E4300 + Ultra 120 Extreme

[Coffee29]

Cornelious:

I was aware of Uncle Web's test and it prima fascia seemed like sound advice. However, after I read the intel forums and the responses from intel techs which can be found here---> http://softwarecommunity.intel.com/i.../30228638.aspx

I realized that the test is pointless because temps can never be known and mean nothing. Intel does not set desktop c2d's with tjunctions. The individually calibrate each c2d by a tjmax which is represented as 0 on the DTS reading. When this point is reached the TCC trip activates and shuts down the chip.

Because there is no documentation at this time concerning temp readings on the desktop C2D the temps can not be known. Applying the tjunction formula temp=tjunction-DTS, which is what coretemp does, can not work for a chip with no set tjunction. Even if there was a tjunction, and intel has said that there is not, without documentation you have no way of knowing if it would be one that scaled with temp, ie. changed depending on load. Thus no temperature can be determined on these chips. The uncle web test proves absolutely nothing. The only accurate thing you can do is read DTS numbers, higher is better, because they are 99% accurate and they are the only thing being reported by the chip.

I suggest you read the discussion that took place over this on this website, here is the address---> http://www.xtremesystems.org/forums/...00+temperature

My findings and then a google on them is how I came to these forums in the first place.

Embrace DTS.

[Cornelious0_0]

Interesting, I'll look into it.

[Ad1tya]

Hmm, Quite right.

Actually, Coretemp 0.95 Crashes on my system, but the first time I tried it, it was on for about 20 seconds, and I did notice 15C higher Load temps and a 100C TJunction.

I have no clue what to believe though. Everest, Speedfan, TAT and CoreTemp 0.94 all reported my temps to be 45/47C Loaded, but Core Temp 0.95 reported 60/62C Load. I just thought that Its a core temp 0.95 bug.

Hmm, How the hell are we supposed to tell temperatures now :|.

[david256]

I don't understand all your high temps at the moment my E6600 @ stock is idling @ 38C.

Overclocked to 3.2 1.35v idles @ 45C and under load @ 53-55C (playing games etc) and never goes above 60C. Fan speeds between 800-1200 rpm (never full speed 1600 rpm). This with my Ultra120, no lapping or removal of IHS.

[Coffee29]

The research necessary will take a bit as there is a lot of documentation to look through, but it boils down to the this:
1. You can't tell temperatures on desktop C2D's now nor ever. Only A DTS signal is sent from the core and these programs, including speedfan and coretemp 94 are using the following formula temp=tjunction-DTS. The ones you are using assume a tjunction of 85 and the 95 that crashes and TAT use a tjunction of 100. But these tjunctions, according to INTEL themselves ONLY apply to mobiles.
So you might ask why did coretemp and the rest assume tjunctions when they know they are invalid? Surely they know?
The answer is that they do know. However, if they did not do this they could not give you temperatures (who care if the temps are wrong) there is no way to give temps so they use the mobile formula and simply guess at a number 85 or 100. This is scary because not only do you not know which number your comp would be at, there is no reason to believe that it is either! Or that it does not scale with load! And now that INTEL has said that this number does not even exist for desktops you are basically using these programs and getting false temps on ALL of them. None of them are right because they all use a formula that doesn't apply to your chip with a number that doesn't exist in reality for your chip...
2. This is not necessarily a bad thing. I have read that on previous processors temp programs were notoriously off the mark. The advent of DTS is a great thing because that DTS number is 99% accurate. These programs cannot and do not change it. The only variable is that stupid tjunction they are making up in the equation. So the answer is simple, take out the eqaution and only use the accurate DTS.
3. DTS works like this---> each processor is tested at the manufacturer and they find its limits. They assign a TCC or a trip point for your individual processor based upon your specific processors limits. The DTS reads how many degrees C you are from that TCC. For example, an idle DTS of 65-70 is what I get. As I increase load I will see it move to the 30's. You want to stay as far from 0 as possible, but many believe, (they do not know) that a DTS of 20 or above is ok. Time will tell.
So the good news is we have a more reliable method to tell the state of the chip, we only have to figure out its scale...ie is 20 DTS ok, is 10, ok...etc...The bad news is that you can never ever tell your absolute temp again. It is not knowable unless there is a hidden MSR somewhere (a hidden signal) that no one has been able to find and that INTEL swears does not exist.
The best plan right now imo is to just pay attention to your pc. IF its crashing a lot it may be too hot. Use your fingers, is the ram to hot to touch for less than 10 seconds? thats too hot. Is your northbridge doing alright? USe DTS, if you can get coretemp 95 to open, mine opens 50% of the time, you can switch it in the settings to report this DTS instead of the temperature guess. Unfortunately there is no other program right now that I know of that will tell you this DTS other than coretemp 95.

I understand that all of this is new, and I don't claim to be an expert. But the documentation from intel is right there and everything I have said here is in black ink. Therefore, if anyone wishes to correct me I encourage it, but please do not do so with he said she said, so and so has a theory....Give documentation so we all know its a fact.

[Coffee29]

Please everyone check out those two links I posted a few posts back. One of them is to the intel forums and you will find what I stated here, stated there by an Intel tech...

[Leeghoofd]

I don't understand all your high temps at the moment my E6600 @ stock is idling @ 38C.

Overclocked to 3.2 1.35v idles @ 45C and under load @ 53-55C (playing games etc) and never goes above 60C. Fan speeds between 800-1200 rpm (never full speed 1600 rpm). This with my Ultra120, no lapping or removal of IHS.

Indeed it is known if we may believe the temps red out that these cheap core2's are really hot stuff, they overclock well but generate laods of heat too... I would have opted for an E4300 to get a q6600 later on but I think I will opt for an E6600 and go for 3200mhz with much lower temps then these....

[Coffee29]

Leeg:

The E4300 are almost the same exact chip as the E6600's. There are only two major differences of which I am aware. The E6600 uses solder between the IHS and the core which would transfer heat better than the E4300 because the E4300 uses grease. Second the E6600 has more cache...which actually may cause more heat.

Most people I have talked to state that the two chips put off very similar heat. Please correct me if you knwo something that I do not.

Either chip, E4300 and E6600, as well as all other desktop c2d's suffer from the same problem of not being able to have the temp acurately taken. They ALL use DTS. Because the archetecture is virtually the same they should theoretically put off similar heat. But because we cannot know what the temps on any of them are its hard to say.

I recommend lots of research on it before you make your choice because if I am right that the E4300 puts off the same heat, you can essentially have an E6600 (minus a little cache) by purchasing the E4300 and OCing it.

[Leeghoofd]

thx for the advice but I could get a sweet deal on an E6600 29B so i got for that one, guy stepping up to a Q6600... thx for sharing the info though...no matter how the temps are read out I feel more at ease when it gives me 50°C laod then 70°C...

[Coffee29]

I hear ya! I like 50 better too. With Coretemp 95 that would translate into the following:

Coretemp 95/TAT (100 tjunction assumed)
70C = 30 DTS (right on the margin of what some people say is ok)
50C = 50 DTS WAY within spec of the chip

Coretemp 94/Speedfan (85 tjunction assumed)
70C = 15 DTS (pushing it)
50C = 35 DTS (within spec)

the goal is to stay above 20-30 DTS so either way you look at it, 50C on either program is within a good range so its something to shoot for.

[//\\//3W]

Good job Coffe29.
I was looking over my temps using SpeedFan, EasyTune, TAT and CoreTemp.
SpeedFan and EasyTune say that my temp are 20°c but TAT and CT say that they are 10°c above SF and ET.
This is really annoying.

So the best way to be sure is use our fingers on HS.

[Ad1tya]

Leeghoofd, I hope you dont get screwed :P.

Coz My damn L629B953, is shady.
3.6 Ghz @ 1.47vCore. 3.8 Ghz @ 1.51vCore and 3.9Ghz @ 1.53vCore.
4Ghz not possible even at 1.57vCore.

Complete utter bull. My L626A on the other hand was amazing :P. 3.6 Ghz @ 1.42volts! I cant believe I sold it for the damn L629B crap!!

[Battle_Rattle]

Taco.... have you said what your ambient temps are?

Cause that seems hot.... Lap IHS and HS and be sparing on the TIM

[Leeghoofd]

My current 32a does 3.6 too at 1.42 volts, 3.8 at 1.5 volts I have the choice between 2 new ones (I have to assemble 2 new rigs) and the second hand one... must be one good one in this batches that can do about the same, though I'm goin for a modest overclock as this pc will be my 2nd rig at my folks place...

[adamsleath]

i have an e4300...seemd great...now its bsod central

i wonder whether ive overvolted it to death?

now im on 1.58v just to keep the thing stable
ambient temps are very low at the moment - 17-20c and coretemp is showing between 30-35c idle and below 60 during orthos.

so i dont believe i have overheated the chip...nevertheless; i was at 3.4ghz evrything was fine for about 3 days @ 1.46v...then i had a bsod
...now i've been up and down with voltages and speeds....even stock speed is troublesome...

i have the impression that i have damaged the chip with too many volts.

it crashes very quickly at lower volts - say 1.25-1.35vcore

like i said im at 1.58v at stock clocks just to keep it stable atm

something has definitely changed.

ram seems fine at 2.15v...i'm 99% certain it's the chip.

[Khapz]

my results testing 1.2v in bios mult 6x
room temp = 25~30ºc

SS on full load



coretemp .94 are correct?