Nope, because those are straight on-die sensors readings.
With calibrating temps you're fooling yourself because the only thing that matter for your CPU thermal behavior is distance to TJMax.
Printable View
Realtemp is only running at 60% load average when running IBT at maximum with 16 threads on i7920.
Why?
If you leave threads on auto, it will only load 4 threads and give 50% loads. On mine setting threads to 8 gives 100% load, as does 16...though i920 only has 8 threads, so should just set on 8. Are you using most recent versions of both realtemp 3.30 and IBT v2.3. Also not sure what you mean by avg....ibt is an intermittent loading program, so will load at 100% for time, then drop off load, then repeat.
http://img43.imageshack.us/img43/9761/ibty.jpg
Yes but these sensors are bugged, not accurate. Therefor the Distance to TJmax is not accurate. This is the reading before adjusting the temps. If your core temperature's are off so is your distance to TJ Max. The 2 add up to 100 no matter what, if one is off they are both off.
http://img253.imageshack.us/img253/8126/realtemp.png
Barso: I know older versions of RealTemp had some issues reporting the correct load when running some high stress testing programs like LinX and IBT because those guys tended to run at a higher priority which prevented RealTemp from doing its sampling. I'm not a fan of programmers that do this but I was forced to increase the priority of RealTemp to compete with these programs for some CPU cycles for RealTemp. If you are having problems with RealTemp 3.30 then post some screen shots.
http://www.fileden.com/files/2008/3/...alTempBeta.zip
OC Nub: I think what burebista is trying to say is that the Distance to TJMax numbers that RealTemp reports are the raw data coming from these sensors. I've decided to report that as is. Ultimately, whether this data is bugged or not and whatever TJMax really is, this data is what controls thermal throttling. For that reason, users need to see the data coming from these sensors as is. When Distance to TJMax counts down to about 2 or 3 on a Core 2 based CPU, that's typically when thermal throttling begins. Throttling is not based on the actual temperature but only on whenever these sensors count down to almost zero. With Core i7 CPUs, they seem a little tighter and throttling won't begin until they hit exactly zero.
The TJMax spec is not consistent for each core, which Intel sort of agreed with at their IDF conferences last year, so the exact temperature of when thermal throttling starts is not exact either. If TJMax is 100C for two of your cores and 105C for the other two then thermal throttling will first start to happen on your core 2 and core 3 at a hair before 100C and throttling of the other two cores won't start to happen until the actual temperature is closer to 105C.
All RealTemp does is it allows the user to convert the data from these sensors into some somewhat meaningful core temperature numbers.
Thanks, that makes sense. I normally wont stress it at anything over 75c and then only for short times. If in Prime the temps reach 75c at the end of an 8k loop, then drop back down I will let it run but if it becomes consistently 75c its time to stop, at least for me. Who knows maybe the sensors are correct, still a good chip. I will take uneven temps all day long with a good oc on low cvore.
RealTemp takes a good stab in the dark at converting random sensor data into some meaningful temperature numbers but the exact temperature isn't that important. It's all pretty simple really. Run your CPU as cool as possible and the more you'll be able to reliably overclock it.
I learned the above fact during a Prime Small FFTs run on my E8400 C0 at 4050 MHz. It was running great for over 9 hours so I disconnected the CPU fan to see what would happen.
http://img30.imageshack.us/img30/960...osfailtemp.png
http://img263.imageshack.us/img263/9...edfangraph.png
Core Temp was using TJMax=105C instead of 100C back then so you need to subtract 5C from that temperature.
I set SpeedFan to TJMax=95C which was also wrong so you need to add 5C to the numbers in that graph for the correct temperature.
It only took about 5 minutes for a nice stable Prime run to error out. At 58C, it was 100% stable. When the core temperature got up to 77C, it crapped out.
Core i7 can run reliably when overclocked to 4+GHz at over 90C. I found with Core 2 based chips that about 70C was the upper limit with a decent overclock. The harder you push, the cooler you'll have to run them to remain stable but I think most people here already know that.
The old "disconnect the CPU fan while running Prime Small FFTs" trick is a good way to find out how much temperature head room you really have and how stable your system really is. :rofl:
Hmmm, now if I shut off the water pump it might be an interesting race as to which would melt down first, the 4870s or the overclocked i7 :). Of course I do have RealTemp set for a shutdown at 90c.
I will let you guys be the guinea pigs and take your word for it. Afraid to unplug my pump while running.
Pull off your waterblock, then you'll see thermal protection in all its glory. :D
I find it amazing that some people post on forums "I read an OC guide and went into the BIOS and disabled C1E, EIST and thermal control like it said. Please help me OC...". Do people actually recommend disabling the only thing between you and a burnt chip in overclocking guides?
Has this bench changed in recent versions? It was my understanding that the bench scaled pretty much linearly with mhz but I had a hard time passing rge's score.
http://lakesidepc.com/rtb.jpg
Hi loonym. Very nice XS Bench score. :up:
I almost never see any results from this bench so I haven't kept close track of what programming changes have effected it.
The calculation it does is small and fits in the cache so it is 100% CPU MHz, and cache speed sensitive. Compared to Core 2 chips, Core i7 can definitely crunch numbers faster at the same MHz.
The most recent changes were the ability to Cancel out while the bench is running. I know this slowed things down so it's likely that you had to work harder and overclock higher to match rge's old score.
You are now officially the new and improved RealTemp 3.30 - XS Bench world record holder. How does it feel to be famous? Unfortunately, in the land of free software, fame doesn't translate into much $$$$$.
It will probably be a week or two or three but I'll add your name and new record to the next release of RealTemp. The code is in the middle of a general house cleaning and I've been too lazy to work on it too hard during the summer. I won't be making any changes to the XS Bench code so your score will remain valid.
I'll send you a PM when the next version is ready so the world can see your new record, unless rge finds some colder air and gets his new i7-950 a little faster. Now he'll be motivated when he finds out. :rofl:
I was just wondering, do you know if your core sensors are stuck at -8C? In theory, Distance to TJMax can only count up to 127 so the lowest possible reported temperature would be -27C. 45nm core 2 sensors didn't have a chance in hell of getting near that temperature without sticking so I was just wondering how low a Core i7 sensor can go before crapping out.
Hi, i am using Real Temp for quite a long time, i always preffered it over core temp and other monitoring software. I have one small suggestion - is it possible to add an option "Run RealTemp on startup"? When i am doing stability tests or just testing some settings, i always boot up and first thing i do is open a monitoring software, so i thought it could come handy. Cheers.
Task Manager or registry edit. Pretty easy to do it yourself.
AFAIK Kevin is not a fan of software writing to registry.
Drag a link to RealTemp.exe into your Startup folder. In XP, click on the Start menu and it should be easy to find your Startup folder.
Remember, just a link to the exe that you want to start. Don't drag the whole program folder into there. :)
In Vista or Windows 7, this folder is a little more hidden but it's still there. Have a look in this folder:
C:\Users\user name\AppData\Roaming\Microsoft\Windows\Start Menu\Startup
I've found that this is the best way to start a program with the fewest issues and if you ever get sick of RealTemp, it's easy to go back into this folder and delete the link to it. I prefer not to add any items to the startup area in a users registry so they can drag the RealTemp folder to wherever they like on their hard drive.
The Task Scheduler in Vista or Windows 7 is also a great idea to control start up items but it's a little too complex for some. Here' a tutorial if you want to learn about it.
http://www.vistax64.com/tutorials/13...eate-task.html
These i7s run pretty cool if you undervolt them nicely. 0.16V below VID and the stock cooler manages this in Prime95 with HT and Turbo enabled:
http://i176.photobucket.com/albums/w...ry_cool_i7.png
Granted I'm running in the 70s and even up to 80C at stock.
Have installed Rivatuner 2.24 and RealTemp 3.30 and the Plugin RTcore but it only shows two of my four cores on the Q6600
RTCore.dll
http://img195.imageshack.us/img195/5...coreplugin.png
http://img233.imageshack.us/img233/5...repluginac.png
It is the same thing when I use the Plugin for Everest, only core 0, 1
How do I make it see all four cores?
Does RealTemp 3.30 show all 4 of your cores? Do all 4 cores show up in the Device Manager under Processors?
I haven't seen this problem before with the RTCore.dll plugin and it's strange that the Everest plugin would be doing the same thing.
Open up the Settings window in RealTemp and tell me what it shows for APIC ID. It should be something like 0123.
It shows the four cores in RealTemp and Device Manager and Everest but not with the plugins in Rivatuner 2.24
Device Manager
http://img219.imageshack.us/img219/1108/dmcores.png
RTsettings
http://img219.imageshack.us/img219/8158/rtsettings.png
Everest ...which shows 17C idle temp on Core 4 but thats normal ;)
http://img29.imageshack.us/img29/9012/everestcores.jpg
...all four cores shows up in Everest
RealTemp
http://img41.imageshack.us/img41/186/rtcores.png
I'm running forceware 185.85
Might need to reinstall Rivatuner
Hello unclewebb. I am running Windows 7 RC1 and have tried both methods, adding to startup folder and using task scheduler but whatever I try, it refuses to run at startup. If I use the task scheduler, when I look at the list of running processes, Realtemp is listed but the program itself does not actually autostart. Do you have any suggestions for me?
Where do you have the RealTemp folder located? Do you have UAC enabled? Give me a few more details so I can try to duplicate the problem you're having.
hi Uncle, just a Q: i don't understand what does the 'idle calibration' can you explain it? thanks
The temperature sensors do not respond linearly to changes in core temperature. The Core i7 sensors are very good and changes in actual core temperature seems to track changes in these sensors very closely but the 45nm Core 2 sensors can be horrible.
Intel calibrates these sensors somewhere around 100C and I find that core 0 is usually very accurate at that point. The further you get away from the calibration point, the more the amount of error increases. Even though TJMax is correct, the reported temperature at idle might read 5C or 10C too low or too high. Every CPU is unique so I gave users a way to correct for this error at idle by using some Idle Calibration adjustments.
If users saw errors in their reported temperature at idle, they used to increase or decrease TJMax but all this did was move the sensor error from idle up towards TJMax so now their load temperatures would be reported incorrectly. Idle calibration lets you correct the temperature curve of your sensors without having to change TJMax which will screw up your full load temperatures.
The data from these sensors might change 9C or 11C for every 10C change in core temperature. The +/- 10% accuracy number was a rough ballpark number from Intel's presentations last year at IDF. The amount of error might be worse than that. When you get 60C away from the calibration point at idle, it's easy to see that a +/- 10% error in these sensors can become significant.
Most users don't bother to calibrate their sensors because it's not an exact science. It was just nice that Intel finally admitted that these sensors are far from perfect with errors at the TJMax calibration point, errors in the slope of the temperature curve as well as sticking sensors at low temperatures.
thanks uncle ;)