I request again, if possible, to have an idle calibration separate for any core, in order to correct my DTS sensors in E8400 that work but one of them is very badly calibrated.
Thanks, ;)
Printable View
I request again, if possible, to have an idle calibration separate for any core, in order to correct my DTS sensors in E8400 that work but one of them is very badly calibrated.
Thanks, ;)
+1. thanks for all your testing so far and looking forward to your Q6600 G0 results.Quote:
Originally Posted by Brama
Lol, this is funny.
I have my CPU clocked now at 3 Ghz, 1.25V
Speedfan and Real Temp 2.2 reports core temps of 19 degrees C
http://img246.imageshack.us/img246/6228/realtempap2.jpg
My water temperature is 23.5 degrees C
I know that they can't idle below ambient , rather water , shouldn't water heat those cores ?
So ... Please tell me why is this!? :D
Clearly the DTS sensors in your E6600 don't measure linearly over the full temperature range. Change the idle calibration until you have something sensible.
Re-read the instructions in the 1st post and adjust 'Idle Temperature Calibration' as necessary.Quote:
Originally Posted by Spacehead
Spacehead: I've been banging my head against the wall for the last year trying to convince people that the data coming out of the DTS sensors does not change linearly with changes in core temperature, especially at idle. Your post confirms that. Head to the first post in this thread and have a look at the nice graph I created to help explain this. My E6400 - Revision B2 does the exact same thing as your E6600. It reports idle temperatures that are too low and impossible to achieve.
Since developing RealTemp, I have not put my E6400 back in for an in depth study but I can say the (++) Idle Calibration setting was developed with my E6400 in mind. Using that setting should get you very accurate temperatures from Idle to TjMax at 85C and even beyond that. I just released v2.21 yesterday with a couple of bug fixes which should correct some issues that KTE discovered during his recent high temp testing between TjMax and thermal shut down.
Why don't you do us all a favor and drop your MHz down to 266x6=1600 MHz and drop your core voltage down to 1.10 volts and show us your idle temps with no calibration correction and then with a (++) correction. After correction your idle temps may not be perfect but they should be a lot closer to reality.
Maybe I'll do a thorough test of my E6400 similar to the one I just did with my E2160 while waiting for my Q6600 G0 to arrive. :D
Hmm, I'm using it right now and I'm getting 37 or 39C (Core 1) and 36C (Core 2) with Ambient or Room Temp. at 75F in East Texas. I guess that sounds about right with my Artic Freezer 7 Pro with Artic Silver 5 (Installed it Yesturday) so It should go lower than in the next few days. Oh and btw, its E6300 at Stock Speed with Voltage at the Lowest as possible on ASUS P5B Deluxe. Will overclock once my Thermal is necessarily to overclock my CPU.
Anyways, Read the Graph, I don't really understand it but I like to hear that is more accurate than Core Temp Software.
Thanks for sharing with us on this program! :)
I feel your pain, and it's only been a couple weeks for me. I've been following this thread closely, and it seems this pops up almost everyday.Quote:
Originally Posted by unclewebb
I can understand not wanting to read the whole thread, but when the info is clearly presented in the very first post it makes you wonder if some folks just d/l it, install it, and then post something. Come on people! At least read the first post!
unclewebb, maybe you should copy/paste the first post into a readme file and include it w/ the download. I'm sure a lot of folks won't read it, and will still post the same question, but it might get through to a few people.
Sorry for ranting! I just want unclewebb to stay focused on the ultimate goal and not be bothered w/ answering the same question over & over again. ;)
Great idea that I've been meaning to do for a while but haven't got around to yet. I enjoy the testing and data gathering too much that I get side tracked once in a while. :DQuote:
Originally Posted by jason4207
Thanks for coming to the defense of RealTemp. I'm trying to make this program the first program people go to when they want to know the core temp of their Intel Core processor. I'm a couple of years behind CoreTemp in development time and I'm still missing a feature or two but in terms of accuracy, I think RealTemp is already a step ahead for most processors.
i think your program offers more versatility...thank you for putting so much effort into getting it perfected!:clap:Quote:
Originally Posted by unclewebb
your post on the previous page along with KTE's was very helpful for me to understand thermal dynamics;)
I thank Unclewebb again and again for the development of this usefull program and for considering my request to add 2 curves with different slopes in the program; preliminary test is giving very similar temps in my strange E8400 that has the DTS of the 2 cores giving very different data.
Q6600 (GO) Real Temp vs Core Temp
Ambeint 18C
H20 cooled Triple Radiator 3x 98CFM
LAING D38 Vario #5
Swiftech Storm
P35C-DS3R (BIOS F10)
Open air (no differential between case temp and ambient)
Shown: IDLE at 9x333FSB (default Vcore1.28V)
Shown: LOAD using Prime95 (4xcores) dissipating approximately 128W
Where can you download version 2.21?
Page 1Quote:
Originally Posted by Richard Dower
Don't trust those idle temps until you calibrate them as per the 1st post (or several other posts since then).Quote:
Originally Posted by Liquid3D
he is on water, not AIR!
How many other users have E8x00 CPUs with two sensors that work but with one sensor that significantly trails the other until higher temps are reached like Brama had? He sent me a RealTemp log file that was an ugly mess at idle and the two cores are still way out even at 50C:
13:47:05 -6 25
13:47:30 -5 25
13:47:35 -4 27
13:47:40 -6 26
13:47:45 -6 26
13:47:50 -5 26
13:47:55 -6 25
13:48:00 -7 26
13:48:05 -6 26
I came up with a workable solution for him and his displayed temps are much better balanced now on his E8400 from idle to full load. Hopefully he can post a few lines of how his log file looks now. If your DTS sensors are seriously messed up like this then send me a PM and maybe I can write a custom profile for you as well. I need some more real data before I decide if it would be possible for users to create their own custom profiles.
He still can't have below ambient temps, though (unless that's chilled water). He said ambient was 18, and RT is reporting 16. Idle temps obviously need to be calibrated.Quote:
Originally Posted by Ace-a-Rue
How do you know what the actual DTS core temps are if DTS feedback is wildly wrong?Quote:
Originally Posted by unclewebb
I have an E8400 which shows the same problem. I'm currently in talks with Intel on it whether there's a fix to it, how accurate and trustable it is or whether they will provide me another one in replacement of it, so not yet testing it (have another CPU in that I'm testing).
Using "idle system" described on message #1.Quote:
Originally Posted by KTE
Please, let us know the policy of Intel about substitution of processors with stuck or crazy DTS sensors.
Some lines of my log after the very cool customization of unclewebb :DQuote:
Originally Posted by unclewebb
21:47:30 46 47
21:47:35 45 47
21:47:40 47 48
21:47:45 45 47
21:47:50 45 46
21:47:55 46 46
21:48:00 46 48
21:48:05 46 47
Now it looks like a normal cpu ;)
KTE: When working with Brama on his messed up E8400 he assumed, based on idle and load temps, that his second core was reporting what looked like appropriate temperatures while the first core was reporting temps that were WAY too low. The log file he originally sent me was showing a delta of 30C to 33C at idle and a delta of about 15C at full load.
I pulled out Excel and plotted some points of both his cores and it appeared that both sensors were working but the slope of the two curves or rate of change of the DTS data vs the rate of change of core temperatures looked considerably different.
I created a linear correction formula for Brama that is giving him excellent results from his normal Idle ambient temps to his typical full load temps using OCCT I believe. Without a proper calibration with an IR gun or calibrated thermal diode we'll never know how accurate his temps are but I think they are far closer to reality than what they were. He definitely was not idling at 5 to 7 degrees below zero like it was previously showing.
It will be interesting to find out what Intel has to say about this issue. The DTS sensors were not designed to report idle temps and as long as they work for thermal throttling and shut down control then I don't think Intel is really obligated to replace a CPU. If you complain enough they might replace CPUs on a one on one basis to keep their customers happy. Maybe they could hire me and I'll show them some junior high school math they could use to at least make the DTS data look believable near idle. :D
I'm going to post a log file later today showing how closely the two sensors in my 65nm E2160 track each other through several idle to full Orthos load runs. It very clearly shows how DTS sensors are supposed to work. Time to go look for a Q6600.
KTE, if any of your talks include technical information, would you mind posting their content here? I am assuming of course that you are communicating through correspondence.Quote:
Originally Posted by KTE
OK, thanks for the explanation unclewebb. :)
Dual core would be easier than quad core in that you could try and calibrate each core to be as similar as possible while keeping a linear correlation with load based off the working core DTS feedback. The temperatures between both cores at full load won't be much different as long as you can establish the TjMax and be idling above ambient at low TDPs. So, good going to resolve the issue there. I suppose it's most difficult only when both of the core DTS are wrong, a complete wild shoot in that scenario.
Intels stance is, the CPU won't be replaced unless the DTS are completely faulty and either of throttling/thermal control shutdown do not occur at their intended margins. But I'm not going to them as a normal customer, I'm going through as a major corporational partner as they work with my uncle very closely for over 15 years now, he's a personal friend with some there.. hence I get a little better treatment. :D
I've had my last two replaced already, the entire first delivery batch was replaced TBH. The boss didn't accept them as these parameters are very crucial in a professional environment and so Intel promised delivery of the C1 step first batch to us which fixes these DTS issues. However, we've not received them yet, I picked up an E8400 lying around but it's an old batch and still idled 10-15C sub-ambient with a 105C TjMax. I haven't played with it yet, only booted it at work in a spare system, my own P35 MB at the flat didn't support it without a new BIOS flash and I have no spare PSU for the system yet. So just waiting a little while before I test it after some guidance from the Intel S.FAE. Basically, the ES Wolfdales they had were known to have good working DTS. So he's just going to compare my temps to those of one working at the same Voltage x MHz using the stock cooler open air, and then I can have a good idea of the correct temps (approx anyway). Will report back when we're making progress.
Did you replace the file in the links? I re DL it and its still 1.1. The program when its opened is also 1.1. The notes arent updated either...