I just read Computronix's awesome Core 2 Duo temperature guide. I just completed my first build, so I am a complete n00bie, but trying to learn quickly. Computronix's guide was outstanding :trophy: . I did a lot of searching, and there is nothing comparable anywhere else. After reading this I realized that most of the discussions I have seen on Core 2 Duo temperatures are nonsense: many people are comparing apples to oranges! (i.e. comparing someone's Tjunction readings to someone else's Tcase readings). As a n00bie who feels overwhelmed with information that is sometimes not very reliable, all I can say is THANK YOU for taking the time to write this!
In any case, after reading the guide I wanted to set some benchmarks for how my system behaves so I ran some tests. I am using TAT and Speedfan to read the relevant values (temperature, voltages) and I started getting some strange results. I know from the thread that the Asus P5B (my MoBo) seems to have some problems with accurate temperature readings, so maybe this will help shed some light on some of the problems and possible solutions.
First of all, here are my system specs
Ambient = 20C
Chipset = P965
C2D = E6600
CPU Cooler = Intel stock HSF
Frequency = 2400
Load = TAT @ 100% for 10 minutes
Motherboard = Asus P5B Deluxe
Everything is inside an Antec P180 case, so the PSU and HDD are in a separate chamber from the CPU/Mobo. The CPU/Mobo chamber has 2 120mm fans at the top/rear of the case, both in exhaust (the standard fan locations for the P180B). All fans set to manual and each fan moves about 39CFM of air. I spent a couple hours doing cable management, so the inside of my case is fairly clean. All this is to say that I think I have some good airflow inside the case.
I first took some readings with C1E ENABLED and Vcore set to "auto", my system's default. The result of having C1E enabled with automatic voltage control is that when the CPU is idle, its frequency and VCore drop dramatically. When I did this, I got the following results (Tcase was read using Speedfan. Tjunction was read using TAT and speedfan, both agreed to within 1C):
Tcase : 33 idle, 60 load
Tjunction : 36 idle, 64 load
Vcore : 1.14 idle, 1.27 load
CPU frequency: 1600 idle, 2400 load
The difference in Tjunction betwen idle and load is about 28 degrees. This is slightly higher than Computronix's rule of 25, but this is probably due to the increased voltage between idle and load. What is strange is the difference between Tcase and Tjunction: only 1C when idle, increasing to 4C at load. This is very far from Computronix's indications that the difference should be around 15C.
To try to see which of the two was more accurate I followed Computronix's suggestions to disable C1E and set Vcore manually, and ran another test. The results get even more strange:
Tcase : 33 idle, 33 load (no, this is NOT a typo)
Tjunction : 42 idle, 58 load
Vcore : 1.24 idle, 1.22 load
CPU Frequency was constant at 2400.
Tjunction shows an increase of about 16 degrees under load. This is less than before, but it is almost certainly due to the fact that now I did not have the 0.13V increase in Vcore between idle and load that I had in the first test. Tjunction under load is also less than before, but this is probably due to the fact that Vcore at load was .05V less than before. The strange thing is that now Tcase did not move at all. I did not believe this was correct so I tested again: same result.
Based on all this I assumed that the measurements for Tcase in Speedfan for this board are somehow messed up, probably by the BIOS. I went back to the BIOS and after some searching and reading decided to enable two options that had been disabled: ACPI 2.0 support and PECI. I enabled both and tested again:
case : 34 idle, 50 load
Tjunction : 42 idle, 59 load
Vcore : 1.24 idle, 1.22 load
Everything stayed the same as before (within 1C), except that now Tcase under load seems to be more reasonable. So, for P5B owners, it seems that enabling PECI and/or ACPI 2.0 will allow for better temperature readings in Speedfan.
Computronix: does all this make sense? Is there a "logical" reason why one of these two options could have this effect? Could having them enabled have a downside?
Finally, I have a fairly consistent difference between Tcase and Tjunction of about 8 or 9C. It seems that Tcase should be about 5C lower, or Tjunction should be about 5C higher. I know I can enter offsets in Speedfan, but I don't know which one is the more accurate reading. Should I lower Tcase, or increase Tjunction? Is there a good way to try to assess which is the better reading?
Thanks again for the excellent guide, and TIA for you help!
EDIT: I am using the latest BIOS, version 1004.
In any case, after reading the guide I wanted to set some benchmarks for how my system behaves so I ran some tests. I am using TAT and Speedfan to read the relevant values (temperature, voltages) and I started getting some strange results. I know from the thread that the Asus P5B (my MoBo) seems to have some problems with accurate temperature readings, so maybe this will help shed some light on some of the problems and possible solutions.
First of all, here are my system specs
Ambient = 20C
Chipset = P965
C2D = E6600
CPU Cooler = Intel stock HSF
Frequency = 2400
Load = TAT @ 100% for 10 minutes
Motherboard = Asus P5B Deluxe
Everything is inside an Antec P180 case, so the PSU and HDD are in a separate chamber from the CPU/Mobo. The CPU/Mobo chamber has 2 120mm fans at the top/rear of the case, both in exhaust (the standard fan locations for the P180B). All fans set to manual and each fan moves about 39CFM of air. I spent a couple hours doing cable management, so the inside of my case is fairly clean. All this is to say that I think I have some good airflow inside the case.
I first took some readings with C1E ENABLED and Vcore set to "auto", my system's default. The result of having C1E enabled with automatic voltage control is that when the CPU is idle, its frequency and VCore drop dramatically. When I did this, I got the following results (Tcase was read using Speedfan. Tjunction was read using TAT and speedfan, both agreed to within 1C):
Tcase : 33 idle, 60 load
Tjunction : 36 idle, 64 load
Vcore : 1.14 idle, 1.27 load
CPU frequency: 1600 idle, 2400 load
The difference in Tjunction betwen idle and load is about 28 degrees. This is slightly higher than Computronix's rule of 25, but this is probably due to the increased voltage between idle and load. What is strange is the difference between Tcase and Tjunction: only 1C when idle, increasing to 4C at load. This is very far from Computronix's indications that the difference should be around 15C.
To try to see which of the two was more accurate I followed Computronix's suggestions to disable C1E and set Vcore manually, and ran another test. The results get even more strange:
Tcase : 33 idle, 33 load (no, this is NOT a typo)
Tjunction : 42 idle, 58 load
Vcore : 1.24 idle, 1.22 load
CPU Frequency was constant at 2400.
Tjunction shows an increase of about 16 degrees under load. This is less than before, but it is almost certainly due to the fact that now I did not have the 0.13V increase in Vcore between idle and load that I had in the first test. Tjunction under load is also less than before, but this is probably due to the fact that Vcore at load was .05V less than before. The strange thing is that now Tcase did not move at all. I did not believe this was correct so I tested again: same result.
Based on all this I assumed that the measurements for Tcase in Speedfan for this board are somehow messed up, probably by the BIOS. I went back to the BIOS and after some searching and reading decided to enable two options that had been disabled: ACPI 2.0 support and PECI. I enabled both and tested again:
case : 34 idle, 50 load
Tjunction : 42 idle, 59 load
Vcore : 1.24 idle, 1.22 load
Everything stayed the same as before (within 1C), except that now Tcase under load seems to be more reasonable. So, for P5B owners, it seems that enabling PECI and/or ACPI 2.0 will allow for better temperature readings in Speedfan.
Computronix: does all this make sense? Is there a "logical" reason why one of these two options could have this effect? Could having them enabled have a downside?
Finally, I have a fairly consistent difference between Tcase and Tjunction of about 8 or 9C. It seems that Tcase should be about 5C lower, or Tjunction should be about 5C higher. I know I can enter offsets in Speedfan, but I don't know which one is the more accurate reading. Should I lower Tcase, or increase Tjunction? Is there a good way to try to assess which is the better reading?
Thanks again for the excellent guide, and TIA for you help!
EDIT: I am using the latest BIOS, version 1004.
Facebook
Twitter
Instagram