Real Temp - New temp program for Intel Core processors

[unclewebb]

Thanks for that pic some_one. Now that the multiplier is correct, I'll have to have a second look at the BCLK MHz part of the code. I wouldn't be surprised if my program says WTF? when it sees that the multiplier is more than twice what the BCLK is.

The Core i has 8 bits set aside to represent the multiplier so that gives a theoretical range of 0 to 255. Whether the CPU or a bios would ever let you use that range is hard to say. Maybe they were planning for the future when they designed that. Let's hope.

[radaja]

is there a way to make realtemp work with a pentium 4 530?

[unclewebb]

There are no core temperature sensors on a Pentium 4 so there are no sensors for RealTemp to read.

[radaja]

i see.thank god i only need to use it till tomorrow.thanks for the quick answer

[athletex2]

Any new development on RealTemp GT? The GT executable seems to be at 3.4 instead of 3.52. Is there a more recent GT executable?

[unclewebb]

I have been working on other projects for the last 6 months so RealTemp GT has not been updated. Probably someday but no idea when.

[athletex2]

I have been working on other projects for the last 6 months so RealTemp GT has not been updated. Probably someday but no idea when.

Gotcha-- no sweat. I appreciate all the work you've done.

[icecpu]

help please, why latest realtemp show only 3 core ??
all other programs I use show all 6 cores.



Uploaded with ImageShack.us

[burebista]

Because you need RealTempGT for Gulftown.

[icecpu]

got it , thanks

[SiGfever]

What is the recommended "Set TJ Max" for a i7 920 D0?
Thanks.
SiG

[unclewebb]

Intel writes a number that they call Temperature Target into each core of each Core i CPU that they produce. Most monitoring software, including RealTemp, has decided to assume that this is the same as TJMax but actual TJMax might be slightly higher than this number. On these CPUs, TJMax and Temperature Target seem to be very close, especially on core 0. Intel does not document their procedure for setting this value so no one knows how accurate the value in this register is.

The default value for most i7-920 CPUs is usually 100 but I think actual TJMax can vary a little from core to core on the same CPU. The Core i7 X58 CPUs have very consistent sensors so there's no need for any sort of calibration. You might as well just use them as is.

[SiGfever]

Intel writes a number that they call Temperature Target into each core of each Core i CPU that they produce. Most monitoring software, including RealTemp, has decided to assume that this is the same as TJMax but actual TJMax might be slightly higher than this number. On these CPUs, TJMax and Temperature Target seem to be very close, especially on core 0. Intel does not document their procedure for setting this value so no one knows how accurate the value in this register is.

The default value for most i7-920 CPUs is usually 100 but I think actual TJMax can vary a little from core to core on the same CPU. The Core i7 X58 CPUs have very consistent sensors so there's no need for any sort of calibration. You might as well just use them as is.

It seems to read what I think is most true for a well watercooled system if I set it to 95.

[unclewebb]

Intel says that their "temperature target" number is the minimum temperature so actual TJMax can be slightly higher than 100C but it won't be lower. My best guess is that actual TJMax varies between 100C and 105C on many of the Core i7-900 series CPUs with core 0 being close to 100C, core 3 being close to 105C and the two center cores being somewhere in the middle. My theory is that Intel might deliberately offset TJMax slightly so all 4 cores don't reach the thermal throttling point at the exact same time. By offsetting TJMax, you get a more gradual decrease in computing power so it's less noticeable to a user.

The sensors that Intel uses are far from perfect so it's certainly possible that your sensors read a few degrees high. If you bought 10 different i7-920 D0 CPUs and mounted them one after another on the same board, it's not likely they are all going to read exactly the same. There will be some variation but Intel has never said how much variation should be expected.

The reported temperature coming from a batch of Core i7-920 CPUs would be very close together compared to the previous Core 2 45nm generation. Those crappy sensors were all over the place from one CPU to the next.

Send me a box of Core i7 CPUs so I can start testing this theory.

[Wishmaker]

Hey Uncle,

I found something interesting when using real temp on my older laptop .

[burebista]

Try ThrottleStop for your laptop.

[Ajhayter]

Not entirely sure what this shows. Looks to me in graph form that the decay is fairly linear across all cores, but there is a few degrees disparity between core 0 and core 3. Is this of any significance?

Cooler is Noctua C12P in a LL PC-P50 case, with all fans on minimum (Zalman controller). Temps seem about right based on web reviews.

Thumbnails, click for biggies

[unclewebb]

Wishmaker: Try adding this to the RealTemp.ini configuration file.

MSRMulti=1

That might fix the problem you are having. Your older CPU might not have the internal timers that RealTemp uses. You can also check off the TM Load option in the Settings window for a load meter that should be similar to the task manager load meter. At least the temperatures and bus speed look about right.

RealTemp 3.59.3
http://www.fileden.com/files/2008/3/...alTempBeta.zip

Ajhayter: I don't believe that TJMax is set equally across all 4 cores of a Core i7-920. I think the difference you are seeing in your graph is completely normal and it's definitely nothing to be concerned about. My best guess is that if TJMax=100C for core 0 then most of these CPUs have core 3 set to about 105C with the other two cores somewhere in the middle.

[Wishmaker]

Hey uncle,


1 hour after I reported that issue, I heard something snaping in that laptop. I opened it and I saw that one part of the sinking was off the GPU!!! Funny thing from Fujistu not to add 4 screws around the GPU part of the sink. I know this sounds weird but the cooling in the laptop has 4 screws for the CPU plate and that is it. Imagine something like this : the gpu was glued somehow to the sink and when the glue lost coehsion the sink lifted up. In my Asus laptop I have 4 screws for the CPU and 4 screws for the GPU. So yeah, the GPU burned out on that thing .

[TheScavenger]

Talk about a dumb place to cut corners...

[stasio]

i7 Turbo GT Ver. 1.1

http://www.mediafire.com/?w5lpqqpy6wwq8wi

RealTemp 3.59.4

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

-added a font picker for the system tray / notification area

[CryptiK]

With that latest version stasio just linked, the multi of my Ci7 920 is jumping around from 14 to 19.0 (its set to 19).

UncleWebb did you ever get any closer to being able to read a Ci7 processor VID? As discussed awhile back, it seems gigabyte mobos can read it (as must any mobo really to set the appropriate vcore) so it stands to reason it must be stored somewhere in one of the registers.

[The Viper]

does anyone know if the TJmax are different between the i7 930 and the Xeon W3530?

[WoodButcher]

Geez Uncle, who'da thunk it, 168 pages, a freakin novella, You makin a living off this yet?
I need your help. I set up real temp way back when and that file has been moved or deleted but the startup still runs. Where in win7 is this command so I can clean it up?

[werdwerdus]

look in the task scheduler?