Real Temp - New temp program for Intel Core processors

[Cumulonimbus]

Thank you to give us RT GT 3.70. Has RT now command line Switches to enable / disable C-States and Turbo?

[nksharp]

Whats the easiest way to make the program start on launch?

[Sauron_Vie]

windows autostart ?

[unclewebb]

Cumulonimbus: I don't ever use the command line so I don't plan to add any command line switches to RealTemp.

nksharp: For Windows autostart, I use the Task Scheduler.

http://forum.notebookreview.com/hard...ml#post6865107

If you are an Administrator on your account and you are not using UAC then try dragging a link to RealTemp into your Windows Startup folder.

[SS_The_Demon]

Hi, need some help here. I just bought a used Q9450 (rest of the system is in sig) which I installed yesterday and very much intend to overclock. Upon installing the CPU I first cleaned it and also my HSF with Arcticlean, or whatever it's called, and put some AS5 on the CPU. I set all voltages in BIOS to minimum value, except the vcore and multi which I left on auto. I also set speedstep and the other CPU features to enabled, and set the FSB to 333. When I got into Windows I started Real Temp and noticed my temps were really weird, which they never were with my old 65nm E6600. I ran IBT and it seemed stable, so I started reading a bit about the C2Q sensor problems and got here. So I ran the sensor test in RT next, with the settings described above:



At this point I thought maybe something was wrong with my installation so I turned the case on the side, no difference. I then removed the HSF, cleaned it and the CPU, and then put some NT-H1 on it. At this point ran the sensor test again, using the same settings as before and here are the results:



And nothing. Rebooted into BIOS, set vcore to 1.3875V (stock seems to be 1.2375V according to RT) and booted into Windows. Ran the sensor test a third time and guess what? This:



THE F is going on? What readings am I supposed to trust here, none of them?

[Peakr]

Will Sandy Bride-E support be coming soon to RealTemp?

[donmarkoni]

Will Sandy Bride-E support be coming soon to RealTemp?

SB-E IS supported. Go one page back and read.

[unclewebb]

What readings am I supposed to trust here, none of them?

The temperature sensors that Intel used on there 45nm Core 2 processors were junk. They get stuck at lower temperatures and have a pile of error when trying to report full load temperatures. Intel was always a little sheepish about just how bad they were. I have never seen any official specs for how much error they have but I know from experience that it is significant.

You are correct. On some of these 45nm CPUs, you can't trust any of these sensors. Intel only intended these sensors to be used for thermal throttling and thermal shutdown control and for those two purposes, these crappy sensors are more than good enough. Intel spent a few more pennies on the Core i sensors but they are still only intended to be used for thermal control. None of them were ever intended to be used for 100% accurate temperature reporting.

As far as I know, RealTemp should work on the Ivy CPUs too.

[SS_The_Demon]

The temperature sensors that Intel used on there 45nm Core 2 processors were junk. They get stuck at lower temperatures and have a pile of error when trying to report full load temperatures. Intel was always a little sheepish about just how bad they were. I have never seen any official specs for how much error they have but I know from experience that it is significant.

You are correct. On some of these 45nm CPUs, you can't trust any of these sensors. Intel only intended these sensors to be used for thermal throttling and thermal shutdown control and for those two purposes, these crappy sensors are more than good enough. Intel spent a few more pennies on the Core i sensors but they are still only intended to be used for thermal control. None of them were ever intended to be used for 100% accurate temperature reporting.

As far as I know, RealTemp should work on the Ivy CPUs too.

So in my case, you don't think I can trust any of the readings, right? I mean considering that I raised vcore from 1.2375 to 1.3875 and the temp hardly rose on any of the cores, with the exception of core 1 which rose 8 degrees. I would think that the temp should rise alot more then that considering the huge increase in voltage? Not that it seems to matter because I can't seem to get the damn thing stable at even 3.3 GHz...

[unclewebb]

Core 0, the core on the far left of RealTemp is probably the best of a bad bunch but there is no way to tell for sure. The 2 cores on the left appear to have some significant slope error where they change at a different rate than the temperature changes and core 1 has some issues with getting stuck at lower temperatures.

If you can't get one of these CPUs stable at 3.3 GHz then you probably have other problems besides core temperature. Run CPU-Z while Prime95 is running and see what is reported for actual voltage. Some boards are not great when it comes to overclocking Core 2 Quads and crap out when the bus speed goes over 400 MHz. I have an old Asus board that can run a Core 2 Duo reliably at a bus speed of over 500 MHz but falls on its face when trying to overclock a similar Core 2 Quad. I ended up buying a QX9650 for this board so I could overclock by adjusting the multiplier higher which allowed me to keep the bus speed at 333 MHz for better stability.

[SS_The_Demon]

Core 0, the core on the far left of RealTemp is probably the best of a bad bunch but there is no way to tell for sure. The 2 cores on the left appear to have some significant slope error where they change at a different rate than the temperature changes and core 1 has some issues with getting stuck at lower temperatures.

If you can't get one of these CPUs stable at 3.3 GHz then you probably have other problems besides core temperature. Run CPU-Z while Prime95 is running and see what is reported for actual voltage. Some boards are not great when it comes to overclocking Core 2 Quads and crap out when the bus speed goes over 400 MHz. I have an old Asus board that can run a Core 2 Duo reliably at a bus speed of over 500 MHz but falls on its face when trying to overclock a similar Core 2 Quad. I ended up buying a QX9650 for this board so I could overclock by adjusting the multiplier higher which allowed me to keep the bus speed at 333 MHz for better stability.

I just lapped the CPU, which seemed to lower the temps a bit, but again it's hard to tell since the sensors don't seem to work. Testing 3.3 GHz at the moment, using 413 x 8. Several days ago I had it running at 458 x 7 so it doesn't appear to be the board that's holding me back. Using a P5E Deluxe btw. Vdroop seems to be the biggest problem atm. Setting a voltage of 1.3875 in BIOS which I'm using for 3.3 GHz currently, let's me idle @ 1.368V according to CPU-Z. While running IBT it drops down to 1.304V however, with LLC disabled that is. And I would prefer not using LLC since it causes my CPU to degrade even more. And since Intel's spec says max voltage shouldn't be more than 1.3625V I'm hesitant to raise it more, since I'm already slightly above that at idle. So sure, I probably could go higher if I wanted to, unless temps become a problem, which I have no idea of knowing.

[WaterFlex]

Dear, unclewebb. Couldn`t you write Real Temp Mac OS version? Is it real?

[Eggy]

How is the 22nm support on the latest Real Temp, i know that it reads the temperature and everything appears to be OK, but myself and many others are quite surprised in terms of the temperatures reported. I've tested it with 2 3770K's and a 3570K. On the 3770K's i'm seeing close to 80C with just 1.35v, i'm using chilled water so the water temp is 10c under load, in other words i have a 70C Delta T between the reported temps and the water block at just 1.35v. On the 3570K i got a load temp that peaked @ 68C (62.5C on average between the 4 cores) with 1.4v, with 2600K with HT off i got an average or 37.75C with the same voltage. Is it really possible that the 22nm chips are 25C hotter on average between all 4 cores compared to the 32nm chips. See my test results under:

Load with Prime95

ff1dd25b_Thermal.jpeg

@ 1.1V IB runs 4,25C hotter then SB with HT off. (16.03%)
@ 1.2V IB runs 10,75C hotter then SB with HT off. (36.44%)
@ 1.3V IB runs 15,00C hotter then SB with HT off. (44.11%)
@ 1.4V IB runs 24,75C hotter then SB with HT off. (65.56%)

Also the TJ Maxx is reported as 110C, where is this taken from, and could it be that the programs are reading wrong from the sensors considering that this are all new chips that is not yet released? (Getting the same temps in Core Temp btw)

Thanks.

[stasio]

TJ Max for Sandy is 98C.
TJ Max for Ivy is reported as 105C.

[WFO]

IB runs much hotter than SB. It's been reported everywhere. Even here.

"Ivy Bridge exhibits much higher temperatures during full load due to its 22nm process, which will probably only get better though cooling optimizations and better contact between the HIS and the CPU Die."

http://www.xtremesystems.org/forums/...s+Overclocking

[TEDY]

what is latest version?

[Russ_64]

Real Temp 3.70
January 16, 2012

http://www.techpowerup.com/downloads...Temp_3.70.html

I think it is this one.

[-X-hellfire]

It would be nice to have a Min/Max CPU temp when hover on systray like the GPU Min/Max, will we see it in the next update, tnx?

[CrazyNutz]

Hi Uncle

I would like to ask do you have anything documenting RTCore.dll. I am writing a plugin for LCD Smartie to get CPU core temps, and I would love to use your library if that is ok with you.

Thanks

[Jonny5isalivetm]

Hi

If I remember correctly Real Temp gave an approximation of watts usage on Sandy Bridge ?

Can we has this for Ivy ?

Thankies

[unclewebb]

If I remember correctly Real Temp gave an approximation of watts usage on Sandy Bridge?

Intel came up with a great idea so monitoring software could easily read power consumption data directly from the CPU. This data is used internally by Sandy Bridge and Ivy Bridge CPUs to control the amount of Turbo Boost. Unfortunately, this data is based on the VID voltage and not the actual core voltage. If you go into the bios and change the CPU core voltage manually, this estimated power consumption number will not be accurate. I will try to add this data back to RealTemp in the near future for Ivy Bridge CPUs.

RTCore.dll was designed to work with RivaTuner. When RivaTuner was replaced by MSI Afterburner, I stopped working on the RTCore.dll plugin. Reading the core temperature of a Core i CPU is trivial. You would be better off just writing your own code but you can try and use the RTCore.dll however you like.

[donmarkoni]

Intel came up with a great idea so monitoring software could easily read power consumption data directly from the CPU. This data is used internally by Sandy Bridge and Ivy Bridge CPUs to control the amount of Turbo Boost. Unfortunately, this data is based on the VID voltage and not the actual core voltage. If you go into the bios and change the CPU core voltage manually, this estimated power consumption number will not be accurate. I will try to add this data back to RealTemp in the near future for Ivy Bridge CPUs.

What if you take the power consumption, divide it with CPU VID, and multiply it by vCore? Wouldn't it give approximate real power?

[thinking out loud]If you read first one, then before you read second one everything drops, you could get abnormal result, very high power consumption.[/thinking out loud]

[unclewebb]

If you were doing some Prime95 or similar testing where the VID and actual voltage were fairly consistent, you probably could come up with some sort of correction factor.

Power consumption in a CPU is proportional to voltage squared. If VID voltage was 1.20 volts and actual voltage was 10% higher at 1.32 volts then the correction factor would be:

1.10 x 1.10 = 1.21

In that case, actual power consumption would be about 21% higher than what RealTemp is showing.

Does any software show actual power consumption based on actual voltage? Any monitoring software that uses the Intel method and the Intel recommended power consumption register is not accurate.

[Jonny5isalivetm]

Okay thanks for your efforts, excellent program :-)

[Nightblade]

Hi Unclewebb, does RT support Rainmeter? If not, may I request it?