Real Temp - New temp program for Intel Core processors

[WoZZeR999]

Interesting information. Is that the difference between Thermal Monitoring 1 and 2? 1 lowers multi, 2 lowers vcore.

Anyway, I agree with Wood Butcher, go enjoy your summer, I don't see anything in the request queue for RT.

[zorzyk]

I don't see anything in the request queue for RT.

Hm... RT doesn't make coffee yet

[-X-hellfire]

Hm... RT doesn't make coffee yet

you could probably turn a Quad core into at coffee cup heater if take of the TRUE and put your coffecup there

...then you can monitor very accurately with RT how hot coffe you get, maybe only 3 instances of Prime95 will get your coffee perfect, 85C I think

If the Quad Q6600 G0 are throttling at 95C then it might be a better idea to get the older B3 which throttling at 105C

[nstarz]

Hi, I tested and tried to get my CPU calibriated to the lowest I can as per instruction. My previous post was http://www.xtremesystems.org/forums/...postcount=1647

Here is a screenshot of the undervolted CPU. I ran it at 4am Room temperature is 24C



So it seem to be 3 degree cooler than before. But I think that is just cause the room temperature is colder too. The CPU is virtually idle. Is there anything else I can do or is this CPU can't report anything under 50C?

[WoodButcher]

nstarz, speedfan, coretemp, RT all report high temps, reading your thread at SPCR you did all the right things, did you check the chip and sink for flat? get a backplate mounting system? Try another heatsink? My guess?,, hardware prob not software.

[unclewebb]

So it seem to be 3 degree cooler than before.

The Minimum temperature shown in your second post is 51C so there's an extra degree for you! This second test confirms what I originally believed, your sensors are not stuck at 55C. If they were stuck they wouldn't have been able to go even 1C lower.

Your reported temperatures are still about 20C higher than what they should be. I asked you before about how do you apply thermal paste and how much do you use. It's probably not that but you have to start with simple things when trying to figure out a problem like this. Does TaskManager show a flat line that looks like a heart trace of a dead guy when you are idle? Some people that think they are idle have some junk running in the background creating heat.

Next thing to look at is what WoodButcher mentioned about flatness. If the IHS on top of your processor is not flat or if your heatsink is not flat then there won't be proper contact and heat transfer.

Another test you can try is running Prime95 small FFTs on both cores at default MHz and core voltage. Run Prime for about 5 minutes and compare your full load temps to your idle temps. When you have poor heat transfer, it usually shows up in a larger than normal temperature change during a stress test like this. It's possible that the IHS is not making proper contact with the cores but I wouldn't start to think about that possibility until you look into some of the other things I mentioned.

If all this is too much work then I guess there is always EBay and you can try again.

WoZZeR999: During my test above, RMClock 2.30 showed that both TM1 and TM2 were enabled. I'll have to do some research about what each one is supposed to do unless CompuTronix comes along to help me out. C1E and SpeedStep were both turned off. There's still one or two more features to add to RealTemp before I go into retirement!

A Q6600 powered coffee warmer sounds very interesting. I'm sure some rich guy at a trade show would pay top dollar to be the first on his block to have one. I'll take mine with two cream, one sugar and two rounds of Prime small FFTs. Perfect!

[-X-hellfire]

unclewebb, sorry for the late response, just got fed up fiddling with pictures etc, which describes the oc etc in bios, had a few more crashes too, so I need to get the C1E speedstep vdrop under control or disable it

[Dua|ist]

unclewebb
Regarding TM1 & TM2. AFAIK TM1 is the "older" version, that was introduced in the first Pentium IVs. It's supposed to force the CPU skip some cycles per second in order to cool itself down. The actual frequency remains the same but the effective frequency is the lower the hotter the CPU is. TM2 was introduced later (in Prescotts perhaps, as it uses same principles as EIST and C1E). TM2 first lowers the multiplier to its minimum, if that's not enough it lowers the VID (just like in your observations). TM2 is more efficient in cooling the CPU down thus it is triggered first upon the overheat. And if TM2 can't prevent CPU temps from raising further, TM1 is started. RightMark will show two frequency graphs in that case: the actual frequency and the effective one (which is lower).
P.S. Thanks for Realtemp! It's becoming better and better.
P.P.S. Sorry for my bad english.

[CompuTronix]

unclewebb,

The following link - http://www.digit-life.com/articles2/...res-core2.html - is to an excellent article written at XBD Digit Life on 12/12/06. The author uses a QX6700 B3 and RM Clock for testing, which includes detailed decriptions and analysis of the functions and interactions between Intel's TM1 and TM2 technologies, as noted above by Dua|ist.

The article is well written, very informative, and shows the B3 throttling at 81c. Also, there's an APIC ID table, and a brief dicussion which I'm sure you'll find interesting, as it helps to make sense of the "shifting Core temperature" phenomenon.

Enjoy!

Comp

[unclewebb]

Thanks Dua|ist for your explanation.

I also knew CompuTronix would have something up his sleeve to explain TM1 / TM2. Some people still want to hang me from a tree for suggesting something crazy like TjMax=85C for the B3 Quads. I'm not sure what TjMax the latest version of RMClock is using for the B3 but it probably got bumped up to 100C like most other programs are using.

RealTemp uses the APIC ID table to sort out the core order on the Quads. I hope Everest and CoreTemp follow along because it can be very confusing when the temps of individual cores start moving around from one day to the next. This issue has obviously been around for a long time now. Thanks for that link.

Edit: I tried using RMClock 2.30 but it can't find any DTS sensors on my chip so it doesn't show any temperature data. Version 2.35 is the same but it doesn't let me exit the program. It just hangs and I have to kill it in the Task Manager. I think I'll stick with RealTemp.

Here's my Core 2 Solo Q6600 with 4 cores:

[nstarz]

nstarz, speedfan, coretemp, RT all report high temps, reading your thread at SPCR you did all the right things, did you check the chip and sink for flat? get a backplate mounting system? Try another heatsink? My guess?,, hardware prob not software.

Yes I am using a Thermalright LGA775 Bolt-Thru-Kit. The chip and sink looks flat. Is there a method to test the flastness? I use a razor blade to apply thermal paste evenly. I did it 3 times now and with a new Artic thermal paste too.

It looks idle at 0-3%. when I check temperatures again.

Im running Prime and the temperature is now around 60 C.
It was 51 C at idle. My current room temperature is 26 C.

Is this the worst (highest) temperature (51 C) stuck for E8400 so far?

[HDCHOPPER]

What happened? I thought you were out enjoying the summer,,, RT works, you've done a beautiful job, none better in fact, it's time to chase bikinis!

here ! here ! Unclewebb rocks !

[WoodButcher]

Yes I am using a Thermalright LGA775 Bolt-Thru-Kit. The chip and sink looks flat. Is there a method to test the flastness? I use a razor blade to apply thermal paste evenly. I did it 3 times now and with a new Artic thermal paste too.
A straightedge held perpendicular against the surface, w/ a light source behind

Im running Prime and the temperature is now around 60 C.
It was 51 C at idle. My current room temperature is 26 C.

60c is not unusual for air at load, what did the stock cooler give you? as to the idle temps maybe bad sensors, I don't know, 9deg c difference between idle and load is watercooling not air. If your chip throttles w/ stock cooling send it back, (don't mention the blow-dryer aimed at your heatsink!) if that does not fly RMA the board. As I said, from reading your thread at SPCR you tried all the right things, You have an internal hardware problem. RMA!

Is this the worst (highest) temperature (51 C) stuck for E8400 so far?

I don't know, in reality your idle temps are irrelevant, the sensors are there to prevent overheating to protect your chip.

...

[unclewebb]

A holiday means I only spend half the day on the computer instead of the whole day. If anyone was wearing a bikini in this neighborhood I'd probably have to call the cops. Mostly retired people swinging golf clubs around here!

Just a couple of minor features to add before the next official release. CompuTronix thought a temperature alarm would be a good idea. I'll probably start with something simple and wait for some feedback. The other minor item I'm adding is a user selectable anchor position. Right now when you double right click on RealTemp it will jump to the top left corner of your screen. The next version will let the user decide what position RealTemp jumps to after a double right click.

nstarz: Yep, I think 51C is a new record for a stuck sensor. Congratulations!!!

This quote needs repeating:

I don't know, in reality your idle temps are irrelevant, the sensors are there to prevent overheating to protect your chip.

As long as your computer is stable and not catching on fire then there's nothing to worry about. Intel seems to have decided that accurate idle temperatures are more like a bonus feature. 95% of their retail customers could care a less so it doesn't make any sense for them to invest too heavily into producing sensors to keep the XS types happy. I've got an E8400 that can run cool and quietly, 24/7 at over 4 GHz. I can live with less than perfect sensors.

[nstarz]

Thank You for all the help. I guess I can choose to close all my temperature programs or roll the dice on an RMA.

Sorry it took so long for me to get all of this.

[WoodButcher]

Sound familiar Uncle? somewhere in this novella you said that.

RT won't work on older Intel, I have an 840EE, any way to make that happen?

[unclewebb]

RealTemp needs DTS sensors within the CPU to work and those are only in Core based processors. I've read Intel manuals that said the DTS sensors were introduced in the late P4 processors but I haven't found any yet. You can run my RT MSR Tool and Read MSR 0x19C and see if there is any data in there. It will likely be zero which means that there is nothing RealTemp can do. After you download the MSR tool just copy it into your RealTemp folder so it can find the WinRing0 library.



My old P3 laptop doesn't show any data that RealTemp can read. If you find any info hiding in this MSR in your P4 then tell me what it is so I can see if it is temperature data or not.

[WoodButcher]

Thanks, next time I play w/ that box I'll do that.
That pc is my teenagers so it isn't a priority, just curiosity.

[psychok9]

I was on another forum this evening battling against typical core temperature FUD. I did some testing which really shows and explains thermal throttling so I'm going to re-post it here. I enjoy programming RealTemp but it's always nice when I actually get the chance to use it. ***megacut***

So, is 95°c the real TjMax of Q6600 65nm G0 revision?

Thank you very much!

[Leeghoofd]

So, is 95°c the real TjMax of Q6600 65nm G0 revision? Thank you very much!

Nope think it's 105°C... 95°C for 45nm revisions... otherwise I've been reading wrong temps on all of my GO Q6600's...

[psychok9]

Nope think it's 105°C... 95°C for 45nm revisions... otherwise I've been reading wrong temps on all of my GO Q6600's...

On unclewebb test's, throttling do not enter at 93°c?

[WoodButcher]

Nope think it's 105°C... 95°C for 45nm revisions... otherwise I've been reading wrong temps on all of my GO Q6600's...

RT is reading my G0 @ 95 and I've adjusted idle by +1. perfect!

@ uncle, Here is the 840EE

[Leeghoofd]

On unclewebb test's, throttling do not enter at 93°c?

So it's 93 then ? not 95 ? GO Q6600 have been around along time, no idea why Coretemp and other temp related programs didn't adapt sooner then... maybe there are several TJcasemax versions out there...Intel is surely hiding stuff from us which has become more apparent with 45nm...

I love the work done here and give a big thanks to Unclewebb, but have to remain sceptical as all GO Q6600 owners in the past were 10 degrees off then ?

[WoZZeR999]

Uncle did testing on an actual Q6600, I believe that most other programs used Intel docs to find the Tjmax. I will use RT until someone can prove otherwise that it isn't 95c.

[Leeghoofd]

Look I'm not doubting anything, Pure logic sense now : I just would find it weird that my Q6600@3.6 at 1.42 volts would idle at 23-25°C then under watercooling if it's based on 95 and not on 105°C... Sadly I cannot prove anything as I sold that Q6600 ( think Vid was 1.2125 ) but I know what I've seen during 2 watercooling setups, idle temps in coretemp (based on 105) were 32-35°C idle which sound more beleiveable than 22-25°C unless I live in some cold country and my ambient is sub 20's...

Think my logic makes sense not ?