E8400, pretty hilarious temps (cold)

chaloux

n00b
Joined
Oct 17, 2006
Messages
56
What is everyone else using/what version to test temps of E8400.

My box is outside with good airflow. It's currently -3C outside.

I haven't even overclocked yet, and at stock settings this is what I'm seeing:

Coretemp .96.1: 35C no matter what
Speedfan 4.33: -2C idle, 7C load
Everest 4.2: doesn't show idle, 7C load

Is it actually running this cold? My old E4400 would run at 17C idle and heat up to 40C at load... I have a hard time believing the CPU is that close to ambient.

dope.jpg
 
The E8400 have temp reporting problems currently. I wouldn't necessarily believe anything I saw. The 35C load with those ambients seems reasonable though - especially compared with the motherboard and GPU temps. Those northern winters are great coolers, eh...
 
what is that program that is running to the right?

ice seen it before with other people who are testing their temps under load?

whats it called?
 
Alright, it seems crazy, but I am going to go with the Speedfan temps. The reason is that I am now at 3.7ghz (1.175V), and the load temp is now 10C... scary! :p

But at least the temp is going up with increased clocks, right? I really hope some sort of software/bios fix comes out for this issue.

The E8400 have temp reporting problems currently. I wouldn't necessarily believe anything I saw. The 35C load with those ambients seems reasonable though - especially compared with the motherboard and GPU temps. Those northern winters are great coolers, eh...

Oh yeah they work great, except for the changes in temperature, but I keep things fairly conservative so even if the temperature outside does go up, I'm still fine.

The only thing about the Coretemp reading is that it is 35 all the time. Never changes, idle or load.

illuminate said:
what is that program that is running to the right?

ice seen it before with other people who are testing their temps under load?

whats it called?

It's called Prime95. Do a quick google search, and it should pop up right away.
 
thanks for the app name

i would go with the normal speedfan. get the beta. my 'normal' one says that i am at 21C with both cores.

the beta and core temp both say 36 for each...
 
Thanks. I just got the beta, 4.34 b38, and temps are still reading 10C.
Core 0 says 30, Core 1 says 30, never changes.

Interestingly, I tried coretemp .95.4, and instead of a Tjunction of 105, it reads 100. My core temps are 30 instead of 35. So 5C decrease in Tjunction and core temp. hmmmm. Something is certainly fishy.

I am sticking with Speedfan because the BIOS reads about the same as my Idle temps (-2 or -3).
 
At stock speeds, my E8400 idled at 21C with ambient 18C.
-running Thermaltake Ultra Extreme 120 wotnot cooler with MX2 paste and 68CFM 22dB fan.
After clocking (and raising the voltage) to over 4.2GHz, it idles at 26C to 29C which is what I would expect so the temps look about right for mine.

I get 3C over ambient at idle at stock, you're pretty close to that so I'd say the temps are valid, accounting for say +/- 2C error.
 
thanks for the app name

i would go with the normal speedfan. get the beta. my 'normal' one says that i am at 21C with both cores.

the beta and core temp both say 36 for each...

I assume 21c is an idle temp. If so, your room needs to be about 16c maximum, for you to have that low of a temperature at idle.

I'd go with coretemp.
 
You have your computer outside? Like it's just chilling in the snow? haha. Or is it in like an open window or something? Just curious. I'm just like picturing this computer sitting outside in winter weather, with a bunch of cables running up through a window into a room where you have the monitor and such, haha.
 
Interestingly, I tried coretemp .95.4, and instead of a Tjunction of 105, it reads 100. My core temps are 30 instead of 35. So 5C decrease in Tjunction and core temp. hmmmm. Something is certainly fishy.
That's just the way coretemp works - it subtracts the DTS value from the Tjunction to display a temp - so change Tjunction and you change the reported temp. Click the "show delta to Tjunction" setting and you won't have to worry about it.
 
You have your computer outside? Like it's just chilling in the snow? haha. Or is it in like an open window or something? Just curious. I'm just like picturing this computer sitting outside in winter weather, with a bunch of cables running up through a window into a room where you have the monitor and such, haha.

Pretty much. There are some precautionary measures taken...

comp.jpg
 
Well I figured I'd update this thread again with some more useful information.

I am now at 3.9ghz with a vcore of 1.25 (picked cause it seemed like enough steps to be stable).

Coretemp, speedfan, and CPUID Hardware Monitor are all now reporting a RISE in temp as I load the CPU. What threw me off before was that there was simply no change in the core temp. I guess this can be attributed to cold ambient temperature, but to see no load temp increase... kinda crazy. At higher clocks and voltages it makes sense that it would increase. Either that, or the fact that I have restarted my PC a few more times fixed something.

SS idle:
idle39.jpg



SS load:
load39.jpg
 
I reckon the air was so cold when you took the earlier temps that the temperature reporting tools/hardware hit the lowest value they are capable of showing while idle and under load.
This explains why you saw no change.
Using more power (higher VCore or clocks) or having higher ambient temps is allowing the temps to register above the minimum.

btw to answer your earlier request, I use the Gigabyte overclocking tool that came with my mobo to measure temps and Coretemp 0.96.1 to get absolute values and the delta to max.
(Subtract 15C from the Delta to max to get the max temp you should ever hit, not that it matters in your case :D)

Some more info in case you get tempted to go for max clocks:
I read an Intel document a few days ago that says it is detrimental to the CPU to go over 1.45V vcore.

Before I read that, I tried 1.5V and did noticed it follows the law of diminishing returns in a big way above 1.45V, the higher voltages did nearly nothing to help my overclock.
I still had over 20C delta to go (after the 15C adjustment).
 
this makes me want to cancel my order and wait for the new batch to come >.<

Just because of the questionable temp reading? I can't argue with 900mhz and .125 volts increase. I am nowhere near my personal max (of 1.4v) and close to the "average" OC of 4ghz. I'm 63C away from the delta to max so I have a way to go.

Nenu: I will probably attempt to hit a max clock with around 1.4 or 1.45 volts, but only just to see what the chip is capable of. I will probably leave the chip around 4ghz and whatever voltage that requires for everyday use. My guess is it'll be somewhere around 1.3v or just above.

Then when spring/summer comes, I'll have to re-do everything as it will probably come inside where it'll be cooler.

Thanks everyone for your responses. :D
 
Heheh nice setup but what happens when it's not winter any more? Lower speed or ?

By most accounts the Wolfdales can hit 4+GHz on air with normal ambient temps, so chances are he'll be able to keep that overclock, although he might need to bump the voltage a little :).

Edit: About the temps, take a look at this thread, there may be some issues with the thermal sensors on the Wolfdale chips.
 
Thanks for the headsup although I think there is some confusion.
I dont doubt from some of the posts that the 45nm chips are suffering even when not pushed too hard.
I'm going to back down my overclock a little until this is cleared up.

The temp reporting issue seems to be related only to non X38/X48 motherboards.
I havent seen the complaint from anyone with X38 or X48.
I have X38 and believe my mobo reported temps are just right.
Currently: Idle 26-29C, Load 51-55C with 1.45V @ 4.275GHz
 
Back
Top