugs that we find here on the forum? I wasn't paid enough then and I'm not being paid at all now to debug and create workarounds for products that I've already lost money on.

[wytnyt]

its amazing how some have had so many problems with the software/hardware,,i honestly think it has to do with a ''compound effect'' as in less than 15 minutes im up n running with set profiles

 

i even used one of my computers to run the various versions of software and had favorable results..

[Tech_Greek]

Amen to that. I already did my time as a hardware and software tester for credit card swipe terminals and I hated it. Why must we settle for the beta versions of software for hardware that we have already dropped $100+ for? And then why are we expected to test it for Corsair and post the bugs that we find here on the forum? I wasn't paid enough then and I'm not being paid at all now to debug and create workarounds for products that I've already lost money on.

 

Yep. I own an IT Consulting company and do custom builds all the time. As of right now, I can't recommend nor install it in anyone's system because it'll end up coming back to me the first time the software breaks like it does.

 

wytnyt, some more so than others are having issues due to other things going on in their system but on a fresh install of Windows 8.1 with all patching done and nothing else but Corsair Link you shouldn't have to be doing the fixes that I've had to do.

[Geth]

Hey, Tech, I'm still not having my fans sync with the curve on startup/logon something like 95% of the time. The massive outline that I previously posted above works to keep the CL stable and resync the fans when they get lost but that's it. Any ideas?

[Tech_Greek]

Hey, Tech, I'm still not having my fans sync with the curve on startup/logon something like 95% of the time. The massive outline that I previously posted above works to keep the CL stable and resync the fans when they get lost but that's it. Any ideas?

 

Please post a picture of the your Link software tabs

 

(Hint: Use the Microsoft Snip It Utility)

[Geth]

Please post a picture of the your Link software tabs

 

(Hint: Use the Microsoft Snip It Utility)

 

Here's my setup.

[Tech_Greek]

Your fans are currently grouped with the h100i coolant temperature and pump speed which will mess with it.

 

Group it to the i7-4770k (Group 10) and then set your curves. Should solve that one!

 

Also, if you want to use the temperature scaling on the h100i LED properly, group it to the i7-4770k - this is a core avg that is much more accurate than the regular CPU temp.

 

Sorry about the delay, we were swamped this week.

[Geth]

The fans are grouped with the H100i temp because I want them to react to the coolant. Binding them to the CPU temp makes them jump too often because the CPU temp changes so quickly frequently and can go from 30 straight to 34 or higher and then back to 30 at idle.

 

However, what curve values would you advise if I were to adhere it to the CPU while using SP120s with a max usable RPM of 2100 (much faster and CL is known to drop them back to < 800)?

 

The image shows my system at 100% load. It's winter and the room temp < 20C here, so that's why it's not exceeding H100i = 34C and i7-4770K = 62C with current setup. If I opened the case, it would be cooler still.

 

Also, CL always finds the H100i at startup; it's the SP120 fans that it often misses until I reload CL altogether and... (see post 9).

 

[LeRoy3rd]

Here's the thing, the coolant temp stays very stable, which is why the fans don't change often when linked with the coolant. The CPU temp, on the other hand, can go from cool to dangerous in less than a second. The only thing keeping your CPU from melting down is the air moving through the radiator... So, if you want a silent PC either:

 

a) don't ever let the CPU get hot (aka never use the PC)

b) get quieter fans

c) Set the curves properly

d) continue to improperly cool your CPU (it will take care of itself! ;-) )

 

Seriously though, you NEED to have your CPU temp linked to your H100i fans. The one thing I noticed is that your curves are set WAY too low (you're going to approx 100% at 39c?? I've seen many air cooled systems that aren't that cool at IDLE temps!). Do some benchmark/CPU burn-in tests, and watch your CPU temps... This is the most accurate way to dial in your settings. A couple of notes to make it easier:

 

I usually set my fan speed to the minimum while at idle speed. Test fan speeds and see how low you can go while still keeping them on (do this for each fan separately as it will be slightly different for each fan), write this down as minimum. Next, set to max, look at RPMs, and write this down as max (do this for each fan separately as it will be slightly different for each fan). Next, set the fan speeds to be a minimum when the CPU is at idle temps (whatever your idle temps are). Then watch the temp and make sure that it doesn't go up at idle with this fan speed. If it does, adjust the idle speed higher accordingly until you have a stable idle speed.

 

Next run something HARSH like prime95 on "In-place large FFTs"... I do 3D modeling and rendering, which uses 100% CPU on EVERY thread, and it doesn't come close to creating the heat that this test does. Make sure your fans are running at max during this test... Watch temps to make sure they don't go too high. Usually you'll want them around or below 70 degrees c at this point, though with a test like this, even around 75c is acceptable.

 

You now have a minimum and maximum fan speed. Set minimum at idle temps, max at about 70-75c. The others? Adjust to taste. As an example, I've attached a screenshot of my fan speeds. It should be noted that your results will be different since you're using the stock fans (I'm not). Also, these are set to give me the best overhead for overclocking (needed for 3D rendering, even a 1% overclock will literally save me HOURS in a render), so you can probably set your speeds even lower. Finally, it should be noted that even at these settings, at full speed, my rig is very quiet, I specifically replaced the fans it came with to give my ears some rest. If you do replace your fans, find something with a high static pressure, and low noise, you'll probably spend around $20 each (including tax and shipping), but you'll be much happier.

 

 

-LeRoy3rd

[wytnyt]

The fans are grouped with the H100i temp because I want them to react to the coolant. Binding them to the CPU temp makes them jump too often because the CPU temp changes so quickly frequently and can go from 30 straight to 34 or higher and then back to 30 at idle.

 

 

this is the correct way to watercool

if the cpu temp is getting too hot ,raising the fan speed will not help if the water is already cool.

if this is the case then something is causing this such as too high a oc,a voltage setting,,,ect

many think using the cpu is best but quite frankly their just ms-informed...

[Tech_Greek]

this is the correct way to watercool

if the cpu temp is getting too hot ,raising the fan speed will not help if the water is already cool.

if this is the case then something is causing this such as too high a oc,a voltage setting,,,ect

many think using the cpu is best but quite frankly their just ms-informed...

 

I disagree. I set up server cabinets and IT Infrastructure for a living (I went into business when I was 23) so this is my world that I stay in and that I excel at - not trying to blow my own horn here but just giving you a back drop of why I'm posting this.

 

Would you rather engage the AC inside of your house when it hits 110 F and cool it down to 73 F or would you rather kick the AC on at 75 F and work it's way back down to 73 F? The principal is still the same, you're still transferring the heat to liquid - just takes a much longer path and has freon to aid cooling BELOW the ambient temperature around it. Your cars AC also engages the eFans on high even when it's cold outside or on a cold start in the afternoon. The airflow helps bleed off temperatures versus no air flow at all.

 

You've got to take into account that once your cores spike (which they do, instantly when OC) that your temperature is not going to spike in the water cooler just as fast, you would be beating the thermal transfer that isn't possible even on copper that's been polished. If you've already got airflow pushing air through the radiator you're going to easily bleed off the heat as it rises through the loop versus trying to blow it off.

 

Another easy analogy: Would you rather accelerate before a hill in the car so that the car can use less power to get up the hill or would you rather wait until you hit the hill to hit the gas? If your answer is wait until the hill, your bad gas mileage probably reflects your decision. :biggrin:

 

I also managed to get my little 3570k to 5.0 Ghz without delidding it on a H100i with a 50-60* distance to tjmax under load. People thought I was crazy for doing Quiet PWM SP120 in the front versus AF140 or AF120, but considering my fans only kick on when gaming and I've hit the speeds I've hit I think I'm doing pretty good. Hell, I managed to get to 4.8 with the same temperatures with an h55 and one SP120 Quiet fan - the only reason I even went for the h100i was because I was browsing through Best Buy and had some rewards zone points to spend.

 

So, this is why I group my fans to my core temps and why my LED is grouped with it as well. I don't want to wait on the coolant temperature to peak to an unsafe temperature before I shut the system down, I want an instant read out and instant speed reaction on the fans.

 

Tldr? Don't fight peak temperatures, bleed them off as they rise, prevent spikes and increase total efficiency.

 

I tend to go against the 'common belief' system that most people follow if it doesn't seem right to me. I was told there was zero way I could get below Ambient temperature in my server cabinets without CRAC (or even get TO ambient air temperatures in the cold/front end of the cabinet).

 

This is at 73* F inside the room...that's 65* F on the server intake with a 2* split from the bottom to the top (top was 66, bottom was 64).

 

http://imageshack.com/a/img908/1218/PbcuXh.jpg

[wytnyt]

I disagree. I set up server cabinets and IT Infrastructure for a living (I went into business when I was 23) so this is my world that I stay in and that I excel at - not trying to blow my own horn here but just giving you a back drop of why I'm posting this.

 

Would you rather engage the AC inside of your house when it hits 110 F and cool it down to 73 F or would you rather kick the AC on at 75 F and work it's way back down to 73 F? The principal is still the same, you're still transferring the heat to liquid - just takes a much longer path and has freon to aid cooling BELOW the ambient temperature around it. Your cars AC also engages the eFans on high even when it's cold outside or on a cold start in the afternoon. The airflow helps bleed off temperatures versus no air flow at all.

 

You've got to take into account that once your cores spike (which they do, instantly when OC) that your temperature is not going to spike in the water cooler just as fast, you would be beating the thermal transfer that isn't possible even on copper that's been polished. If you've already got airflow pushing air through the radiator you're going to easily bleed off the heat as it rises through the loop versus trying to blow it off.

 

Another easy analogy: Would you rather accelerate before a hill in the car so that the car can use less power to get up the hill or would you rather wait until you hit the hill to hit the gas? If your answer is wait until the hill, your bad gas mileage probably reflects your decision. :biggrin:

 

I also managed to get my little 3570k to 5.0 Ghz without delidding it on a H100i with a 50-60* distance to tjmax under load. People thought I was crazy for doing Quiet PWM SP120 in the front versus AF140 or AF120, but considering my fans only kick on when gaming and I've hit the speeds I've hit I think I'm doing pretty good. Hell, I managed to get to 4.8 with the same temperatures with an h55 and one SP120 Quiet fan - the only reason I even went for the h100i was because I was browsing through Best Buy and had some rewards zone points to spend.

 

So, this is why I group my fans to my core temps and why my LED is grouped with it as well. I don't want to wait on the coolant temperature to peak to an unsafe temperature before I shut the system down, I want an instant read out and instant speed reaction on the fans.

 

Tldr? Don't fight peak temperatures, bleed them off as they rise, prevent spikes and increase total efficiency.

 

I tend to go against the 'common belief' system that most people follow if it doesn't seem right to me. I was told there was zero way I could get below Ambient temperature in my server cabinets without CRAC.

 

 

your analogy of various sorts really has nothing to do in comparison to computers

while i dont have your vast amount of experience in IT ect...,i can unequivocally say without a doubt that you have the wrong perspective in water cooling a pc.

do you honestly think turning up your fans is going to cool an already cooled water loop?

have you ever noticed your cpu heat dropping by turning up fans?,,i highly doubt it

i ask anyone to load their computer up ,then crank up the fans and see if cpu temps drop,,perhaps a couple degrees at best if even that.

this idea of cooling the cpu your way is a misconception by those that are used to cooling by air and WC is totally different.

in laymen terms

your water is say 35c,,you hit a hard load and the cpu instantly heats up

your water is still going to be 35c so what good does it do to spin up the fans?,,if its at its lowest temp of 35c at idle,then the fans wont lower it under load.

now keeping the hard load will obviously eventually heat the water so fans are needed to maintain a lower water temp as the temp raises.

its amazing how the proper concepts of watercooling eludes so many people

 

another very simple analogy

water temp is 35c

you load it and keep the fans low and the water temp maintains 35c

are higher rpm fans needed?

of course not

but IF AND WHEN the water temp rises will rpm's be needed

 

keep a open mind and think about it...;):

[Geth]

Here's the thing, the coolant temp stays very stable, which is why the fans don't change often when linked with the coolant. The CPU temp, on the other hand, can go from cool to dangerous in less than a second. The only thing keeping your CPU from melting down is the air moving through the radiator... So, if you want a silent PC either:

 

a) don't ever let the CPU get hot (aka never use the PC)

b) get quieter fans

c) Set the curves properly

d) continue to improperly cool your CPU (it will take care of itself! ;-) )

 

Wow. Seriously? You expect someone to accept your advice when you're a condescending dick about it?

 

Seriously though, you NEED to have your CPU temp linked to your H100i fans. The one thing I noticed is that your curves are set WAY too low (you're going to approx 100% at 39c?? I've seen many air cooled systems that aren't that cool at IDLE temps!). Do some benchmark/CPU burn-in tests, and watch your CPU temps... This is the most accurate way to dial in your settings. A couple of notes to make it easier:

 

Look at my previously attached screen. That was taken during a benchmark using Prime95. Not sure exactly what you're tying to say about air-cooled systems (since I'm using watercooled) but I wouldn't expect them to be 39C at idle... because they're aircooled. Mine (the CPU), however, idles between 28C and 36C. Remember that my fans are currently set to the coolant temp, not the CPU.

 

I usually set my fan speed to the minimum while at idle speed. Test fan speeds and see how low you can go while still keeping them on (do this for each fan separately as it will be slightly different for each fan), write this down as minimum. Next, set to max, look at RPMs, and write this down as max (do this for each fan separately as it will be slightly different for each fan). Next, set the fan speeds to be a minimum when the CPU is at idle temps (whatever your idle temps are). Then watch the temp and make sure that it doesn't go up at idle with this fan speed. If it does, adjust the idle speed higher accordingly until you have a stable idle speed.

 

Next run something HARSH like prime95 on "In-place large FFTs"... I do 3D modeling and rendering, which uses 100% CPU on EVERY thread, and it doesn't come close to creating the heat that this test does. Make sure your fans are running at max during this test... Watch temps to make sure they don't go too high. Usually you'll want them around or below 70 degrees c at this point, though with a test like this, even around 75c is acceptable.

 

As I said previously, my temps do not exceed 62C at full load, and that isn't even using the full capability of the fans.

 

You now have a minimum and maximum fan speed. Set minimum at idle temps, max at about 70-75c. The others? Adjust to taste. As an example, I've attached a screenshot of my fan speeds. It should be noted that your results will be different since you're using the stock fans (I'm not). Also, these are set to give me the best overhead for overclocking (needed for 3D rendering, even a 1% overclock will literally save me HOURS in a render), so you can probably set your speeds even lower. Finally, it should be noted that even at these settings, at full speed, my rig is very quiet, I specifically replaced the fans it came with to give my ears some rest. If you do replace your fans, find something with a high static pressure, and low noise, you'll probably spend around $20 each (including tax and shipping), but you'll be much happier.

 

Mate. I am using SP120 PEs (not stock fans) at the moment because they are the quietest things available without resulting in resonating vibrations through the rad (even with padding and gaskets on it) at medium speeds and outright rattling at high speeds, and it's a lottery with the SP PEs because some do rattle at max right out of the box. In the past month, I have gone through NB-eloop B-12Ps, Gentle Typhoon AP-15s, H100i stock Corsair SP120Ls, and Noctua NF-F12 iPPCs (IP52). I can tell you the problem with each of these high-quality and expensive ($22 - $35 each) models. Suffice to say returning to using SP120 PEs was no easy decision and was definitely not one made out of laziness. I have gone great lengths to optimize my rig's noise-cooling ratio, focusing specifically on acoustical noise and static pressure.

 

I'd be interested in knowing exactly what fans you're running on your rad, as you're either not running them at their max potential, or their max RPM is < 2000, which would leave the strength of the static pressure a bit lower than what is possible if you were to use other fans (ie the NF-F12 iPPCs boast a 2000 RPM and 3.94 static pressure and the SP120Ls a 2700 RPM with a 4 static pressure, which have a high-pitched whine/hum and rattle when upside-down respectively at max speed). I'd hazard a guess that that's why your rig is so quiet and your CPU temp maxes at around 75C.

 

 

I'm not new to building computers. I've been doing this for years (that includes using CL custom curves) and the only reason I'm asking about setting the rad fans to the CPU temp is because I've always set them to the coolant (as that's what the fans are actually cooling directly, and it's what many people do), and this is the first time someone has suggested using the CPU temp instead, not because I'm new and inexperienced.

 

(And when you're replying to someone asking for help, remember that first impressions apply on a forum just as they do face to face. Being condescending early on will often paint you as such for the remainder of the interaction.)

[LeRoy3rd]

Wow. Seriously? You expect someone to accept your advice when you're a condescending dick about it?

 

 

Whoa, whoa, whoa! Slow down there. I was neither trying to be condescending, nor trying to be a dick. When I sit down to write a response I try to put in as much information as I can while keeping my post light. I have no way of knowing your background or which fans you've tried (this wasn't the purpose anyway). You'd asked how to set the curves to not change so dramatically when using the CPU temp as the bind, I did what I could to write something that would be understood by anyone reading this ( you didn't submit a resume with this).

 

I apologize if you thought I was talking down to you, or being a dick, but I assure you, my purpose in taking the time to write a reply was only to help.

 

-LeRoy3rd

[Geth]

I disagree. I set up server cabinets and IT Infrastructure for a living (I went into business when I was 23) so this is my world that I stay in and that I excel at - not trying to blow my own horn here but just giving you a back drop of why I'm posting this.

 

Would you rather engage the AC inside of your house when it hits 110 F and cool it down to 73 F or would you rather kick the AC on at 75 F and work it's way back down to 73 F? The principal is still the same, you're still transferring the heat to liquid - just takes a much longer path and has freon to aid cooling BELOW the ambient temperature around it. Your cars AC also engages the eFans on high even when it's cold outside or on a cold start in the afternoon. The airflow helps bleed off temperatures versus no air flow at all.

 

You've got to take into account that once your cores spike (which they do, instantly when OC) that your temperature is not going to spike in the water cooler just as fast, you would be beating the thermal transfer that isn't possible even on copper that's been polished. If you've already got airflow pushing air through the radiator you're going to easily bleed off the heat as it rises through the loop versus trying to blow it off.

 

Another easy analogy: Would you rather accelerate before a hill in the car so that the car can use less power to get up the hill or would you rather wait until you hit the hill to hit the gas? If your answer is wait until the hill, your bad gas mileage probably reflects your decision. :biggrin:

 

I also managed to get my little 3570k to 5.0 Ghz without delidding it on a H100i with a 50-60* distance to tjmax under load. People thought I was crazy for doing Quiet PWM SP120 in the front versus AF140 or AF120, but considering my fans only kick on when gaming and I've hit the speeds I've hit I think I'm doing pretty good. Hell, I managed to get to 4.8 with the same temperatures with an h55 and one SP120 Quiet fan - the only reason I even went for the h100i was because I was browsing through Best Buy and had some rewards zone points to spend.

 

So, this is why I group my fans to my core temps and why my LED is grouped with it as well. I don't want to wait on the coolant temperature to peak to an unsafe temperature before I shut the system down, I want an instant read out and instant speed reaction on the fans.

 

Tldr? Don't fight peak temperatures, bleed them off as they rise, prevent spikes and increase total efficiency.

 

I tend to go against the 'common belief' system that most people follow if it doesn't seem right to me. I was told there was zero way I could get below Ambient temperature in my server cabinets without CRAC (or even get TO ambient air temperatures in the cold/front end of the cabinet).

 

This is at 73* F inside the room...that's 65* F on the server intake with a 2* split from the bottom to the top (top was 66, bottom was 64).

 

I get and appreciate your analogies, especially the driving one, as I often question the intelligence of other drivers in front of me who slow down before steep hills instead of increasing their forward momentum at they approach.

 

Assuming that my CPU temp is always between 30C at idle and let's say 70C at load (has never reached than yet), what kind of a curve would you suggest if I were to bind the rad fans to the CPU temp? IE. a smooth flat curve, one that becomes steeper as temp increases, one that steepens and then flattens out again near the max RPM, etc.? Or what if I took my current curve for the H100i temp and applied it to the CPU temp and changed the temps to range from 29 to 60?