Solo-Q9650 build running a bit hot?
Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee
Solo-Q9650 build running a bit hot?
I was hoping to avoid using intake fans on a solo build, but based on temps in a stripped down config its looking like the intake fans may be required.
Here's my build. Any suggestions on getting things cooler would be appreciated.
Intel Q9650
Ninja - 2000 with Scyth Fan installed
Asus P5N-D
Antec Solo with Stock exhaust fan
4 GB Ram
Asus EN9600GT Silent
Thermaltake Toughpower 650W PSU
No HDs installed yet...I'm just testing initial temps.
Booting linux via USB key and running memtest for a few hours I get temps of
38C for MB and 60C for CPU (from bios) with rear fan on low.
Bumping the rear fan to high will drop temps about 4C.
Here's my build. Any suggestions on getting things cooler would be appreciated.
Intel Q9650
Ninja - 2000 with Scyth Fan installed
Asus P5N-D
Antec Solo with Stock exhaust fan
4 GB Ram
Asus EN9600GT Silent
Thermaltake Toughpower 650W PSU
No HDs installed yet...I'm just testing initial temps.
Booting linux via USB key and running memtest for a few hours I get temps of
38C for MB and 60C for CPU (from bios) with rear fan on low.
Bumping the rear fan to high will drop temps about 4C.
-
- Moderator
- Posts: 7681
- Joined: Mon Dec 09, 2002 7:11 pm
- Location: Maynard, MA, Eaarth
- Contact:
The CPU heatsink is only about an inch away from the rear exhaust fan and I have the CPU fan mounted on the font side of the heatsink blowing air across the heatsink and towards the rear exhaust.NeilBlanchard wrote:You could try using a duct from the CPU heatsink to the rear exhaust, to lower them?
-
- Moderator
- Posts: 7681
- Joined: Mon Dec 09, 2002 7:11 pm
- Location: Maynard, MA, Eaarth
- Contact:
A 38C rise over ambient when idling implies:
- your temp reporting s/w isn't calibrated, and/or
- your heatsink isn't mounted properly, and/or
- you have world class levels of poor airflow
Something is really off..I'd advise using Real Temp (calibrated) to get a better feel for the core temps. Then, we can see where things lie.
Here's some recycled advice from March
Using Real Temp (CPU temp monitoring freeware):
First step - remove your case door.*
A. Set (45nm) CPU core in BIOS to run at 333MHz and x6 when idling
B. Set Vcore to 1.1V for idle measurement:
1. With your current settings, look at CPU-Z. Write down the Vcore (idle).
2. Reboot, and go into BIOS. Look at your Vcore setting. Write down the offset between this setting and the CPU-Z reading. If your Vcore is set to auto...change it to 1.15V and then boot PC. Read CPU-Z and then write down the offset.
3. Change Vcore in BIOS by this offset or something close until CPU-Z reads 1.1V or something slightly lower.
This sets the baseline for idle temp. Let it run for a couple of minutes.
C. Real temp idle calibration.
1. Measure the ambient temperature inside the case and read the Real Temp core temp. If they are within, say 8C**, stop. If not, go to step 2.
2. Go to the settings screen in real temp and put in about a -5 degree offset.
3. Read real temp and case ambient temps again. If they are within 8C stop. If not, change the offset.
4. Rinse, repeat until all cores are within 8C. (My offsets ended up being -6C and -5C).
D. Go back into BIOS and configure your CPU for desired speed, voltage. Close your case door.
That should do it.
* you may have better circulation with the door closed, but this is the way the author set up his data gathering and offsets. So, it's best to calibrate the same way.
** better to have Real Temp overestimate your load temps than underestimate them...if you feel good about your setup, you can always reduce the Tcore-Tambient difference.
- your temp reporting s/w isn't calibrated, and/or
- your heatsink isn't mounted properly, and/or
- you have world class levels of poor airflow
Something is really off..I'd advise using Real Temp (calibrated) to get a better feel for the core temps. Then, we can see where things lie.
Here's some recycled advice from March
Using Real Temp (CPU temp monitoring freeware):
First step - remove your case door.*
A. Set (45nm) CPU core in BIOS to run at 333MHz and x6 when idling
B. Set Vcore to 1.1V for idle measurement:
1. With your current settings, look at CPU-Z. Write down the Vcore (idle).
2. Reboot, and go into BIOS. Look at your Vcore setting. Write down the offset between this setting and the CPU-Z reading. If your Vcore is set to auto...change it to 1.15V and then boot PC. Read CPU-Z and then write down the offset.
3. Change Vcore in BIOS by this offset or something close until CPU-Z reads 1.1V or something slightly lower.
This sets the baseline for idle temp. Let it run for a couple of minutes.
C. Real temp idle calibration.
1. Measure the ambient temperature inside the case and read the Real Temp core temp. If they are within, say 8C**, stop. If not, go to step 2.
2. Go to the settings screen in real temp and put in about a -5 degree offset.
3. Read real temp and case ambient temps again. If they are within 8C stop. If not, change the offset.
4. Rinse, repeat until all cores are within 8C. (My offsets ended up being -6C and -5C).
D. Go back into BIOS and configure your CPU for desired speed, voltage. Close your case door.
That should do it.
* you may have better circulation with the door closed, but this is the way the author set up his data gathering and offsets. So, it's best to calibrate the same way.
** better to have Real Temp overestimate your load temps than underestimate them...if you feel good about your setup, you can always reduce the Tcore-Tambient difference.
"COOLING...............................IDLE DEGREES ABOVE AMBIENTmetron wrote:For a Q9650 what would the expected rise in temp above ambient be?
High end water...............................6C above ambient
High end air (true push/pull) ........6-7C
High end air (1fan).........................7C
Mid air (zalman 9500)....................8-9C depending on fan rpm
Intel stock cooler...........................10-11C"
For Core 2 processors, when set to the low freq/low core voltage idle state I showed earlier.
A 2C drop implies the case isn't that starved for airflow.metron wrote:Removing the case door dropped the temps by 2C.
<shrugs> 45nm die temp monitoring isn't as straightforward/linear as it used to be. So, there is a lot of variability in the accuracy of measurement from BIOS to OS monitors to reality. Your mileage may vary.metron wrote: Is monitoring the temp with the bios not recommended?
metron wrote:Are there any non windows solutions for the instructions suggested above?
Don't know. But, if you don't want to mess with Windows, at least see if going to the underclocked/undervolted idle state makes a big difference in the BIOS based idle temp. Keep the case door open as a reference point, set the BIOS, reboot and then monitor.
If the temp doesn't drop very much, then it implies the problem is the CPU/heatsink mounting.
The question is where do you take this measurement? I've checked various places inside the case and there is a huge variance in temp difference which is to be expected. Of course if you measure near the vid card you get a much higher themp then if you measure near an area of the MB with no components.1. Measure the ambient temperature inside the case and read the Real Temp core temp. If they are within, say 8C**, stop. If not, go to step 2.
Also, don't know if this test works or not, but I turned off the rear exhaust fan as well as the CPU fan and temps shot up to 71C in a minute or two. I then put the fans back on and temps dropped back to 57C in about a minute. Based on that I'd guess the heatsink is working correctly.
Doesn't this rely on a couple of assumptions?CA_Steve wrote:C. Real temp idle calibration.
1. Measure the ambient temperature inside the case and read the Real Temp core temp. If they are within, say 8C**, stop. If not, go to step 2.
2. Go to the settings screen in real temp and put in about a -5 degree offset.
3. Read real temp and case ambient temps again. If they are within 8C stop. If not, change the offset.
4. Rinse, repeat until all cores are within 8C. (My offsets ended up being -6C and -5C).
1) that the heat sink is installed with very good contact
2) that there isn't anything on the motherboard that would cause the cpu to be hotter without increasing the ambient temp much (if at all). For example, a heat source sitting directly under the CPU.
What we know:
- your idle temps are displayed as 58C
- your load temp is shown as 60C with low airflow and 56C with higher airflow.
- with no airflow temps were 71C - when you opened the case, temps dropped 2C
- temps were measured via BIOS screen
I think your CPU cooler is working fine and your airflow is fine. You just have temp measurement issues with your mobo. You could just ignore the actual number and walk away now
I'm still curious as to what your core's baseline idle temp is. Underclock the CPU as shown earlier and set the Vcore to 1.1V. If it's a measurement issue, your CPU temp will probably display in the 50's in BIOS. If it drops significantly (say to 30C), then we'll have to rethink this.
Ambient temp measurement: It's just a measure of the ambient temp inside the case. There will be hot/cool spots and really it's dealing with the temp of the ambient air b4 it flows over/thru your CPU cooler...but let's stay at a higher level. Good airflow, with the door open, may be just a few degrees C above your room temp.
I don't have a temp probe, so I just take an educated guess. I look at the HDD temps after a cold boot (where the PC's been turned off and had some time to get to room temp). With the case door open, I let the PC idle for 5 min and then look at the HDD temps again. I have airflow over my 2 HDD's and don't expect them to get more than 3C over case ambient.
- your idle temps are displayed as 58C
- your load temp is shown as 60C with low airflow and 56C with higher airflow.
- with no airflow temps were 71C - when you opened the case, temps dropped 2C
- temps were measured via BIOS screen
I think your CPU cooler is working fine and your airflow is fine. You just have temp measurement issues with your mobo. You could just ignore the actual number and walk away now
I'm still curious as to what your core's baseline idle temp is. Underclock the CPU as shown earlier and set the Vcore to 1.1V. If it's a measurement issue, your CPU temp will probably display in the 50's in BIOS. If it drops significantly (say to 30C), then we'll have to rethink this.
Ambient temp measurement: It's just a measure of the ambient temp inside the case. There will be hot/cool spots and really it's dealing with the temp of the ambient air b4 it flows over/thru your CPU cooler...but let's stay at a higher level. Good airflow, with the door open, may be just a few degrees C above your room temp.
I don't have a temp probe, so I just take an educated guess. I look at the HDD temps after a cold boot (where the PC's been turned off and had some time to get to room temp). With the case door open, I let the PC idle for 5 min and then look at the HDD temps again. I have airflow over my 2 HDD's and don't expect them to get more than 3C over case ambient.
Just thought of something:
For the load test...when you say you look at the CPU temp via the BIOS screen, does this mean you:
- boot the PC via USB stick
- run memtest
- reboot, go into BIOS and look at the temp?
If so, you aren't looking at the load temp. You are looking at the remainder of the load temp after the time it takes to reboot has passed. In a well cooled system, the core load temps will drop almost to idle temps in a matter of seconds. I just tried it running Prime95 and monitored via speedfan. 5 seconds to drop 15-20C.
For the load test...when you say you look at the CPU temp via the BIOS screen, does this mean you:
- boot the PC via USB stick
- run memtest
- reboot, go into BIOS and look at the temp?
If so, you aren't looking at the load temp. You are looking at the remainder of the load temp after the time it takes to reboot has passed. In a well cooled system, the core load temps will drop almost to idle temps in a matter of seconds. I just tried it running Prime95 and monitored via speedfan. 5 seconds to drop 15-20C.