RealTemp General Discussion

[unclewebb]

Your memory is running at 560 MHz according to AIDA64 but it if you look at the Extreme Memory Profile, it should be capable of running reliably at 800 MHz with much tighter memory timings.

Different users have different goals when overclocking. Some are happy at the level you are at while others aren't happy until they are on the edge of self destruction. If you are happy at the present level then you could probably lower your core voltage which should help reduce your full load temperatures. If you intend to overclock further then increase the bus speed some more.

If you need some more information about overclocking then try posting in that forum. There are a lot of TechPowerUp users with a lot more experience than I have when it comes to overclocking the Core i7-920.

 

[Arctucas]

@unclewebb,

I believe I have found a bug in the 3.67 version.

I ran the XS Bench, and the result reports my CPU speed incorrectly.

Anyway, thank you for continuing to provide an excellent application.

 

[unclewebb]

The MHz number in the XS Bench section is based on the 25 multiplier which is the maximum turbo multiplier for a Core i7-950. Does it always report this Speed or was that a one off freak occurrence?

Are you using any other monitoring software on your computer? Intel provides some extremely accurate timers in their CPUs for monitoring purposes but unfortunately they are not protected so other software can manipulate these timers which prevents other monitoring programs from getting reliable data out of them. Everest/AIDA64 randomly starts and stops the system monitoring timers that I used to use so I've been forced to switch to a secondary set of timers. Unfortunately I don't yet know of any other software that is using these timers so that could also be the source of this problem.

I'll have a look at my code and see if I can see any obvious problems. Thanks.

 

[Arctucas]

The MHz number in the XS Bench section is based on the 25 multiplier which is the maximum turbo multiplier for a Core i7-950. Does it always report this Speed or was that a one off freak occurrence?

I downloaded and ran the 3.67 version. I re-ran it three times when I noticed the frequency discrepancy and got the same result each time.

I shut the PC off for a couple hours, restarted and ran it again and it reported the correct frequency, but the score is different (better actually).

Are you using any other monitoring software on your computer? Intel provides some extremely accurate timers in their CPUs for monitoring purposes but unfortunately they are not protected so other software can manipulate these timers which prevents other monitoring programs from getting reliable data out of them. Everest/AIDA64 randomly starts and stops the system monitoring timers that I used to use so I've been forced to switch to a secondary set of timers. Unfortunately I don't yet know of any other software that is using these timers so that could also be the source of this problem.

I have AIDA64 v1.60.1339Beta installed. That causes a conflict, then?

I'll have a look at my code and see if I can see any obvious problems. Thanks.

I appreciate your help, thanks.

 

[unclewebb]

Your screen shot above shows the 25.0 multiplier which is usually a good sign that some other monitoring software is not using the shared CPU timers in a friendly manner.

If you're bored and want to do some testing, you can disable the AIDA64 side gadget and also make sure that AIDA64 isn't running in the background. You can run RealTemp with the log file option on for 5 or 10 minutes without AIDA64 running and then do the same thing again with AIDA64 running and you might be able to isolate the problem.

If RealTemp randomly reports the full 25.0 multiplier only when AIDA64 is running then you won't be able to run these 2 programs at the same time.

 

[Arctucas]

Your screen shot above shows the 25.0 multiplier which is usually a good sign that some other monitoring software is not using the shared CPU timers in a friendly manner.

If you're bored and want to do some testing, you can disable the AIDA64 side gadget and also make sure that AIDA64 isn't running in the background. You can run RealTemp with the log file option on for 5 or 10 minutes without AIDA64 running and then do the same thing again with AIDA64 running and you might be able to isolate the problem.

If RealTemp randomly reports the full 25.0 multiplier only when AIDA64 is running then you won't be able to run these 2 programs at the same time.

Hmm... I have not used RealTemp and AIDA64 at the same time, and have never used the sidebar gadget. In fact, I have disabled the Widows Gadget Platform altogether.

 

[Arctucas]

I tried running AIDA64 and the RealTemp XS bench simultaneously:

I appears that for whatever reason, the multiplier issue resolved itself.

 

[peterwise]

Q8200 processor core differences

I last posted a month on so ago under the thread 'Real Temp Feature Requests'.

I decided that I should finish my RealTemp 'SensorTest' calibration before trying ThrottleStop 2.99.6 as recommended. This was only after discovering a number of strange things about the processors on my Intel Q8200 chip which it may well be worth exploring.

Most of my problems happen when running Trainz 2004 (a railway simulator) or, at least, that is the program that shows up the problem most clearly. Other applications like Firefox and ArcGIS are clearly impacted (Firefox starts using 25% total processor, especially when left on the Gmail page for some time, until it is restarted), but Trainz 2004 actually hangs the system such that even ctrl-alt-DEL or Alt-Tab will take 5 minutes to respond (one piece at a time). The program admittedly was probably not written for multiple processors and noticeably runs entirely on one processor at, say, 90-100% of that processor's time (or up to 25% of the Quad's total time) with an associated process running maybe just 0-2% of the Quad's total time.

Because I seemed to be having problems with CPU0 and/or CPU1 (it appeared that the temperature sensors for one or both may be stuck at low temperatures they always seemed to show 62C even when busy), so I tried moving all my users processes off CPU0 and CPU1. To my surprise they both show temperature BELOW 62C (e.g. 56C). Trainz 2004 did seem more stable on those processors as long as I had the RealTemp alarm set for a RealTemp displayed temperature of 69C and paused and waited when that sounded. If the room temperature was 19C, the alarm never sounded and the sim kept going forever. At a room temperature of 21C the alarm usually sounded (and the sim sometime failed within an hour, but NOT EVERY TIME) if I had the Intel stock heatsink's at fan 90-100% and an external fan pushing air directly onto the processor through the (closed) case. It's this random element that gets me, plus the fact that the temperatures of CPU0 and CPU1 will go down but (show) rarely up. I am using EasyTune 6 to control the CPU fan. I wonder would it be taking an average of the temperatures, the maximum or simply monitoring one or two? If fan speed is being controlled from the measured temperature of CPU0 and CPU1 only then this could explain the strange stability to those temperatures and also may help to explain the randomness of the shutdowns (they were even more random when any of the four processors could be used by the sim.

I have a general question about RealTemp: what is the meaning of the (in my case) four 'Movement' readings on the sensor test page (which for me never show numbers greater than 0 for CPU0 or CPU1)? Also, I note my "thermal status" is "OK" (for all four), but given the difficulty of getting the Intel stock heatsink and fan locked down to the motherboard in four places at the same time (with about the same pressure) and so getting an evenly thin spread of gel across the surface, I am wondering how the four processors are actually placed on the Q8200 chip. Are they in four quadrants from the centre, as one might expect? If so, it would be really easy to see how some processors could be behaving quite differently especially as, in my case, when re-fitting the Intel stock heatsink/fan, one particular clip always seems to stick and click in later and only after more pressure has been applied to it than the other three. This would seem to almost guarantee some differences between quadrants in thermal conductivity to the heat sink.

Having said all that I have run the Sensor Test several times making adjustments after each and am not sure I'm actually getting anywhere, would someone look through my results on the attached Excel spreadsheet and let me know? As you will see CPU2 and CPU3 are virtually indistinguishable, CPU0 seems to want to stay non-linear, but CPU1 is doubly so. I have only managed to bring the difference between the distance to TJmax of CPU1 (when compared to CPU2/3) down from 4C to 3C by the most stringent of modifications to Idle1 and TJmax1 (as per my spreadsheet).

At the end of my tests I also measured the temperature inside the case (and the measured temperatures and CPU2/3 haven't gone back down to the "normal" 61-62C since I closed the case an hour ago), and also I wanted someone to give me some idea of what the Nvidia System Monitor (that I used to use to measure temperatures) might be trying to say. It numbers the processors CPU1-4 not CPU0-3 and one would probably assume that their CPU1 equals our CPU0, but somehow the way the temperatures change seems so different - as if they are referring to different processors (not just "distance to TJmax" differences)! Can anyone explain?

Should I be assuming that my "real" processor temperatures (at 0-1% total CPU usage) should be nearer the 56C than the 62C mark (at c. 20C external and 22C in-case) ?

Any help will be most welcome.

Peter.

 

[peterwise]

Q8200 processor core differences

The attachment - sorry!

 

[peterwise]

Q8200 processor core differences

Oh yes! I also noticed another strange thing about my shutdowns, by the way. I get far LESS when MORE processes are running! I presume this means it's not (generally) the absolute temperature of any of the cores that is the problem, but rather the differences between cores. I should also point out that if whether it is CPU2 or CPU3 that is doing the lion's share of the processing, the temperature of BOTH always goes up together, whereas CPU0 and CPU1 register the same old "cool" 61-62C.

I know someone will ask me: I believe have (some time ago now) tried running the sim on CPU0 or on CPU1 and I have a feeling (at least one of them worked OK), but I'd have to record my results more carefully. I definitely remember doing something like that expecting it to fail immediately and being surprised - but it was probably really late when it happened and I was too tired to think clearly about it, much less to note it all down carefully!

Peter.

 

[unclewebb]

This article over at Tech Gage has a nice picture of how a Core 2 Quad is laid out.

http://techgage.com/article/intel_core_2_quad_q9450_266ghz/

A Core 2 Quad internally consists of two separate Core 2 Duos placed side by side with a little bit of space between the pair in the middle. The two individual cores in each Core 2 Duo are so close together that when you fully load one core, the core right beside it will heat up to almost the same temperature over time even if it is doing absolutely nothing. The other Core 2 Duo under the heat spreader will also heat up too.

Most users have never bothered using RealTemp to calibrate the temperature sensors. Intel never designed these sensors to be used for 100% accurate core temperature reporting. They are only designed for thermal throttling and thermal shutdown control and all of the ones that I've seen are adequate for that purpose. When you have sensors that stick at lower temperatures like all of these sensors do, it's pointless trying to come up with some sort of perfect calibration so I don't bother with calibrating these sensors. Your testing shows that your first 2 cores, Core 0 and Core 1, are both getting stuck at 57C and are not capable of reporting any temperatures lower than that. You can't calibrate sensors that are stuck.

Depending on Windows, most single threaded applcations that are run on a Quad Core processor have the task randomly scheduled on a variety of different cores. This can be happening hundreds of times a second so the Task Manager is usually too slow to accurately show this. You can use the Task Manager Set Affinity... function to lock a task to a single core but you often times get lower performance when you do this.

I have no idea where your problems lie. Trainz 2004 may not be a 100% stable program or your computer might not be 100% stable. I usually run testing programs like Prime95 or LinX to try and get an idea how stable my computer is. If my computer is stable and I have trouble with one program; it's usually the program that is the problem.

If you are a fan of Trainz then why not consider upgrading to a newer version to see if they have got the bugs worked out.

http://www.auran.com/TRS2004/default.htm

When you install a heatsink with the Intel style push pins, you should try to push two pins in at a time. Push the two that are diagonally opposite to each other. If you push in one pin at a time, the last pin can be a pain in the butt trying to get to seat. I installed one of these on a friend's Gigabyte board and it took all of my 200 pounds to get pin #4 to finally go in. When using a heat sink that has push pins, I try to install the heatsink with the motherboard outside of the case so I can look on the back side of the motherboard to make sure all 4 push pins are fully seated.

If you ever decide to reinstall your heatsink, remember to completely clean your CPU and heatsink and use some new thermal paste. That might improve your temperatures. Your core temperatures seem a little high but they are well within the Intel specifications for this CPU. Do some testing to see what is causing your problems. Lots of memory fails over time or is not set to the correct voltage or the correct timings in the first place and might need to be adjusted. CPUs are very reliable compared to memory and power supplies and video card drivers and old software from 2004.

 

[peterwise]

The "two pairs" type architecture that the picture shows and you describe does explain what I'm seeing. I presume 0 & 1 are on one side and 2 & 3 are on the other.

Is there no way of just switching out or turning off CPU0 and CPU1 for ALL processes - I already do it for individual ones. (I've asked before here I think). Actually I know of at least one process than will not take anything but CPU0. If I change the affinity for GUI.EXE (presumably a GUI for Gigabyte's GSvr.exe) it just quietly slips back to CPU0, Situations like this, as well as no doubt parts of the XP Pro system software, would mean that anything interfering in this way (if it is even possible) would have to effectively map or renumber the cores and the maximum number of cores e.g. 2 & 3 of four cores would become 0 & 1 of two cores.

Actually, I just ran the sim for several hours with all my own processes ignoring CPU1 (only) and all worked OK. CPUs 2 & 3 warmed up only about 4 degrees C (CPU0 pretended it was cool with the whole affair) and, for the first time I noticed that one CPU wasn't taking a full load almost alone - the load was more or less evenly dispersed among the three. Is CPU0 still problematic you think? It seems like my main problem is CPU1.

Why did you say that 0 and 1 are getting stuck at 57C? Where did that number come from? Is that (TJmax0 - Dist0) or (100 - 43.0) towards the end of my spreadsheet, but then what about the Idle0 of 4.0?

Some time ago now, I did change carefully clean off and replace the gel with fresh gel after I read an excellent review/tutorial on the subject by Olin Coles at

http://benchmarkreviews.com/index.php?option=com_content&task=view&id=170&Itemid=38

I really liked his experimental and investigative approach and it gave me confidence to give it a try and err rather on the side of too little than too much gel. In the case of my processor and stock heat sink, one small drop dead centre seemed appropriate.

You said that "all of these sensors" stick at low temperatures. Do you mean that all the Intel Core2 processors sensors tend to (presumably) stick. Which affordable processors do not and take a balanced approach in their architecture. My grandson might be buying me a motherboard for my birthday, because the one I'm using is his hand-me-down and he felt it might have been damaged (anyway it's v1.0 of a Gigabyte board that they never made a later version of)! Also it has just 2GB of 2x333 (667) Mhz memory which may be a bottleneck for a 2.33GHz Quad processor. However, you seem to be suggesting that I should throw away my processor and get another anyway. It all sounds expensive!

I have tested the memory with free DOS type utilities, but would welcome advise of what type of (ideally free and Windows) utilities I could use. Wouldn't Prime95 or LinX just test the processor's stability? I think I know about the processor now - it passes Intel's "Processor Diagnostic Tool" tests, but they do not seem that stringent anyway.

Actually, the reason I am now BACK on Trainz 2004 is that it DOES work (most of the time) and, no doubt, because of my PC's problems, my Trainz 2010 has stopped working and would not reinstall (I tried five times). It is very database intensive and my PC keeps messing up the installation and patching procedure (which runs overnight!).

I've thought about getting Trainz 12, just coming out now, but want to ensure my PC is working right first.

Thanks anyway.

Peter.

 

[dlongnecker]

unclewebb
Please update the RealTemp GT to the 3.67 version for the six cores Intel CPU.
Now I necessary to use the GT 3.59

What do you mean by this comment? I am running RealTemp 3.60 with a 6 core processor and get a message that says the processor detected is not supported. I don't see a 3.67 version anywhere..

 

[stasio]

What do you mean by this comment? I am running RealTemp 3.60 with a 6 core processor and get a message that says the processor detected is not supported. I don't see a 3.67 version anywhere..

RealTemp 3.67
http://www.mediafire.com/?n99nq4kn95u6i6a

RealTemp GT 3.61
http://www.fileden.com/files/2008/3/3/1794507/RealTemp_361.zip

 

[dlongnecker]

RealTemp 3.67
http://www.mediafire.com/?n99nq4kn95u6i6a

RealTemp GT 3.61
http://www.fileden.com/files/2008/3/3/1794507/RealTemp_361.zip

Thanks.....none of those versions worked with my processor. Guess I am missing something.

 

[unclewebb]

If RealTemp does not work with your processor why not post a CPU-Z screen shot so I can see what processor you have? Does RealTemp start up and display an error message? Does it immediately crash? Are you using some overly aggressive anti-virus software that is blocking the open source WinRing0 driver that RealTemp uses? Do you have Administrator privileges? When you say, "none of those versions worked", that could mean just about anything.

It seems to work OK on Cryptik's Core i7-990X

 

[dlongnecker]

Sorry if I wasn't specific enough. Attached is the screen shot from the CPU-Z and the error message I get when I run the RealTemp program.

 

[unclewebb]

Sorry for the confusion. RealTemp has always been for Intel Core processors only. AMD CPUs are not supported because I haven't owned any AMD hardware in years. Don't get me wrong. The Phenom II X6 is a great CPU at a great price but I've never had the time or money to start a new development program to support them. Maybe give Core Temp a try.

 

[Mussels]

yeah, realtemp has never worked on AMD. you need coretemp for AMD support.

 

[Eric_On_Web]

Not working on my skt 478 also.

 

[burebista]

From here:

Main Features

Reads temperature information from all Intel Core based processors. Pentium 4 processors are not supported.

 

[Eric_On_Web]

Thanx and thats sucks

 

[halninekay]

It's a great tool! Thanks for that. I tested it on my first and brand new Intel rig.

 

[lyna]

CPU Cool Down Test

Hi, My notebook tries to make scandisk each time when windows starts, but it can't, it always stay the same black screen which is seen just before login page. And nowadays, it shut down randomly, I think cause of temp. I made CPU Cool Down test and took the result seen below. Can any one help me to understand that result ? is my CPU under threat ? can be any relationship making scandisk with that result ?

 

[unclewebb]

Your laptop has a serious overheating problem. The LOG word means that it is thermal throttling at full load which slows it down to a crawl. The Cool Down Test shows that it is throttling so bad that it is only running at half of its intended speed.

Pull your laptop apart and fix this cooling problem. It would be a good idea to reinstall the heatsink with some fresh thermal paste and clean out the dust bunnies.