Testing: Thermal Interface Material vs no Thermal Interface Material. Athlon 64 3200+
I recently got bored and had an Athlon 64 3200+ 2.0GHz processor lying around. So out of curiousity, I decided to do some testing. I replaced the Athlon IIx4 620 in my Asus M4N785-M motherboard with the 3200+ and the 3200+ stock HSF. Actually, to be more accurate, it was the stock fan for an Athlon 64 x2 socket 939 that my friend had. I used Zalman ZM-STG1 tim. The rest of the relevant system setup was:
AMD Athlon 64 3200+ stock voltage and clockspeed.
Asus M4N785-M AM2+ motherboard
4GB DDR2 800MHz memory- mushkin and OCZ 1GB sticks
Antec Three hundred case with side panel open. Top 140mm and bottom 120mm fans on low.
To test for idle I let the computer idle at the desktop once windows had completely loaded, for 1-2 minutes. For load, I let prime95 run for 5 minutes and took the highest ”core” reading. HWmonitor was used for testing. For the second test, I removed the tim and properly cleaned the HSF and CPU heatspreader. I then reseated without applying any thermal compound.
Here are the results with the zalman compound installed properly: stock speeds. 2GHz. TIM
Quick stable overclock. 2.5GHz. no overvoltage. TIM
no TIM: stock voltage and speed
I then decided, what the hell. I’ll test with the HSF loosened. The following are the results with the HSF completely loosened and hanging onto the clip. The case is upright, so gravity is not holding the HSF to the CPU. The results are the same as when I held the HSF away from the CPU but still holding onto the loosened clip so all there is between the HSF and CPU is air.
Max CPU temperature without HSF:
The reason for the idle range is because I was too impatient to let it settle down. So I counted the highest temperature reached starting at 30C and climbing. I then started at 115C and let it climb down and recorded the 96C.
I noticed no signs of thermal throttling during the previous test. Either the CPU speed changing in CPU-z or some other kind of slowdown.
Following this test, I unclipped the HSF and brought it away from the CPU so it was completely clear. Within a second my monitor showed “no input”. I put the HSF back on and there was no change. I rebooted with the HSF on and everything was fine again. So thermal protection for AMD Athlon 64 CPUs is above 115C apparently.
I just find it interesting that the difference between no TIM and a good TIM was 4C at the most in this test. I may run this same test with my Athlon IIx4. Of course, I will not be removing the heatsink during that test lol. If I do, I'll post my results here in an edit.
Edit: forgot to include Ambient temperatures: 19C
By comparison, at this temperature, my Athlon II 620, hits 51C on prime95. At 25C, it hits 55C load with the same thermal compound and th Athlon IIx4 stock HSF.
It's a slow single core processor and I believe those had the big thick heatsink on top of the chip. Not to mention 5 minutes is not long enough to measure temps in prime95. Once the processor heats up, that slowly raised the temps of the other components inside the case, which in turns heats up the processor even more. Try 1 hour.
The results will be much different with your X4 620.
Indeed. However, as I mentioned before, there are two massive fans above and at the back of my case, and my case side panel was open.
My buddy has a machine shop and I was considering taking lapping to the extreme and using the Bridgeport. In theory by doing this to both the heat sink and CPU, no TIM would be needed. In fact it should work better without it, as the surfaces would be perfectly flat.