SourceQuote:
Originally Posted by The Coolest
Printable View
SourceQuote:
Originally Posted by The Coolest
:woot: now we can see how good our watercooling really is :woot:Quote:
Originally Posted by The Coolest
Very nice!
Good to see Intel support 3rd party software.
And happy birthday to you!
:party::cheer::bday::bounces::bday2:
Great news, I'm glad Intel won't try to hide it anymore.
This is Great News !
I bet intel wont reveal the true TJmax and say a number that is like 10C lower than the real TJmax.
Cool! I'm interested in seeing how my processor is doing.
*btw haha at your avatar BuBBle.D
So... you know some thing we dont, or you just being pessimistic:down:Quote:
Originally Posted by BuBBle.D
and tjmax effects us how ?
nvm im an idiot, some reason i was drawing a blank on tjmax was ... derrr O_o
thermal probes work on a non linear voltage scale of how far its lower than the thermal junction, so if we know the exact specs its much easier to make proper temp monitors
right now all temp monitoring programs use TJmax settings to refference the temp of the proc, the problem is that we dont know what the real TJ max is for the current processors and thus we cannot say for 100% certain that a temp from coretemp or realtemp is correct, we are just making educated guessesQuote:
Originally Posted by Lestat
i think it has to do with clothes and such :p:Quote:
Originally Posted by Lestat
:owned:Quote:
Originally Posted by Spawne32
We have its sister TKMax here. ;)
Incoming dumb question: What was all the talk about the recently newly adjusted TjMax then? My E8400 now used 100 instead of 105 in CoreTemp, as I'm sure you all know.
Yes, indeed.Quote:
Originally Posted by xoqolatl
From what I gathered from the last presentation about the Nehalem is that the TjMax value is going to be programmed into the CPU's IA32_TEMPERATURE_TARGET MSR for accuracy's and ease of use sake.
Why would they do that? There's no benefit for them in lying.Quote:
Originally Posted by BuBBle.D
We only got values for the Core 2 45nm range, this time the information will also cover Xeons and 65nm parts.Quote:
Originally Posted by inCore
Ah right, suppose it's no wonder that I didn't notice. Thanks for clearing that up.
No i dont know sh*t. Im just pulling pessimistic conclusions outta my pocket.Quote:
Originally Posted by billdavis
Like it or not, thats the way i am (sometimes).
Sure there is. People would panic and lower their 24/7 OC according to the temperature. Lower OC = less RMA even though there are only a few dead c2d cpus.Quote:
Originally Posted by The Coolest
Intel's not going to lie.
They might change their minds and say, "nope, not giving it up!"
But they're not going to just blatantly lie, that doesn't even make sense. I'm a pessimist to the Nth degree too, but that's just silly.
October 20 – 21, 2008.. oh dear. They don't need an idf to release this inforrmation.
Well i sure hope they wont lie. We wouldnt benefit from it.
:party::rofl::ROTF:Quote:
Originally Posted by Spawne32
This is great news, I'd love to have guess-free temperature readings.
I assume they wont lie, because the whole purpose of releasing the info is to give more accurate details, not less, but then you never know, just look at how they lie with tdp's.Quote:
Originally Posted by BuBBle.D
so QPI is fundamentally diffrent from HTT isnt a lieQuote:
Originally Posted by Sly Fox
the i7 is faster in single threaded apps
IMC is bad, IMC is good
and there are more not from the last year or so
Intel doesn't support OC anyway, you can't RMA once you've overclocked....Quote:
Originally Posted by BuBBle.D
No company does, atleast not officially, however the effort to prove it is high and they dont want to loose customers, so they often give an rma anyway and as you can see on xs many customers exploit that to the full extent.Quote:
Originally Posted by SNiiPE_DoGG
It's fairly difficult to kill a chip with overclocking (at least it used to be until recently), and we're such a miniscule market, that intel doesn't really care about allocating its RMA funds to a handful of xs members who want to abuse it. It's better than having the negative PR of unhappy customers. A lot of xs members have a bit of an influence, too.
and intel likes when we lap or use extreme cooling since it changes the appearance too, they also like that we now shun AMD when for all intensive purposes if u cant OC they are better for the price (flame suit). in all they just want a dead/almost EOL line to be kept alive so no1 starts complaining when there are no more large cash chips in 6 monthsQuote:
Originally Posted by iddqd
there also may be heat problems with the voltages in the i7 so they dont want u overheating it with bad temp monitoring
I'm not sure this is true, at least here in Austria. It seems they are close price/performance wise, so maybe you were right and your comment is indeed inflammatory.Quote:
Originally Posted by zanzabar
Supporting Fugger at events....are you sure Intel doesn't support OC'ing? :shocked:Quote:
Originally Posted by SNiiPE_DoGG
http://documents.irevues.inist.fr/bi...79/1/TMI23.pdfQuote:
Originally Posted by Lestat
Tjmax is used to calculate the actual temperature from the DTS, it is an offset from a 7-bit word from Tjmax. This is accurate for mobile Core Duo and Core 2 Duo chips, since they are designed without an IHS and the Tjmax is specified for those die...
People assumed, since they are the same architecture, same die, that the mobile Tjmax is the same for desktop chips... this may or may not be true. So it will be nice to have the official Tjmax in hand to allow temp sensing programs to report accurate core temps without second guessing the calibration.
Here is a nice explanation of how thermal diodes work: http://www.maxim-ic.com/appnotes.cfm?an_pk=689.
yeah and extreme cpu versions with open multy are for what :)Quote:
Originally Posted by SNiiPE_DoGG
Earn tons of cash?
Okay, surely something can't be right with the TjMax being 105*c for the 45nm E7200 - i've got ORTHOS fired up here at 4Ghz @ 1.4375v and it's reporting load temps of 86*c!!! This is with a Tuniq Tower 120, AS5 (Yes its properly seated) and the fans in my P182 on full - my laser thermometer reports the heatsink base at 38*c - what's going on?!
well when are we going to get it?
Awesome!
Now we can get some true temperature readings.
(Not sure why they didn't do this at launch though...)
*grumbles*
you have to manually set your tjmax to 95 in either coretemp or realtemp the newest core temp reads more properly than realtemp at least imoQuote:
Originally Posted by Kai Robinson
95? hang on, i thought it was 105*c?
I think its 100c for the 45nm processors.
yes this is the current "estimated value" but we will all know for sure soon !:)Quote:
Originally Posted by BuBBle.D
intel already gave aways the Tjmax for the 45nm procs, so its no estimation ;) :Quote:
Originally Posted by SNiiPE_DoGG
Quote:
45nm Desktop Dual-Core Processors
•Intel®Core™2 Duo processor E8000 and E7000 series 100°C
45 nm Desktop Quad-Core Processors
•Intel®Core™2 Quad processor Q9000 and Q8000 series 100°C
•Intel®Core™2 Extreme processor QX9650 95°C
•Intel®Core™2 Extreme processor QX9770 85°
Odd, I wonder what the q9650 really is, considering the QX is 95...but the Q 100C.Quote:
Originally Posted by Hornet331
Is that Tj or Tcase ?Quote:
Originally Posted by Hornet331
It was Tj, it was part of a intel presentation, maybe i can find it again.Quote:
Originally Posted by JumpingJack
edit:
found mirror of the pp:
http://denn1s82.de.funpic.de/dennis/...TS001_100r.pdf
slide 13.
It still doesn't sound right :/
well its information directly form intel, i think they know there stuff. :p:Quote:
Originally Posted by Kai Robinson
Not doubting that - but my thermocouple and my laser thermometer don't lie...seems like a stupidly large temperature gap.
you cant measure Tjunction, what your a measuring is Tcase. ;)Quote:
Originally Posted by Kai Robinson
IR will read shiny surfaces artificially low because of reflectivity. Also you are measuring the heatsink which is lower temp than tcase which is lower temp than tjunction.
If you put a piece of masking tape over the heatsink where aiming you can get accurate temp of heatsink...but on load there will be big gap.
But if you want to measure tcase, remove heatsink, undervolt and underclock (so does not overheat prior to reading), put thin piece masking tape on cpu (otherwise you will get artificially low temps) and put IR gun on top of cpu. That way you can calibrate your cpu temp to actual tcase temp if you desire (speedfan will give you cpu temp) and calibrate core (tjunction) temp ~5C higher than that.
To check accuracy of your sensors with IR:
In attached pic with tjmax of 95, tcase (IR) gives same temp as tjunction (software). Though we now know ~5C gradient exists at idle with no heatsink, underclocked state of measurement. So with correct tjmax 100, tcase would read 72 when tjunction read 77.
(reference articles for gradients are in realtemp thread)
Got an email from Benson, there was an error in the original presentation.
Official release will be made in a few days.
Here is the correct information:
65nm Desktop Dual-Core Processors Stepping: B2 G0
• Intel® Core™2 Duo processor E6000/E4000 series 80 90 °C
• Intel® Core™2 Extreme processor X6800 80 90 °C
65 nm Desktop Quad-Core Processors
• Intel® Core™2 Quad processor Q6000 series 90 90 °C
• Intel® Core™2 Extreme processor QX6000 series 90 90 °C
• Intel® Core™2 Extreme processor QX68XX 80 80 °C
He's still confirming information about the server parts.
I'm still not quite sure about the 80C for the QX68xx parts, but these values look better than the original ones anyway.
Great info, any word on the 9 series?
Thanks
So does that mean Real Temp is more reliable than CoreTemp? I have Q6600 and 90C seems kinda low but I trust it more around there.
i7 965 Extreme? Would really love to know that.
For the Extreme's they should have Green Range, Yellow Caution Range, and Red St00pId range. hehe :D
What about 45nm CPUs?
I thought 100C was officially confirmed for 45nmQuote:
Originally Posted by xoqolatl
45nm cpus were not the issue and should be same, it was the 65nm tjmaxes which made no sense as they suggested core temps 15C below IHS temps. These are a little better for 65nm....but it still does not explain 95C IHS temp on a E6850 GO at DTS=0 with tjmax 90C. And Q6600 B3 stepping measuring 85C IHS at DTS=0 with a tcase max of 62C versus Q6600 G0 measured 95C IHS at DTS=0 with tcase max of 71C, so no way am I buying they have the same tjmax. Though the Q6600 B3 90C tjmax makes sense based on E8400's tjmax of 100 and measuring at DTS=0 95C IHS temp and 97-98C top of die temp when drilled though IHS (both cases suggest a 5C higher core temp than IHS at idle, low volts, which is plausible, heat cant go from IHS to core as would be suggested by some of the other still too low 65nm tjmaxes).
If those are correct tjmaxes for 65nm, then unlike 45nm, many still have 5-10C absolute temps higher than tjmax from DTS sensors being "individually calibrated to avoid throttling below tcasemax". But the inconsistencies still seem fishy.
yeah... 90 on a G0 Q6600 sounds low...that would mean my TRUE is cooling it to ambient air, or actually slightly below, as opposed to about 9 degrees above ambient case temp
I hope CPUz, CoreTemp, etc. will all update their programs with this information because it can really and I mean really help the overclocking community
hmmmmm, I think that 90 may be wrong, my brother just changed his realtemp to 90C and it said his q6600 was idling at only 17C, it is at stock and under water at the moment but that is well below ambient.
Mine would also be under ambient . . . .. LOLOLOLOL.
I wonder what it really is? In any case, I think I'll be treating it as 95 for now.
TJmax of 90 gives me core temps of 27-30 on a true, with a case temp of approx 32 degrees at the bottom of the case *temp guesstimated with a kama meter temp probe, waiting on my IR sensor for a slightly more accurate reading*
Yeah my readings are in the low and mid 20sC idle and I am clocked to 3.6GHz on my Q6600. If that is true then SWEEEEEEEEEEETTTTTTTTTTT, but I doubt it is. Seems a bit to low.
Well this got me thinking. Aren't temp sensors designed to accurately measure load temps? For example my idle temps fluctuate like mad whereas my load temps don't move around. Sooooo could this also be a lack of precision of the temp sensor? Oh well.
what does the "80 90 °C mean"? one core is 80C and the other 90C? and also, the stepping says B2 and G0. i thought these were for the q6600? i have an e4300 with 2 L2 stepping, so this tjmax does not apply?Quote:
Originally Posted by The Coolest
So is 95c accurate for 45nm quads?
80C for B2, 90C for G0. I assume that the B2 Tj Max is actually the L2 Tj Max for the E4x00, they just didn't bother to label it properly. According to the slides there's also a B2 Q6600 but there isn't really. You need to just make some assumptions with the steppings.Quote:
Originally Posted by WhiteFireDragon
G0 and B3 65nm quads with the same TJMAX of 90C?
That is wrong for sure.
2 years passed now, and noone can know for sure the real temps of their 65nm quads. Nice.
100c is the temp officially for 45nm quad, 95 is the correct for 65nm quads (despite what was posted earlier in the thread)Quote:
Originally Posted by Utnorris
Don't get this they don't support OC and the first people that get's their CPU's [ES] don't buy a peace of hardware for more then a decade aka hard core benchers.Quote:
Originally Posted by SNiiPE_DoGG
I'm not sure about 100 for 45nms. If this is accurate (and presuming my temp sensors are ok), my Q9450 under an EK-Supreme and on an MCR320 is hovering at 60idle, and about 80 load. Feeling the back of the board during operation and around the base of the waterblock, there's no way those temps are correct.
[yoda] to hasty you are, a q9450 you have mhhhhm, broken DTS it has[/yoda] :D:D:DQuote:
Originally Posted by SoulsCollective
You probably have craptacular sensors then. :ROTF:Quote:
Originally Posted by SoulsCollective
Run the new sensor test in RealTemp and see if your sensors are sticking at ~60C (~40 delta to Tj Max). http://www.xtremesystems.org/forums/...postcount=2502
It does seem to be quite a high temp to be sticking at though.
So what is the TjMax for the Q6600? I'm going with 95 until someone confirms Intel's news of 90.
90 is very incorrect, 95 is the right TjQuote:
Originally Posted by MomijiTMO
So, I wanna know the CORRECT TjMAX for:
65nm QX6700 (B3 stepping)
Is it 90C? 95C? or 100C?
Damn intel....
and then have 10 million refunds to process from 3rd party companies like dell for example?Quote:
Originally Posted by BuBBle.D
no, that won't happen.
wow, i'm confused about why it's so hard for intel to just tell their customers the right tjmax. incomprehensible.
Perhaps because it's a "trade secret."Quote:
Originally Posted by RaZz!
because they never ment to be public. :p:Quote:
Originally Posted by RaZz!
Actually they did. They made the IDF presentations because of endless customer enquiries about temps. It doesn't mean they had to tell us everything. It doesn't mean that what they told us is true either. We just have to accept it, ignore it or "fight" for the real stuff.Quote:
Originally Posted by Hornet331
I dunno if they were "mistaken" about the 65nm TJmax then who's to say they are right about 45nm TJmax, because they are supposed to put out far less heat but even with 95 TJmax they put out just the same, and if a 65nm is 95-90 then wouldnt a 45nm be less? this just make me wonder.
Really, if your 45nm is pulling the same heat as a 65nm on same volts or clock, then either you have a bad cpu, bad step, bad sensors or bad cooling.
If you want to compare how hot they are running, you need to know that Tj Max for both is correct or is wrong by the same amount. Then you need to assume the same slope error, which you can't really :p: A "better" idea is to drill into the IHS like rge did and measure the temperature of the top of the die yourself.Quote:
Originally Posted by Simps
I think they are shady about the TJ max due to overclocking a lower priced cpu vs purchasing a higher clocked stocked cpu (though niche crowd) and losing profit.
I don't see how that would affect what they're telling us. A EE chip isn't going to run significantly cooler than a mainstream chip. Intel knows that few people are going to pay for an EE chip, that will never change. The funny thing is that the QX9650 and QX9770 actually have a 5-15C lower Tj Max than the mainstream chips, at least according to Intel.Quote:
Originally Posted by kadozer
Nothing funny there -> higher tjmax means they can use a cheaper cooling solution, they acutally mentioned that in there presentation. :yepp:Quote:
Originally Posted by randomizer
Does the QX9770 come with a more expensive cooler than the Q9550? They are the same stepping, but Tj Max is much different.Quote:
Originally Posted by Hornet331
The QX9770 and QX9775...at bottom of intel specs sheet "These processors ship without a fan heatsink. They require a third party liquid cooling solution." So definitely different cooling.
From IDF and from formula, Tjmax is dependent on power and cooling solution (also on ambient temps but they are same in intel testing). Q9650 is 95 watts and air cooled, QX9770 is 135W and water cooled.
I wanna know the correct TjMAX for: e6850 65nm
yep they do, the QX9770 uses the same cooler als the QX9650. And the cooler is hughe compared to the stock Qxxxx cooler. ;)Quote:
Originally Posted by randomizer
I know this is a bit of thread necromancy, but I've recently been shown this Tom's Hardware news article, which is stating all the TJ Max numbers for every Core2 (and Core2 based Xeon).
I want it to be right, I really do, but it's Tom's, and doesn't agree with everything else that's been posted here...
Eller
Well I can tell you that for the Q6600 they're wrong at least.Quote:
Originally Posted by karl_eller
I mean, I'd love it if that's correct, but that would put mine at sub-ambient sometimes, and I can't force myself to believe. :D
God damn you Intel, why can't you just publish the proper numbers already :(
Eller
I have my doubts they'll ever publish it.
so.. whats the tj max for an i7?
The article was correct about the numbers Intel posted originally, but hasn't been updated with the "fixed" numbers. You should pretty much ignore both as they aren't going to give you accurate temps.Quote:
Originally Posted by karl_eller
They have, but because Tj Max isn't a constant the numbers have little meaning other than a sedative for the masses. :p:Quote:
Originally Posted by Flinch
It is variable. Usually it's at 100C.Quote:
Originally Posted by nosboost300
As the TjMax can be read directly from the CPU in Core i7 there is no more guesswork.