How can I compare my new thermal paste?

[trzasku]

I want to clean my entire case, my Dark Rock Pro 3 and all the fans I have. Then I want to change my old thermal paste to my new MX-2 using the pea method. I want to see if this will make a change in temperatures and how much.

• How should I define the temperature that I have at this moment, before the changes?
• What software should I use to test the temperature my CPU has on load and idle?
• Should I perform all the benchmarks on 100% fan speed or another value? 
• Is the ambient temperature important? Will the CPU be 2"C cooler when the room is 2"C cooler?

[ARikozuM]

Wait, is this about toothpaste or thermal paste?

[martward]

11 minutes ago, trzasku said:

I want to clean my entire case, my Dark Rock Pro 3 and all the fans I have. Then I want to change my old thermal paste to my new MX-2 using the pea method. I want to see if this will make a change in temperatures and how much.

• How should I define the temperature that I have at this moment, before the changes?
• What software should I use to test the temperature my CPU has on load and idle?
• Should I perform all the benchmarks on 100% fan speed or another value? 
• Is the ambient temperature important? Will the CPU be 2"C cooler when the room is 2"C cooler?

1) The temperature should be defined in Celcius I guess.

2) I like to use OpenHardwareMonitor to check temperatures, if you want to check the load temps you can use Aida64.

3) I'd not temper with the fan speeds, you normally don't have your fans spinning at 100% on idle so it doesn't make sense to do it for comparing.

4) The ambient temperature is important, but I suppose you can't scale it like that. Just perform your tests, change the thermal paste and then test again when the temperature is about the same as when you first did your test.

[trzasku]

• OpenHardwareMonitor looks like HWMonitor that I use, is this the same or is OHM better?
• Is Aida64 the best software to see what temperatures my CPU reach on load?
• I usually run my fans on 40% because I have a silent build but for test purouses I think 100% will be more reliable  (because it is more consistent) if I want to see if the new thermal paste will make any difference, am I wrong?

[ARikozuM]

6 minutes ago, trzasku said:

• Is Aida64 the best software to see what temperatures my CPU reach on load?

Aida64 is my preferred software as it includes a stress test utility.

6 minutes ago, trzasku said:

• I usually run my fans on 40% because I have a silent build but for test purouses I think 100% will be more reliable  (because it is more consistent) if I want to see if the new thermal paste will make any difference, am I wrong?

Proper methodology is to use as close to a real-world scenario (or as consistent as possible) as you can get. Running the fans at 100% would be good if you intend on leaving them there after testing. You should turn off any fan curves though as they may produce inconsistencies in your results.

[trzasku]

You Sir are right, I think I will perform both 40% and 100% tests. Another important question.

 

How long should the software run to reach the maximum temperature? Is it safe?

[ARikozuM]

Just now, trzasku said:

You Sir are right, I think I will perform both 40% and 100% tests. Another important question.

Go ahead.

Just now, trzasku said:

How long should the software run to reach the maximum temperature? Is it safe?

If it's just a temp test, go for as long as you can until the temp stays stable for 5 or so minutes. After that, use the same time. Make sure you only stress CPU and FPU.

[done12many2]

3 minutes ago, ARikozuM said:

Go ahead.

If it's just a temp test, go for as long as you can until the temp stays stable for 5 or so minutes. After that, use the same time. Make sure you only stress CPU and FPU.

 

FPU only if you want to have some real fun.  

[trzasku]

I do not even know what FPU is. I will read some articles how to make a load test with AIDA64.

[ARikozuM]

1 minute ago, trzasku said:

I do not even know what FPU is. I will read some articles how to make a load test with AIDA64.

Nobody does... It's just a legend, a fairy tale.

[DHelios]

38 minutes ago, trzasku said:

I want to clean my entire case, my Dark Rock Pro 3 and all the fans I have. Then I want to change my old thermal paste to my new MX-2 using the pea method. I want to see if this will make a change in temperatures and how much.

• How should I define the temperature that I have at this moment, before the changes?
• What software should I use to test the temperature my CPU has on load and idle?
• Should I perform all the benchmarks on 100% fan speed or another value? 
• Is the ambient temperature important? Will the CPU be 2"C cooler when the room is 2"C cooler?

The main problem here, is that you are intending to clean your case and fans. So if you measure now, and then again after you're done changing things, you'll not just measure the difference your thermal compound makes. Things may also be affected by having clean fan blades, clean heatsinks, and the like. Also, you'd be testing baked-in old thermal compound compared to freshly applied new compound. You'll want to clean your system thoroughly, then re-apply a fresh pea of your old thermal paste. Measure this, and use this as your before temp.

 

Use any software that shows the temperature. HWMon will do just fine.

 

Whatever speed you set the fans to should be consistent between tests. Try to keep as many things even as possible, including fan speed. So if you test before at 100%, test after at 100% as well. Same with 40%. If you always have your fans set to 40%, that may be a good setting to test with. As this will not only show if the thermal compound change makes a difference, it'll show if the thermal compound change makes a difference for you.

 

Ambient temperature is important. If you can keep it even, do so. If not, make sure to record the ambient temperature along with your other findings. The values you're looking to compare before and after are the 'rise over ambient': how much hotter is your CPU compared to the ambient temp. So if your room is 20°C and the CPU reads 60°C, that's a 40° rise. If your second test has the room at 23°C and the CPU is 62°C, the CPU is hotter but it's only 39° over ambient, so the thermal solution is doing a better job.

[done12many2]

5 minutes ago, trzasku said:

I do not even know what FPU is. I will read some articles how to make a load test with AIDA64.

 

Floating Point Unit (FPU), which is the part of you CPU design specifically for fast math calculations.  Stressing it alone will produce the highest temps in AIDA64 or any stress test that focuses on in.  You just need to make sure that your cooling capacity is up to par before running this test only for extended periods of time.  It gets a lot hotter.

 

You still haven't addressed the "toothpaste" in your title.

[DHelios]

44 minutes ago, ARikozuM said:

Wait, is this about toothpaste or thermal paste?

In a pinch, regular white toothpaste will substitute for thermal paste.  It's not as efficient as actual thermal compound designed for the job, but it's better than no thermal interface material at all. Your computer may smell minty fresh for a few days, but that will fade.

 

That said... just spend the few bucks to get actual thermal paste =)

[done12many2]

1 minute ago, DHelios said:

In a pinch, regular white toothpaste will substitute for thermal paste.  It's not as efficient as actual thermal compound designed for the job, but it's better than no thermal interface material at all. Your computer may smell minty fresh for a few days, but that will fade.

 

That said... just spend the few bucks to get actual thermal paste =)

 

We were both jokingly pointing out the OP's error in his title, but thanks for breaking that down.  

[DHelios]

6 minutes ago, done12many2 said:

 

We were both jokingly pointing out the OP's error in his title, but thanks for breaking that down.  

I'm quite sure the OP was either joking, mistyped or had a case of bad autocorrect. Still, as far as mistakes go it could be worse. My Athlon XP ran for years with nothing but Prodent between the CPU and HSF =)

[done12many2]

1 minute ago, DHelios said:

I'm quite sure the OP was either joking, mistyped or had a case of bad autocorrect.

 

You can't go wrong with a sentence like that.  You've got all the bases covered.

[trzasku]

7 hours ago, done12many2 said:

 

You can't go wrong with a sentence like that.  You've got all the bases covered.

Wrong. Nothing of this is true, it was a simple clickbait, I am very sorry. But thank you guys for help. Last question.

 

Should I perform the benchamrk on an open or closed case? If I do a stress test then the next test will be done in a hot environment because the air in the case will be still hot. Mayby I shut down the PC and wait for some minutes?

 

And should I turn of my AUTO CO for the test time?

[ARikozuM]

1 hour ago, trzasku said:

Should I perform the benchamrk on an open or closed case? If I do a stress test then the next test will be done in a hot environment because the air in the case will be still hot. Mayby I shut down the PC and wait for some minutes?

Closed case is always a better performer as the air can be channeled through efficiently.

[mad dudy]

do your test outdoors at night, if your stress test rise the ambient temp, your the cause of climate change and should feel bad.