Cool to the touch Ninja despite 1.5V dual core.. Odd or not?
Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee
Cool to the touch Ninja despite 1.5V dual core.. Odd or not?
Hello!
I have a core 2 duo 8400, cooled by a Ninja rev b, fitted with a Thermalright BoltThru kit, with the original Ninja fan hard wired at 5 volts, which if http://www.silentpcreview.com/article832-page3.html is right means 720rpm.
I'm a little confused though. At 1.5 volt in the bios, the cpu runs at 1.41 volt stressed, both cores Prime-ing away. CoreTemp says 64c for both cores, the CPU sensor reported by SpeedFan says 59c. Considering the overvolting this might not be excessive at all for 2x 100% load.
Edit: at stock voltage and speed its 48c both cores stressed. Power consumption should be about 30W according to xBit
However, no part of the Ninja that I can touch is remotely warm. I've sustained small burns from previous heatsink 'touch episodes' in the socket A era I can touch these parts:
- the heatpipes before they bend up
- the side of the 'regular' small heatsink on the base
- the lowest cooling fin
- the top fin
Neither of those is subjectively above 25c. Its not even 'pleasantly warm', its barely warm. Not painfully warm back when my Athlon was 75c and I touched the Thermalright copper heatsink!
Since mounting it with a boltthru kit is really secure, I doubt anything is wrong there. Perhaps it is just the heat being so concentrated on two tiny 45nm cores that the temperature difference between the core of the cpu and the heatsink fins is just expected to be a lot higher than in the old days?
Any thoughts?
Niels
I have a core 2 duo 8400, cooled by a Ninja rev b, fitted with a Thermalright BoltThru kit, with the original Ninja fan hard wired at 5 volts, which if http://www.silentpcreview.com/article832-page3.html is right means 720rpm.
I'm a little confused though. At 1.5 volt in the bios, the cpu runs at 1.41 volt stressed, both cores Prime-ing away. CoreTemp says 64c for both cores, the CPU sensor reported by SpeedFan says 59c. Considering the overvolting this might not be excessive at all for 2x 100% load.
Edit: at stock voltage and speed its 48c both cores stressed. Power consumption should be about 30W according to xBit
However, no part of the Ninja that I can touch is remotely warm. I've sustained small burns from previous heatsink 'touch episodes' in the socket A era I can touch these parts:
- the heatpipes before they bend up
- the side of the 'regular' small heatsink on the base
- the lowest cooling fin
- the top fin
Neither of those is subjectively above 25c. Its not even 'pleasantly warm', its barely warm. Not painfully warm back when my Athlon was 75c and I touched the Thermalright copper heatsink!
Since mounting it with a boltthru kit is really secure, I doubt anything is wrong there. Perhaps it is just the heat being so concentrated on two tiny 45nm cores that the temperature difference between the core of the cpu and the heatsink fins is just expected to be a lot higher than in the old days?
Any thoughts?
Niels
As long as your temps are fine it's no issue att all. Although my ninja (first version, still going strong) does get warm, but only if I run encoding for an hour or two. And thats with noctua 120mm running at 10v. CPU is [email protected] running at 1.52V.
Re: Cool to the touch Ninja despite 1.5V dual core.. Odd or
All mobo's have some level of Vdrop and Vdroop. Some are better/worse than others depending on how well they designed the regulation circuitry. So, you are actually running at 1.4V core, not 1.5Vniels007 wrote: I'm a little confused though. At 1.5 volt in the bios, the cpu runs at 1.41 volt stressed, both cores Prime-ing away. CoreTemp says 64c for both cores, the CPU sensor reported by SpeedFan says 59c. Considering the overvolting this might not be excessive at all for 2x 100% load.
Niels
Vdrop = delta between BIOS setting and Vcore measured.
Vdroop = delta between Vcore measured at idle and at load when Vcore is set to a specific voltage in BIOS and not allowed to vary.
Nice temps...I may have to break down and get a backplate....
The rebuild is complete and I did mount the ninja again. I figured a bit more pre tension on the springs in the thermalright kit might help so i put a ~3mm spacer between the screw head and the spring. This does bend the Ninja mounting plate somewhat but it felt secure.. As expected though the results are more or less the same as before.
Now at 1.55 volt in bios which is 1.5 in windows idle and 1.41 stressed, at 4ghz the max coretemp reported is about 81, with a Nexus at 1000rpm which speedfan sets at 500rpm idle, where it is about 51 degrees. Prime wasn't quite stable before the rebuild but seems to be good now.
I might make a sort of duct so the cpu fan air has to go all the way through the ninja instead of having the chance to escape through the sides.
I'm pretty sure the current setup still can't handle 4ghz in summer.
Now at 1.55 volt in bios which is 1.5 in windows idle and 1.41 stressed, at 4ghz the max coretemp reported is about 81, with a Nexus at 1000rpm which speedfan sets at 500rpm idle, where it is about 51 degrees. Prime wasn't quite stable before the rebuild but seems to be good now.
I might make a sort of duct so the cpu fan air has to go all the way through the ninja instead of having the chance to escape through the sides.
I'm pretty sure the current setup still can't handle 4ghz in summer.
I was kinda hoping that too; some ram / nb setting.. Seems my E8400 (one of the first batches they made) just isn't a speed king. As for ram / nb stability, I lowered the cpu multiplier to 7 (x444) and everything is stable, but at 9x444 it really does need a lot of voltage.
I doubt the step from 7 to 9 would suddenly highlight ram / nb issues but I could be wrong.
I have not tested how fast it goes on stock voltage. At 1.4 volt (1.36 windows, 1.3 stressed) I'm at 8x444, 3.5ghz which also seems to be the minimum it can take to be stable.
There are a lot of voltage things in the bios but reading the web, it seems nobody understand them and I pretty much conclude that my cpu simply isn't one of the higher clocking ones
I doubt the step from 7 to 9 would suddenly highlight ram / nb issues but I could be wrong.
I have not tested how fast it goes on stock voltage. At 1.4 volt (1.36 windows, 1.3 stressed) I'm at 8x444, 3.5ghz which also seems to be the minimum it can take to be stable.
There are a lot of voltage things in the bios but reading the web, it seems nobody understand them and I pretty much conclude that my cpu simply isn't one of the higher clocking ones
Re: Cool to the touch Ninja despite 1.5V dual core.. Odd or
Touching a metal object with a 25C temperature would feel *cold* because your body temperature is 37 degrees (a bit lower in the fingers, but nevermind).niels007 wrote:Neither of those is subjectively above 25c. Its not even 'pleasantly warm', its barely warm.
"Pleasantly warm" is around 45C.
"Warm" is around 40C.
It is a very efficient cooler even without any airflow:) My Ninja2 gets toasty when my CPU gets upwards 60 degrees, but then I have no airflow what so ever in that case, only convection. If the fan is running it always feels cold to the touch and the temps of my (undervolted) CPU never leaves the low thirties.
I checked, the 'auto' voltage is about 1.17 idle / 1.12 load. When I manually set this, it won't even do 3.33ghz without a sudden reboot or if I'm lucky a Prime95 error after only a short time.
At default voltage and speed I get these temps (CoreTemp and RealTemp values, they're nearly identical):
- 500rpm fan: both cores 59c full load
- 1000rpm fan: both cores 57c full load
I still don't understand why, at this default setting, a ~35W cpu with a bolt kit mounted Ninja rev B reaches such temperatures full load at stock voltage and speed!
I'm not temperature obsessed at all. I wouldn't mind pretty much any temperature as long as its stable and quiet. These values simply seem on the high side, limiting my achievable silent / stable overclock. Sure there is no real need for 4ghz, but if solving a cooling problem makes it possible, I'd still like the sound of that speed. I wouldn't even mind the 1.55 volt if that would be required but at 82c temps thats going to be a problem if ambient temp rises 10 degrees in summer..
Edit: improved with this viewtopic.php?p=453408#453408
At default voltage and speed I get these temps (CoreTemp and RealTemp values, they're nearly identical):
- 500rpm fan: both cores 59c full load
- 1000rpm fan: both cores 57c full load
I still don't understand why, at this default setting, a ~35W cpu with a bolt kit mounted Ninja rev B reaches such temperatures full load at stock voltage and speed!
I'm not temperature obsessed at all. I wouldn't mind pretty much any temperature as long as its stable and quiet. These values simply seem on the high side, limiting my achievable silent / stable overclock. Sure there is no real need for 4ghz, but if solving a cooling problem makes it possible, I'd still like the sound of that speed. I wouldn't even mind the 1.55 volt if that would be required but at 82c temps thats going to be a problem if ambient temp rises 10 degrees in summer..
Edit: improved with this viewtopic.php?p=453408#453408
As regards your temps, don't know if you've tried it, but the whole calibration issue mentioned by CA_Steve here might be affecting your readings.
-
- Posts: 64
- Joined: Thu Sep 18, 2008 7:05 am
- Location: Rhode Island
E8400 is a peice of work. I got mine to boot into windows at 4.75ghz on a DS3L P35 gigabyte board with an Ultra-120. 4.5ghz benchable. Those are some sick numbers with some sick temps, lol.
Mine was voltage happy, it wouldn't clock past 3.33ghz on stock but when I really fed it, it would just keep going. Beautiful.
Mine was voltage happy, it wouldn't clock past 3.33ghz on stock but when I really fed it, it would just keep going. Beautiful.
on that crappy picture you can sort of see the config. It is an NSK3480 case with a 120mm bottom intake fan (modded!) just below the vga cooler. You can also just see the harddisk is put on its side in the front, in two 0 shaped foam bits for suspension. The lower front 92mm fan hole is taped shut, the upper 92mm fan grill is cut out hopefully increasing its intake effectiveness as the CPU fan is about at that height. Rear fan grill also cut.
The ATX cable, HD power and VGA power are fed through a hole next to the ATX plug, cables then go to the powersupply on the backside of the case. SATA cables for the cdrom and backup HD go the same route up to the upper cabinet. That HD has an on/off switch in the front panel so it only runs when making a backup..
Bottom intake and rear exhaust are connected to one fanmate with a Y cable and run at about 600rpm, the lowest speed they will consistently start at. The cpu fan runs at 100% (~1040rpm) above 75c cpu temp and 40% (550rpm) below 65c.
Hardware:
PSU: Enermax Pro 82+ 425W (fan runs at 490..520rpm great psu but slight bearing noise)
Mobo: Asus P5 Vm HDMI as reviewed by SPCR
CPU: Intel Core 2 Duo E8400 (9x444(.4)=4ghz, 1.55V bios! yes it needs it
Cooler: Ninja rev B with duct / Nexus and Thermalright Boltthru kit
RAM: 2Gb 533mhz valuestuff (DDR2 1066)
HD: 750GB WD Green 5400rpm
VGA: Gigabyte silent 9600GT but with more efficient Arctic Accelero S1