Real Temp - New temp program for Intel Core processors

[Slay0r]

Well, things to consider: Intel told me, " we don't care how much hotter X core is than Y core, all we are concerned with is if the chip exceeds 70 degrees celcius. " Further, these temps were after I just powered on after being off all night. Windows had just loaded so everything was at less than 5% load on the cpu. Also, what is weird, is realtemp tells me the temps are fine but my 790i BIOS shows it at 50c. I've been told the CPU is slightly loaded in BIOS, though.

The only theory I think is valid and hasn't been disproven is that maybe one set of cores isn't receiving as much contact as the others. I'm talking internally, perhaps, when the CPU is made there's a percentage of offset on how the cores sit. Or, the sensors could just be improperly aligned. IDK, intel only gave me that response and they wouldn't say much more. =/

Bios loads the first core @ 100% on every processor/mobo, this explains the temperature higher than idle.

The discrepances in temps are likely caused by a concave IHS/bad mounting/bad sensors. The latter seems to be very common on 45nm intel processors (duals and quads), and some of the late 65nm quads. You won't get intel to admit that, though.

[varrius]

Bios loads the first core @ 100% on every processor/mobo, this explains the temperature higher than idle.

The discrepances in temps are likely caused by a concave IHS/bad mounting/bad sensors. The latter seems to be very common on 45nm intel processors (duals and quads), and some of the late 65nm quads. You won't get intel to admit that, though.

As long as the temps don't exceed the 65, I'm happy. Remounting 6 times didn't seem to change it. So, maybe a 790i BIOS update will fix things, who knows.

[Retrolock]

Thanks for the update unclewebb! Now I can hide it at my system tray!

[varrius]

CoreTemp added G15 support. Is there any way we can get G15 support? It would be really awesome if I could check my temps while in game. =)

[mrcape]

woohoo @ minimize to taskbar

Sweet program.

[ChrisZ]

Unclewebb,

It is possible to start several instances of RealTemp.
Please check your PM, -- I posted there a function which can prevent this happen.

Best!

[unclewebb]

ChrisZ: One man's bug is another man's feature. Did you ever think that some crazy people might want to have multiple instances of RealTemp running at the same time to do some XS Benching? On a Quad you can run 4 instances simultaneously with no slow down. Thanks for the code you sent. I've always wanted to know how to do that and I'll probably make it an INI selectable option in the future.



Just finished the last major feature this evening before the next official update. RealTemp MHz and CPU-Z MHz have a lot in common while the competition is still a little behind. RealTemp also supports the 0.5 multis in the new 45nm series, at least in theory. A screen shot of an E8500 or any other 0.5 multi CPU while running RealTemp and CPU-Z would be great.

Beta testers can add the following file to their RealTemp directory.
http://www.fileden.com/files/2008/3/...alTempBeta.zip

[unclewebb]

i43: Here's some testing you will find interesting:

The first picture is of my Q6600. It's at idle but doesn't have a heatsink on it so the temperatures are getting up there.



I think looking at it you would have to agree that TjMax must be the same for both sets of cores and all 4 temperature sensors seem to be working properly and displaying similar temperatures. A difference of a couple of degrees is pretty minor.

Picture #2 is of the exact same processor. This time it has a heatsink on it but I now have it at full load runnning Prime on all cores and I've turned off the fan to let the temperatures get up there again. It's now showing the same as your CPU, a perfect 9C difference between the two sets of cores.



When you see someone post the second pic the first thing you think is, "Dumb ass doesn't know how to mount a heatsink." I thought so too until I loosened up my heatsink, turn by turn, until it was less than baby finger tight and yet there was still this difference between the sets of cores. It just goes under the category of s-h-i-t happens when you own a Quad.

When I turned the CPU fan back on in picture two the difference between sets of cores dropped to the more typical 5C. Weird, hard to fully explain but completely normal. It seems to be a heat transfer issue with that second set of cores.

Edit: In this case, it is the second set of cores that are doing a better job of transferring heat away from the core temperature sensors.

[Fritzz]

Looking good once again Unclewebb. I have been trying to spread the word. I have probably put the link to this thread in almost all of my posts in reply to the temp issues people are seeing.

Here's my SS:

[kadir_slayer]

are these temps safe??

[Richard Dower]

2.49 reports the wrong CPU speed...my Q9450 is set at 425x8=3.4GHz, but the program is assuming an 8.5x multi and 3.5GHz?

[unclewebb]

Thanks Richard. I'll have to re-install my E8400 this weekend for some 0.5 multiplier testing.

kadir_slayer: I think over 70C should be avoided while running Orthos. In my own experience, I find that with a brand new processor I can run Orthos stable for an extended period of time at a high temperature like that but soon afterwards, I won't be able to repeat it. I'll need better cooling or less MHz to run the same MHz reliably.

[Milamber]

Anychance we can have realtemp ID the CPU with a logo for that CPU type like CPUz That would look great - vcore detection would also be wicked. temps, vcore and logo

[unclewebb]

Milamber: If you want core voltage and a nice logo how about running CPU-Z side by side with RealTemp? Just a thought.

I thought I'd just add the Core Speed into RealTemp so users would have some useful information to look at inside that box. Once you know what TjMax RealTemp is using, there's not much use in looking at TjMax 24/7.

I think I would need Intel approval before plastering their logos over my program and I don't imagine they like my Test Sensors feature too much and they certainly don't approve of any software that tries to get some meaning out of their top secret digital thermal sensors. They call that reverse engineering and frown.

When using RealTemp, I already know I have an Intel core processor inside or else the program wouldn't be working.

Why doesn't someone do me a favor and post a RealTemp vs CPU-Z screen shot showing the MHz. Boot up with a less than full multiplier as well as with a less than full half multiplier. Your E8500 would be perfect Milamber. I want to see where I screwed up. I might have misread the top secret Intel docs.

Edit: If you use SetFSB or ClockGen while RealTemp is running, you will need to do a quick recalibration of the RealTemp MHz feature. Just right click once anywhere in the main RealTemp dialog screen.

[loonym]

Thanks once again for outstanding work

[kadir_slayer]

where is the link of 2.49version???

[ChrisZ]

where is the link of 2.49version???

As always in the same place:

http://www.fileden.com/files/2008/3/...alTempBeta.zip

[Milamber]

Why doesn't someone do me a favor and post a RealTemp vs CPU-Z screen shot showing the MHz. Boot up with a less than full multiplier as well as with a less than full half multiplier. Your E8500 would be perfect Milamber. I want to see where I screwed up. I might have misread the top secret Intel docs.

Is this what you were after uncle?





Ambient temps were 26 degrees.

[unclewebb]

Is this what you were after uncle?

That's exactly what I needed to see. It confirms that RealTemp is reading the 0.5 multipliers in the 45nm series correctly.
Thanks for your help.

[chris.y2k.r1]

Welcome to the club.

This seems to be a common issue with many Quad core processors and has absolutely nothing to do with how you mounted your heatsink or block. It effects both 65nm and obviously your new 45nm cpu. This issue was brought up on page 29 of the RealTemp novel and discussed, mostly only by me.

http://www.xtremesystems.org/forums/...&postcount=722

It might be a heat transfer issue where both sets of dual cores within a Quad do not transfer heat equally and it also might be a sensor issue. On my Q6600, core0 and core1 as reported by RealTemp were in agreement with what the IR thermometer was showing. Try the test I came up with in post#722 and post your results.

I"ve found this to be worse with XP and almost non-existant with Vista. I believe it's simply work load related. The same thing happens with DC in XP.

UncleWebb, I am humbled and amazed at knowledge, ability to deduce and focus and your programming skills. Thank you, Thank you!

[Milamber]

That's exactly what I needed to see. It confirms that RealTemp is reading the 0.5 multipliers in the 45nm series correctly.
Thanks for your help.

Cool - glad to have been of some help. When you have your website up and running uncle you will need to make a small link on the realtemp app to it, or somehow have the app check for updates upon launch like GPUZ does --- not needed for now tho.

[jcool]

Hey unclewebb,

I have a rig here with Asus 790i mobo and a QX9770, but I don't think temps are being read correctly. CPU is watercooled by an Apogee GTX, the mounting is good (double-checked it), water temp is below 30C and yet it shows temps above 70C for all cores (setting is 4Ghz@1,34V).
Do I have a chip with crappy readouts or does Realtemp maybe use the wrong tjunction?

[bigslappy]

I've got Vista64 & there is a 5 degree diffrence between the two r/t & C/T which is the correct one ? sorry too little time to read all 35 pages , I got a new MoBo to test today

[unclewebb]

water temp is below 30C and yet it shows temps above 70C for all cores (setting is 4Ghz@1,34V).

Is that 70C at idle or full load? It sounds like your QX has sensors that are not functioning properly which has been reported by many users. Stuck and misreporting sensors is a very common problem since the transition to 45nm by Intel.

Try CoreTemp as well to try and confirm. Both programs should report the exact same Distance to TjMax, at all times, full load or idle. You will need to go into the settings in CoreTemp and select, "Show Delta to Tjunction max temp." Post some screen shots so I have some data to look at.

chris.y2k.r1: I've found this to be worse with XP and almost non-existent with Vista. I believe it's simply work load related. The same thing happens with DC in XP.

I originally thought the temp differences within a Quad was an operating system issue but when I discovered that the balancing appears to be directly related to the data coming from the digital thermal sensors, I concluded that it must be a hardware issue. It would be very interesting for someone with a dual boot system to run Prime95 small FFTs, first on XP and then on Vista, to see how one's choice of OS effects temperature balance within their CPU.

I've got Vista64 & there is a 5 degree difference between the two r/t & C/T which is the correct one ? sorry too little time to read all 35 pages

If you don't have time to read what RealTemp is all about then you're better off using CoreTemp.

UncleWebb, I am humbled and amazed at knowledge, ability to deduce and focus and your programming skills. Thank you, Thank you!

A quote like that needs to be repeated! Thanks for the thumbs up.
It just might end up on my eventual website, near the top!

[bigslappy]

didn't mean to offend you UncleWeb I read the 1st 10 pages & understood most of your work but couldn't find out if it worked on Vista 64 ,I like your program as it is running fine on my 64 sys

i'm getting 33,33,32,35 in core temp right now & with your program 28,28,27,30 sooo maybe just idle about 2 pages of IE open right now & both programs open

your programstest >
test went core 0 = 10
core 1 = 9
core 2 = 9
core 3 = 9

all this on rig#1 in my signature

agin sorry if i offended you ,

i have alot of PC projects going all at once here and the weekend is my only time to do it .As I drive all week 60 hrs plus & was just asking if someone could save a guy some time with a quick post about vista64 & your program ..