So, what measures Wolfdale temps accurately?

[CurseTheSky]

I relied on TAT to measure the temps on my E6600 for over a year now. I just plopped an E8400 in (booted right up with P32 bios on the eVGA 680i A1) and TAT immediately exits saying I have an invalid processor. Oh well.

Core Temp 0.96 is reading 45C idle on both cores, which I kinda doubt. nVidia's monitor shows 36C, but 9.3GHz clock speed. I wish.

What's the trusted program these days?

 

[Zap]

Originally posted by: CurseTheSky
What's the trusted program these days?

Uh, placing your finger on the base of the heatsink? I like fingering them. You get the tactile feedback that way. I've noticed that generally speaking a non-overclocked and idle Core 2 Duo or socket AM2 chip runs pretty cool, the heatsink will be no warmer than around body temperature. If the heatsink is too hot to keep your finger on, then it is probably running too hot.

 

[CurseTheSky]

Originally posted by: Zap

Originally posted by: CurseTheSky
What's the trusted program these days?

Uh, placing your finger on the base of the heatsink? I like fingering them. You get the tactile feedback that way. I've noticed that generally speaking a non-overclocked and idle Core 2 Duo or socket AM2 chip runs pretty cool, the heatsink will be no warmer than around body temperature. If the heatsink is too hot to keep your finger on, then it is probably running too hot.

Very true, but I'm also curious to see if my heatsink is mounted well enough, and if there's any temperature improvements compared to my 65nm. Seems to be fine, but a solid number would be nice.

 

[Scottae]

possibly HWmonitor over there where you get CPU-Z
http://www.cpuid.com/hwmonitor.php

 

[graysky]

My understanding is that these new chips have some sort of physical flaw that prevents programs like coretemp, hwmonitor, etc. from accurately reading their temps. I'd post the link if I could find it.

 

[Tempered81]

I prefer Everest. Most programs report incorrectly the idle temps on these cpus. Try coretemp, everest, or hwmonitor while your stressing the cpu; from what i gather Load temperature readings are correct (given they are above 50C)

 

[JAG87]

If the latest build of Everest doesn't read it properly, nothing will.

 

[Scottae]

I agree Everest works great for me

 

[Duvie]

You have to pay for Everest.....so how about a freeware program?

Damned if I am going to pay for a program that gives me a temp.....

 

[ShockwaveVT]

try Speedfan 4.33

also, what's the ambient temperature in your room?

 

[tenax]

IF everest is accurate (not with my q9450 or any other program) there is a tab there under tools called system stability test. it is in the free ultimate version where you can see a graph test of voltages, fan speeds, temps.

 

[Spike]

I know it does not help you alot but after upgrading the BIOS on my IP35-E the Abit tool (can't remember the name, something like EQ) seems to read my E8400 temps accuratly. It shows me idling usually between 32-38 (depending on if I am running any other pc in the room thus increasing the overall temp) and my load temps hit around 48 while gaming or 52-54 while using orthos. Those are all 3.7

My biggest issue is I can't seem to increase my core voltage at all. No matter what setting I do in the BIOS it never reads anything higher than the stock volts. So even after a month my cpu is still not orthos stable though it has never crashed in a game. I want to try and push it more but worry about the voltages...

Anyway, hopefully soon the mobo makers will come out with better software for reading these great chips.

EDIT** As a note I could never get coretemp to read my cpu correctly, it would always give me an error and nvidias ntune still crashes Vista 64. I'll try everest sometime soon.

 

[Spike]

double post... sorry

 

[tenax]

yeah..really a mess right now..depending on the software, my q9450 reads as either 212C, 221C or -30C..and it never budges which tells me the thermal sensor is "stuck" or as is more likely, not designed for the 45nm (65nm version) i also can add that if i test using the various "power" benchmarks in sandra, it doesn't even get a reading..on a bright note, the cpu voltage is reading fine and accurately..and cool (1.160 volts at 8 x 400 overclock)

 

[graysky]

@tenax - how about coretemp 0.96.1? Can you post a screenshot of it's readings at idle, and another under p95 load?

Thanks!

 

[tenax]

i could, graysky..except it never changes..the temp on all cores in coretemp reads as 126C?

(that question mark is not a question to you, it's literally what it reads) the tjunction or whatever it is above those reads as 105C so i imagine that's why the question marks show up after the 126..load or no load..in any software, my temps don't change.

 

[DerwenArtos12]

Originally posted by: tenax
i could, graysky..except it never changes..the temp on all cores in coretemp reads as 126C?

(that question mark is not a question to you, it's literally what it reads) the tjunction or whatever it is above those reads as 105C so i imagine that's why the question marks show up after the 126..load or no load..in any software, my temps don't change.

I guess intel didn't think we were confused enough by the 8400, they had to take it a step further.

 

[tenax]

i can't recall a situation in the past where mobos and most bioses weren't ready before the processors hit the channel..who was in the big rush to do what that this happened? i don't get it.

 

[DerwenArtos12]

I really think that intel just up and forgot to mention the new temperature probes to their tier 1 partners so now all the motherboard manufacturers are trying to scramble to make it work and it's just not happening. Almost makes me wonder if the memo didn't make the full rounds at intel or even if 45nm was rushed to make their tick-tock and it's an actual quirk in the silicon that the temp probes themselves are malfunctioning.

 

[Tempered81]

Originally posted by: DerwenArtos12

Originally posted by: tenax
i could, graysky..except it never changes..the temp on all cores in coretemp reads as 126C?

(that question mark is not a question to you, it's literally what it reads) the tjunction or whatever it is above those reads as 105C so i imagine that's why the question marks show up after the 126..load or no load..in any software, my temps don't change.

I guess intel didn't think we were confused enough by the 8400, they had to take it a step further.

I read about why this happens. There is some hex method of reading the DTS and reporting in a formula related to Tjunction max. So it reads blah blah blah delta to TJmax. It was just incorporated for conroe/kentsfield as the latest and hasn't been updated for 45nm yet. Seems like they should write something to cater to the newest thermal calculation equations.

 

[DerwenArtos12]

Originally posted by: jaredpace

Originally posted by: DerwenArtos12

Originally posted by: tenax
i could, graysky..except it never changes..the temp on all cores in coretemp reads as 126C?

(that question mark is not a question to you, it's literally what it reads) the tjunction or whatever it is above those reads as 105C so i imagine that's why the question marks show up after the 126..load or no load..in any software, my temps don't change.

I guess intel didn't think we were confused enough by the 8400, they had to take it a step further.

I read about why this happens. There is some hex method of reading the DTS and reporting in a formula related to Tjunction max. So it reads blah blah blah delta to TJmax. It was just incorporated for conroe/kentsfield as the latest and hasn't been updated for 45nm yet. Seems like they should write something to cater to the newest thermal calculation equations.

Quite frankly that just seems too simple to me. If that was the case, and intel just forgot to tell it's teir 1 partners that pay them many of hundreds of millions of dollars a year for their chipsets that they were making the changes, even if the manufacturers were left to figure it out themselves, it shouldn't have taken this long. They have had production or near production silicon for months and they have BIOS available to support the chips but, not the temp sensors? At that I'm now hearing about some ABIT boards that the newest beta bios detects temps what appears to be correct but, can no longer change the vCore on wolfdales. Something is afoot, and it stinks of fish.

 

[Tempered81]

I just cannot fathom that it is hardware side. When was the last time intel released faulty or bugged cpus to the public?

 

[DerwenArtos12]

I find it nearly impossible to believe but, what viable alternatives are there? There is not one board on the market that I know of that can detect the temperatures properly without losing other functionality in the bios.

 

[Keysplayr]

Originally posted by: tenax
i could, graysky..except it never changes..the temp on all cores in coretemp reads as 126C?

(that question mark is not a question to you, it's literally what it reads) the tjunction or whatever it is above those reads as 105C so i imagine that's why the question marks show up after the 126..load or no load..in any software, my temps don't change.

Yikes, I could see that being Farenheit though.