Real Temp - New temp program for Intel Core processors

[unclewebb]

I understand the Not Responding!

I knew you did. I just put that in there for everyone else that might be wondering what's going on when that happens.

Thanks for the feedback on version 2.58.1. RealTemp has been 99.9% crash free throughout its lifetime, including the beta releases so I didn't want 2.58 hanging around too long. It's been shot.

[Talonman]

XS Bench is working on Vista 64 bit too. Both in Gamer Mode, and not in Gamer Mode.

[unclewebb]

Now we need some gamers to see if this feature works in modern games. It was OK in 3DMark06 and I did a few laps in Ford Street Racing last night but I'm too old for playing fast moving shoot 'em in the head kind of games!

[rge]

Gaming mode works great on desktop, but the temp display flickers more off than in COD4, and is barely readable, sometimes half is missing or pixeled/grainy. One pic captured it full on, another as flickering off (pics right)

In Assassins creed (left) and crysis it flickers more on than off, so that it is readable and camera captured it on.

Edit...Actually, it flickers with frame rate. Frame rate 30-40, flickers but able to read, frame rate 90 in COD4 it is unreadable most of time until frame rate slows.

[MrHydes]

can real temp read well 45nm?

[unclewebb]

Thanks rge. If the frequency of updates by RealTemp is similar to the FPS of your game then it's quite visible but if the frequencies are quite different then it flickers too much and is unreadable at times. It's not perfect and probably never will be but I think it's a usable feature as is. When you can see it in game, it shows information that is not otherwise available. I'll play around with this option some more and maybe even try to come up with user selectable refresh rates that might help visibility in different games.

Gamer Mode wasn't causing the crashes in the brief 2.58 version. That was caused by my changes to the real time frequency calculations. When this feature was first introduced there were some problems on some boards with the MHz being reported about 3% too high compared to CPU-Z. Can someone that used to have this problem select the CPU MHz option in RealTemp and see if it works properly now or is it still not quite right? This option updates the MHz display automatically if you are using SetFSB. It works 100% on my computer and on the laptop I tried it on.

MrHydes: can real temp read well 45nm?

I believe RealTemp can do a better job at this than other software and users here seem to agree. There are a lot of problems with the sensors in some 45nm chips so if you're having problems, post some results so we can have a look.

[loonym]

The 2.58.1 also fixed the sensor test not working 100% correctly for me with the 2.58 and my 45nm x3360 C1. Once again you've outdone yourself unc

[MrHydes]

I believe RealTemp can do a better job at this than other software and users here seem to agree. There are a lot of problems with the sensors in some 45nm chips so if you're having problems, post some results so we can have a look.

By MrHydez

check it plz

[euonline]

Now we need some gamers to see if this feature works in modern games. It was OK in 3DMark06 and I did a few laps in Ford Street Racing last night but I'm too old for playing fast moving shoot 'em in the head kind of games!

Uncleweb:

Game mod is not working very well in UT3 + Vista64 + 8800GT NVidia (ForceWare Version 175.63)

It keeps blinking during game.


edit: CRISYS, I can only see the temps (gamemod) if played in windowed mode of the game. IN Full screen you can not see temps.


I will test some more games and post here.

Great job, thanks!

[unclewebb]

It keeps blinking during game.

There's no way around the flickering as far as I know so far. The temps have to be updated at the same rate as the game is updating the screen but that is constantly changing and I haven't found any way to exactly sync to the game frame rate.

Depending on your frame rates, in some games it works well enough that it's easy to see your temps while playing while in other games or at different times during some games, you might not be able to see anything. If it provides anyone with some new information that they didn't know before then be happy. It's not designed to be something that you would run 24/7.

I wish there was a more game friendly way to do this and the temp display was rock solid but that's likely not going to happen. Thanks for doing some testing. Maybe some game developer will see this and decide that including a feature like on screen temps would be a great option.

MrHydes: Come on. You have to give me a little more information than that. I don't know anything about your computer or whether you live at the North Pole or on the equator. Have you read anything about RealTemp yet like the documentation page? Are there any specific concerns you have? I enjoy helping users with specific problems and concerns but I'm not psychic.

What sort of cooling, how much core voltage, what about load temps while running Prime95 small FFTs on all cores.

The 2.58.1 also fixed the sensor test not working 100% correctly for me with the 2.58 and my 45nm x3360 C1. Once again you've outdone yourself unc

I hate to be honest but I don't know what I did that effected the Test Sensor feature. I'm kind of a random programmer and quickly forget what I changed yesterday or even last hour. Thanks loonym for continuing to be one of my biggest fans!

[loonym]

I had sensor movement of 1 or perhaps 2 on cores 2 and 3 on x3360 with the 1.58 now with 1.58.1 they are in the 7-8 movement range. It's hard not to be a fan of the kind of common sense, work ethic, and intuition you have shown on this project. When I first saw it I was like yeah ok, another fly by night quicky app but you have surpassed my wildest expectations in developing what I feel is shaping up to be the best intel temp program. Now can you get MBM5 working?

[Ace-a-Rue]

i ditto loonym's comment on your work Uncle!

[Fritzz]

Tried out Gamer Mode in both CoD4 and Half Life 2 Episode 2. I get pretty much the same results in both. It seems like the game is refreshing on the same intervals as RT. It causes my screen to flicker no matter what my FPS are. From 30fps to 90fps it's the same flickering interval. With HL2 there's flickering also, but it's not uniform across the screen. I will flicker in bands across the screen.

I would ultimately like to see it work in conjunction with RivaTuner, or EVGA Precision. They both use a server type program to give you real time readouts while you are playing games. Since I think there are quite a few people out there that use these programs already it would be great to see them integrated together. RivaTuner has a plugin out already that will read the temps of your cpu, but it's no accurate at all. I have no idea how this would be accomplished, but it's something I will put on my wish list.

Keep up the good work Unclewebb, RealTemp is one awesome program, and the following is only going to get bigger.

[unclewebb]

Thanks for the continued encouragement and suggestions. Gamer Mode was my first stab in the dark at trying to get some temps on screen while gaming. It seemed like an interesting idea. I haven't used RivaTuner for a while so I'll have to check out the latest version of it. I'll definitely be looking into other ways to get some stable temps on screen while shooting the bad guys.

work ethic

loonym: Want to drop by and tell my wife about my "work ethic?" She keeps telling me how lazy I am cause all I ever do all day is work on RealTemp.

[ChrisZ]

It causes my screen to flicker no matter what my FPS are. From 30fps to 90fps it's the same flickering interval. With HL2 there's flickering also, but it's not uniform across the screen. I will flicker in bands across the screen.

PMFJI, but adjusting frame rate of RT won't help if the game dynamically changes FPS depending on its load at any given moment. The good news is that frame rate of any game can be *made* constant (i.e. clamped to a fixed ceiling) by forcing the video driver to respect the vertical refresh rate of your screen. IOW, you must do it in the driver settings dialog box first where you set it to override the corresponding game setting.

Ideally, the best technique would be by injecting the RT output directly into video card's frame buffer just before it is switched for display by driver. Unfortunately it would require hooking directly to Direct3D engine (or OpenGL, if your game prefere that) and may open another can of worms by itself.

Never the less, when posting test results, Unclewebb will likely be helped more if you specify whether your driver is configured to synch with vertical refresh (vsync) or not.

All the best for Uncle! I know what it takes to *maintain* the thing.

[W1zzard]

the correct way to implement OSD in games is to hook the directx functions of games and when the game is done with all its drawing calls and calls present() you draw your own stuff on top of that. the best performance and safest way to implement this is as a vector graphic for the numbers so you dont need textures or anything else.

the fastest way to get an OSD in your game is by feeding data to rivatuner's statistics server.

[Brama]

the fastest way to get an OSD in your game is by feeding data to rivatuner's statistics server.

I completely agree as most of gamers using Nvidia use RT and many ATI/AMD guys too.

[unclewebb]

Three cheers for the W1zzard. He's the guy that made the temps show up in the System Tray.

I don't know how I'll thank him if he can show me how to get the temps showing up properly in games.

[clokker]

Quick question, Unk.
I finally went through the calibration procedure yesterday.
Using a (admittedly cheap) laser thermometer, I took several reads of my case exterior to come up with an ambient number and then corrected the idle temps (running at 233 x 6 at .975v) to get ambient + 7°c displayed in RealTemp.
Resetting back to normal speed/voltage, my idle temps are a pretty consistent 10°c lower than those read by CoreTemp...which seems about normal from what I've seen.

BUT...under stress testing the disparity becomes greater, closer to 15°c as shown here...


Obviously, no matter which reading one believes, I have nothing to worry about but I'd like to know if I screwed up the calibration procedure or whether OCCT is just plain wrong.
Any suggestions?

[Brama]

Obviously, no matter which reading one believes, I have nothing to worry about but I'd like to know if I screwed up the calibration procedure or whether OCCT is just plain wrong.
Any suggestions?

OCCT is displaying 10-12° more than RealTemp in my system.

I think RealTemp is the right one.

[clokker]

I think RealTemp is the right one.

Oh, I have no doubt about that.
I'm wondering why the difference between the two grows larger under load...

[WoZZeR999]

RT does not use linear calculations. 1c DTS is not 1c temp difference.

[JVM]

I haven't calibrated Real Temp and don't know that I have to. From what I read, Real Temp gives 10C less than Core Temp--and that is what I get without calibration. I also get 5C less than what is shown in SpeedFan.

I have the Intel Core 2 Duo E8400 with Real Temp showing TJMax of 64-65 and depending on room temperature, my Real Temp readings can be 5-6C to 7-8C above room temperature, which is based on my heating/cooling thermostat reading.

I don't care if calibration would give me lower temps, because I am perfectly satisfied with what is shown uncalibrated, always under 35C at idle with room temperature as high as 25C.

So, the questions are, do I really need to calibrate Real Temp, and are my readings accurate?

[unclewebb]

clokker: You need to show me a couple of screen shots at idle and full load with RealTemp including the settings window opened up beside CoreTemp.

I believe that OCCT offers you the option to select from a multitude of sources for temperature data including it's own internal source. I'm not sure which sensor or program it is reading data from in your screen shot.

With no idle calibration, the difference in displayed temperatures between CoreTemp and RealTemp should be exactly equal to the difference in TjMax that they are using. It's possible to change TjMax in most programs today so I need to see some more data. By default, for the 45nm chips, RealTemp uses TjMax=95C while most other programs are using TjMax=105C.

Reported Temps = TjMax - DTS

It should be easy to see how changing TjMax directly effects your reported temperatures. The reading from the digital thermal sensors (DTS) should be consistent between RealTemp and CoreTemp.

JVM: If you are happy with your displayed temps then be happy. If you want to try to get another degree or two more accurate then read the documentation and try doing the calibration. From what you've said, your temps are probably pretty close as is. Temps aren't too important anyhow. As long as your computer is running stable then there's nothing to worry about.

[Richard Dower]

Uncle, i was talking to a guy who works in DELL. He's in the qualification and testing dept. and he swears by Real Temp...he's long since dumped Core Temp as a diagnostic tool.