Post your Highest OC'

[oldDummy]

A little history:

I had my pentium 233 to 300Mhz.

It wasn't stable but I was really proud of that.

When I posted results on a forum everyone thought I was making it up.

That might have been my first "large" OC.

Don't remember what I paid for that 233 but it was a whole lot of money at the time.

Pentium Pro from 166--> 200 [I think] was an easy OC and saved a ton of money.

[crazybus]

so you ran some "benchmarks", played HL2, transferred files, ripped DVDs, burned DVDs, played BioShock.... do you realize that all of these don't stress the CPU to 100%?

sigh.....usually it's a little stressfull when your running a few of them simultanious while you wait for results from one app or another.

Stressful on I/O perhaps but not necessarily pushing the CPU to its limit. It's not indication of rock-solid stability in any event.

Intel's TaT seems to be the only problem...... does that make TaT matter?..... to me no, that TaT wouldn't allow me to overclock more than 100mhz while every other app is good with it..... does that make TaT matter?..... to me no.

IMHO TaT's worthless.

note the IMHO, it's like basing my computers overclock on running Prime 95...... if it's 100% stable at everything except Prime 95 do I really care all that much?

I run it to test and then consider the results as a possible issue but I usually run my overclocks higher and if I run into an issue then I go with "well Prime 95 told me it would be an issue"... the problem with TaT for me is that a 100mhz being a possible issue really hurts it's value in my eyes when it's the only application showing any potential problem.

I guess you don't put much value in stability. I'd rather lose 5 or even 10% in performance than have a potentially flaky machine. In my experience I system that fails at Prime 95 will eventually crash or corrupt something else.
BTW, if you were experiencing overheating and throttling with TAT, there's something seriously wrong with your watercooling setup or the DTS sensor on that processor. I don't use TAT as representative of temps under load (I use prime95 for that) because no real application will push the cpu that hard, but your cooling system should be able to handle it if you want your chip to last very long.

[Flying Fox]

so you ran some "benchmarks", played HL2, transferred files, ripped DVDs, burned DVDs, played BioShock.... do you realize that all of these don't stress the CPU to 100%?

sigh.....usually it's a little stressfull when your running a few of them simultanious while you wait for results from one app or another.

Stressful on I/O perhaps but not necessarily pushing the CPU to its limit. It's not indication of rock-solid stability in any event.

Intel's TaT seems to be the only problem...... does that make TaT matter?..... to me no, that TaT wouldn't allow me to overclock more than 100mhz while every other app is good with it..... does that make TaT matter?..... to me no.

IMHO TaT's worthless.

note the IMHO, it's like basing my computers overclock on running Prime 95...... if it's 100% stable at everything except Prime 95 do I really care all that much?

I run it to test and then consider the results as a possible issue but I usually run my overclocks higher and if I run into an issue then I go with "well Prime 95 told me it would be an issue"... the problem with TaT for me is that a 100mhz being a possible issue really hurts it's value in my eyes when it's the only application showing any potential problem.

I guess you don't put much value in stability. I'd rather lose 5 or even 10% in performance than have a potentially flaky machine. In my experience I system that fails at Prime 95 will eventually crash or corrupt something else.
BTW, if you were experiencing overheating and throttling with TAT, there's something seriously wrong with your watercooling setup or the DTS sensor on that processor. I don't use TAT as representative of temps under load (I use prime95 for that) because no real application will push the cpu that hard, but your cooling system should be able to handle it if you want your chip to last very long.

It's all a matter of confidence. Some need more re-assurance. Some like you are ok with what you have, which is fine. The problem is that all it takes is one "error" to show you potential signs of data corruption or CPU burning up. You may never hit that error, or it may take a long time such that it will never happen while you own the CPU, or it may still take a long time but while you still own it, resulting in unexplained "random" errors. If you are ok with that, no one is going to stop you.

[Fastfreak39]

E6300 @ 3.3 on a Gigabyte P965 DS3
E3110 @ 3.6 on a IP35 Pro
9600GT @ 750/1000/1800
Don't remember what I had my 7600GT at.

[tsoulier]

Let's see , I posted at the beginning of this thread with a AXPM 2500 from 1.8 to 2.4 = 600 mghz
Then a 3700 + from 2.2 to 2.7 = 500 mghz , Opteron 175 from 2.2 to 2.67 = 470 mghz , Opteron
185 from 2.6 to 2.9 = 300 mghz and now Q 6600 from 2.4 to 3.3 = 900 mghz

So really my very first and last were my highest , the Q 6600 can go way higher but not fighting
heat............

EDIT :: Corrected the AXPM
EDIT :: Corrected the Q 6600 Jebus
thanks

[Usacomp2k3]

Presumably, you should measure the 'highest' in one of 2 respects: raw increase in mhz, or percentage increase in mhz.
According to the latter, your xp-m would have been faster

[BoBzeBuilder]

Hm... I can get a 3D Mark score out of my overclocked 8800GTS @ 815/2050/1120 but its not entirely stable. The CPU is holding me back though.

[boing]

I am running my E6750 at 3,4 GHz with stock cooling.

My 7600GS, passively cooled, is OC'd from to 400 MHz to 510 MHz core and 270 MHz to 420 MHz memory. I'm quite proud of the memory OC on that one. Yes, I know that standard 7600GS cards run the memory at 400 MHz. The passively cooled ones however, do not.

[clone]

Stressful on I/O perhaps but not necessarily pushing the CPU to its limit. It's not indication of rock-solid stability in any event.

I guess you don't put much value in stability. I'd rather lose 5 or even 10% in performance than have a potentially flaky machine.

rock solid stability is a machine that never crashes or hiccups which was the case until.......as mentioned previously Intel's TaT.

2 months no hiccups or crashes while running 3000mhz instead of the stock 2000mhz.... no overheating, no issues whatsoever.....that's a 50% overclock in place of TaT's findings which recommended 5%.

you also mentioned and issue with cooling, the only issue was with TaT, I've been running the latest Core Temp app showing a 20c lower temp for the cpu's and they've yet to exceed 51c even under synthetic load since.

I'll make this the last post in regards to TaT...... to those who've read my posts and responded thanks but it appears many stopped reading right after 1 item or 2 and seemed to miss entirely that TaT was the only app that caused any issue while others were more interested in a lecture than offering an answer.

no grudge thanks all, the fillport finally arrived so my E4500 water cooled system on Gigabyte GA-P35 with 2gb's of Mushkin is the next machine to get pushed beyond stock.

moving to 1/2 lines on this cooler with a D-tek block instead of the PolarFlo assembly with Swiftek pump I'm currently using.

[Jigar]

Q 6600 from 2.4 to 3.2 = 600 mghz

You need to edit it again.. 2.4 to 3.2 = 800 mghz

[ucisilentbob]

C2Q6700
2.67==>3.45

[Fighterpilot]

Sure is a nice chip this one: 1.3GHz overclock with Zalman CNPS 9500

[sandygws]

QX9770 @ 4004.3 MHz

[pikaporeon]

I got 3.49 on the Q9300, totally unstable.
I can get 3.25 stable on the same chip, but I'm at 3.1 because I can downvolt it so low. Folding right now it's at like 45 degrees

[Richie_G]

Not too experienced with overclocking, but when I built this new machine a few weeks ago I chucked in an E8400 since I've heard a lot of good things about this chip. Anyways it seems like I've got a good bin number since I immediately cranked it up to 3.6ghz on a EP35C-DS3R and it hasn't complained at all. I haven't bothered stress testing it to ensure stability because frankly I'm too lazy, but since my machine is essentially a gaming platform I'm not too worried.

I've not bothered seeing how high it'll go because one, I'm not that brave, and two: I don't think I need it. When software starts to demand more from my machine that I start to notice a performance penalty, I'll squeeze it some more. Same with the GPU incidentally.

[Krogoth]

The one E6300 (1.86 Ghz) that I used to have went all the way up to 3.0Ghz (or 63% overclock) with a modest Vcore boost before my Motherboard couldn't handle the FSB. I am sure it would be able to do at least 3.2Ghz.

My current B3 Q6600 (2.4Ghz) can handle at least 3.2Ghz if had a little more volts and much better cooling. It rests right now at 2.93Ghz using stock vcore and 60-70C when fully loaded.

I had also previously overclocked a Venice 3200+ (2.0Ghz) and Opetron 165 (1.8Ghz) to respectfully 2.4Ghz (20% overclock) and 2.5Ghz (38% overclock).

[mattsteg]

I picked up a e2140 for my wife awhile back and it sucks in the overclocking department. 2.8 ghz is all it can do.

I decided to buy an E2180 thinking 3000mhz would be decent given the cost and it would make a good stand by until the 45nm parts arrive but after 3000mhz for 2 months and never feeling near as fast as my E6600 at stock let alone 3000mhz or the 3600mhz I was running it at........ an idiot recommends I use Intel's TaT to test it which promptly overheats it badly each time I'm using it, I've got a good water cooling and at the time am confused but check and recheck the seat after changing the past and making sure on pump rpm... according to TaT 2100mhz was all my cpu should have gotten before burning a hole to china....... so I mention I believe the app is garbage and something is wrong but I didn't do any research and idiot claims he has and stands by his comment "it's an official Intel benchmark and diagnostic app".....after toasting my dual core for another day I delete the app but upon restarting the computer it suddenly won't post any higher than 2600mhz..... within 2 days it's dropped to 2200mhz and now I'm stuck at 2000mhz not entirely stable......I intend to unplug the pump given it was an Intel app that killed my cpu it's their dime and no I'm not concerned and no my conscience is clean on the matter, just waiting for the 2nd system to be built.

No, you don't. If your cooling isn't up to what the processor can do, it's insufficient. Quit blaming others for your incompetence and screwing over the rest of us who aren't thieves.

TAT is one of the most stressful apps that one can use for an Intel CPU. This means your OC does not really hold even you think it does.

the problem here is relevance and having done some research on my own I've found out that TaT is virtually worthless having been developed for mobile cpu's it's temp indications are fundamentally flawed and the synthetic load test seems dubious at best.

It's a worst-case. There's nothing dubious about that. If your cooling capability is borderline/marginal (like yours was) and you run a worst-case thermal loading tool, it's your responsibility to monitor your temperatures using a trustworthy tool.

1st I ran the 2180 at 3000mhz and played BioShock from Start to Finnish, I played Half Life 2 the entire series from start to finnish, I burned and decoded numerous DVD's, burned files like crazy, did file transfers, benchmarked the system extensively........

Few of those are significant loads on a multicore CPU. Several aren't even significant loads on a single core CPU, and mentioning them at all just labels you as clueless.

I did everything that ppl usually do when using their computer with gaming obviously being the most demanding

The problem with that is that gaming isn't demanding[/url] on multicore CPUs, with a handful of exceptions.

although DVD decryption is usually somewhat stressful and for months the system was fine no hiccups, no pauses, nothing .... p.s. I knew ahead of time the cut down cache would hinder performance but I was expecting more and the cpu was never what I would have considered a good value to me.

anyway so I'm living with this cpu and I'm introduced to Intel's TaT application.

running the cpu's at TaT's 100% immediately shoots them to an indicated 85c and they clock throttled........ even running the cpu's at 50% resulted in eventual overheats and clock throttling....... this from cpu's that were powering a system under all sorts of real world loads and ran without a hiccup........50% load is too much by far and BioShock or Half Life 2 gaming for 3+ hours doesn't load the cpu more than TaT in less than 2 minutes?

What does time have to do with it when your cooling system is grossly underspecified to deal with your CPU at maximum load?

...... I'm not saying they weren't being stressed but I am saying what is TaT doing to kill these cpu's like nothing else can....(no question mark it's rhetorical)... and not just somewhat more but the synthetic results didn't reflect real world useage nor even remotely come close.....

That's likely because your real world loads are pretty small.

additionally understand I'm using a PolarFlo water cooling system with dbl 120mm fanned radiator and swiftech water pump..... this same cooling setup kept an Intel P4 running 4200+mhz under 55c, an opteron dual core running 2800mhz under 53c under full "real world" load for both.....

Maybe you did a crappy job installing it.

in the case of the Opteron I burned 9 DVD's in 2 hrs while gaming as much as I could while running WMP, while websurfing, while doing a 9gb File transfer trying to punish the system as it was my fist dual core..... it had 2LG DVD+-RW's, 2gb's of ram and Striped Raptors which I'm sure helped.

Burning DVDs: not a significant CPU load.
Gaming: not a significant CPU load.
WMP: not a significant CPU load for most things, and certianly not for anything that you could do while gaming.
Webserving: not a significant CPU load
file transfer: not a significant CPU load.

Sounds like you were stressing just about everything except the CPU!

anyway so I figure "well let's see what TaT say's I can do for an overclock without overheating?"..... the results 2100mhz... yes folks TaT would overheat the 2180 when under it's 100% load even with water cooling if left running long enough even at a meager 100mhz overclock and what made me consider even more was that without a doubt my cooling system was far more efficient than the stock Intel heatsink... this made me question whether TaT had any value at all but sadly by then it was too late and the damage was done.

to me TaT is entirely worthless given the 2180 would have overheated at it's stock 2000mhz with the stock heatsink had I used the tool making it worthless...... punishing the cpu to gain information is something, TaT isn't IMHO.

Sounds like you're blaming your own faulty setup on TaT.

so you ran some "benchmarks", played HL2, transferred files, ripped DVDs, burned DVDs, played BioShock.... do you realize that all of these don't stress the CPU to 100%?

sigh.....usually it's a little stressfull when your running a few of them simultanious while you wait for results from one app or another.

It's far more stressful on IO. It's not a heavy CPU load. There might be several low-load apps vying for CPU time that make a dual pay off in smoothness, but the CPU still isn't loaded particularly heavily under such a workload.

the problem here is relevance

1st I ran the 2180 at 3000mhz and played BioShock from Start to Finnish, I played Half Life 2 the entire series from start to finnish, I burned and decoded numerous DVD's, burned files like crazy, did file transfers, [b]benchmarked the system extensively

Intel's TaT seems to be the only problem...... does that make TaT matter?..... to me no, that TaT wouldn't allow me to overclock more than 100mhz while every other app is good with it..... does that make TaT matter?..... to me no.

IMHO TaT's worthless.[/quote]If you don't want a robust system with a decent safety margin and have no need to actually use your processor, good for you, I guess.

note the IMHO, it's like basing my computers overclock on running Prime 95...... if it's 100% stable at everything except Prime 95 do I really care all that much?

I'd certainly prefer avoiding that potential for instability and silent data corruption. Plenty of problems caused by CPU instability just aren't apparent immediately, even though they're still doing damage.

I run it to test and then consider the results as a possible issue but I usually run my overclocks higher and if I run into an issue then I go with "well Prime 95 told me it would be an issue"... the problem with TaT for me is that a 100mhz being a possible issue really hurts it's value in my eyes when it's the only application showing any potential problem.

It's the only application that you've mentioned that even approaches loading your cpu. If you can't run it, your safety margin is about nonexistent. That's scary.

Stressful on I/O perhaps but not necessarily pushing the CPU to its limit. It's not indication of rock-solid stability in any event.

I guess you don't put much value in stability. I'd rather lose 5 or even 10% in performance than have a potentially flaky machine.

rock solid stability is a machine that never crashes or hiccups which was the case until.......as mentioned previously Intel's TaT.

Stability doing what? You never loaded it and apparently you don't give a hoot if your data is corrupted or not.

2 months no hiccups or crashes while running 3000mhz instead of the stock 2000mhz.... no overheating, no issues whatsoever.....that's a 50% overclock in place of TaT's findings which recommended 5%.

There's more to instability than crashes. Also, wtf is this 5% "recommendation" that you're pontificating about?

you also mentioned and issue with cooling, the only issue was with TaT, I've been running the latest Core Temp app showing a 20c lower temp for the cpu's and they've yet to exceed 51c even under synthetic load since.

What temp did they reach when testing with TaT, from Core Temp?

I'll make this the last post in regards to TaT...... to those who've read my posts and responded thanks but it appears many stopped reading right after 1 item or 2 and seemed to miss entirely that TaT was the only app that caused any issue while others were more interested in a lecture than offering an answer.

If anything's causing an issue of that magnitude, you're way too close to the edge of instability for the taste of most. The answer remains: make sure your cooling is up to the job.

[clone]

Maybe you did a crappy job installing it.

was decidedly observant during tear down, all was fine.

all components are good to go and going with the new cpu, surface checked the water block it's flat, the install was flawless the cooling ability well beyond TaT, no changes to the hoses, water pump, radiator or routing....simple pull inspection and swap...... the issue was a mix of TaT and crappy CPU.

have since sold the rma'd cpu and replaced it using an E4500 running 2900mhz no issues, running Intel's TaT at 100% temps never went past 31c at 2 hours running time shut it off due to lack of interest.

Quit blaming others for your incompetence and screwing over the rest of us who aren't thieves.

reviving a dead thread in pursuit of the moral high ground seems quite sad.

you never had more than opinion and blind faith.... and of course the generous use of Bold... .... all have failed you, I can understand your bias and logic but have no interest in your tone, that your wrong is just icing on the cake.

seek validation elsewhere you've gained nothing here.

[Fighterpilot]

Well said Clone.
@ nice overclock on the Yorkfield quad there sandygws...congrats!..... I'll bet it goes like a rocket at that speed.

[slot_one]

Opteron 185 to 3.0GHz.
DX4-75 to 100MHz.

[Gerbil Jedidiah]

Q6600 at 3.6GHZ running two Folding at Home SMP Clients 24/7 (This is 100% load) No crashes, just high PPD.

I also OC'd an Athlon X2 AM2 4000+ from 2.0GHZ to 2.97GHZ stable enough to run games and most programs... Had to back it down .05 or so to get stability in Prime95.

[Glorious]

you never had more than opinion and blind faith.... and of course the generous use of Bold... .... all have failed you, I can understand your bias and logic but have no interest in your tone, that your wrong is just icing on the cake.

You returned a CPU you broke as defective. That's wrong. You know that it's wrong because you felt the need to justify the decision. Any argument otherwise is literally saying that software broke your CPU. That's absurb. Your overclocking with improperly installed cooling did that.

This isn't really something particularly suited for this forum, but RMAing equipment that you broke is bad. Don't do it. Furthermore, don't insult our intelligence by claiming that software did it. That's absurb.

Well said Clone.

How so? He still insists that software is responsible for his CPU failure because he doesn't have any idea what a real CPU load is. That's irresponsible and entirely inaccurate. No one on a technical forum should support such a notion.

[Fighterpilot]

I can understand your bias and logic but have no interest in your tone,

I take it that you support the tone of reply then...

[themattman]

That is the highest completely stable OC I can achieve. I am 99% sure it is either the motherboard or a core on the Q9450 that is holding me back. I would like to grab an E8400 to test this out, but until then a Q9450 at QX9650 speeds isn't too bad.

All my BIOS voltage settings are on normal and my vcore is set at 1.18125. Temps are around 50C when running F@H.

[Glorious]

take it that you support the tone of reply then...

Why does that even matter? You're not even disputing that what he said was technically incorrect and that the RMA was dishonest and wrong.

The truth is the truth. Tone doesn't have anything to do with it.

[clone]

You returned a CPU you broke as defective. That's wrong. You know that it's wrong because you felt the need to justify the decision. Any argument otherwise is literally saying that software broke your CPU. That's absurb. Your overclocking with improperly installed cooling did that.

This isn't really something particularly suited for this forum, but RMAing equipment that you broke is bad. Don't do it. Furthermore, don't insult our intelligence by claiming that software did it. That's absurb.

did I break it? Intel saw it differently..... and your opinion means.... to me?

cooling wasn't the issue.

you seem to be under the mistaken impression that your opinion may have some value to me, it doesn't and not to be repetitive but revisiting a dead thread to spout off making baseless accusations in order to gain the moral high ground is just as sad as it was earlier.... hey but why stop now eh continue talking from no position of relevance.

timing and circumstance implied TaT had a significant part to play in the cpu's demise whether it was defective cpu or by a sudden spot overheat due to TaT's specific benching regiment.... either or the issue was Intel's fault, Intel software and Intel CPU..... Intel agreed, end of story..... unless of course you wish to waste more time making baseless accusations in order to prolong it.

as the actions of a troll will dictate obstinance will likely rule the day from this point forward.

p.s. try "absurd" next time as you screwed the spelling of the word multiple times.

[Hance]

OK I have had all of the sniping back and forth that I can take for one thread. Feel free to discuss all things related to overclocking. If you want to have a moral debate take it to R and P. Further off topic posts will be nuked on site.

Hance
Overclocking Tweaking and Cooling Moderator

[Glorious]

cooling wasn't the issue.

At the risk of a nuke, this is 100% technically incorrect. If the chip is overheating under a software load, yes, your cooling is the issue. There is no way that your cooling couldn't be the issue. This isn't opinion, this is quite simply fact. If your cooling system cannot safely handle your CPU at maximum power than you need a better cooling system or you had better stop overclocking.

TaT had a significant part to play in the cpu's demise whether it was defective cpu or by a sudden spot overheat due to TaT's specific benching regiment

The notion that the failure was the result a "sudden spot overheat" is also baseless. The thermal gradient of even the most strenous workload (including TaT) is at most 5 or 6 degrees celsius. Reference Figure 5 at page 4 of "Temperature measurement in the Intel® CoreTM Duo Processor"

http://arxiv.org/ftp/arxiv/papers/0709/0709.1861.pdf

So, at absolute worst, we are talking a difference of about 5 degrees. That's not nearly enough to damage the processor unless you are already running at temperatures that are borderline disasterous (not to mention completely unstable).

the issue was Intel's fault, Intel software and Intel CPU..... Intel agreed, end of story

Overclocking automatically means the failure is not Intel's fault, regardless of all other considerations. Operating equipment outside of its rated parameters means any failure is not the responsibility of the originating company. This is not opinion, this is fact. Not only that, but overclocking explicitly voided your warranty. As per Intel's CPU warranty publically available at:

http://download.intel.com/support/proce ... rranty.pdf

Intel does not warrant that the Product will be free from design defects or errors known as “errata.”
Current characterized errata are available upon request. Further, this Limited Warranty does NOT cover:

...
any Product which has been modified or operated outside of Intel’s publicly available
specifications...

Even if Intel accepted your processor upon return, you were still violating the terms of their warranty. They just didn't know. That's not Intel "agreeing." That's just you defrauding Intel.



Mods, you can consider this sniping, but I consider it to be a necessary refutation to the falsehoods that clone has continually stated in this thread. Everything I wrote is either common sense or 100% verifiable with publically available documentation. Characterizing such a response as "sniping" will only mean that inaccuracies will be allowed to pervade these forums provided their purveyors are suitably indignant when corrected.

[mattsteg]

A few points are worth highlighting for their general relevance to overclocking.

The thermal gradient of even the most strenous workload (including TaT) is at most 5 or 6 degrees celsius. Reference Figure 5 at page 4 of "Temperature measurement in the Intel® CoreTM Duo Processor"

http://arxiv.org/ftp/arxiv/papers/0709/0709.1861.pdf

Heat transfer within a CPU and from the CPU to the external environment is an important overclocking/cooling topic. It's also important to Intel, so they have lots of documentation on it. Heat's generated in specific areas, to a large extent, so specifications like maximum gradient are pretty useful. An overall understanding of how heat is transferred from one stage in the cooling system to the next is quite useful.

the issue was Intel's fault, Intel software and Intel CPU..... Intel agreed, end of story

Overclocking automatically means the failure is not Intel's fault, regardless of all other considerations. Operating equipment outside of its rated parameters means any failure is not the responsibility of the originating company. This is not opinion, this is fact. Not only that, but overclocking explicitly voided your warranty. As per Intel's CPU warranty publically available at:

http://download.intel.com/support/proce ... rranty.pdf

Intel does not warrant that the Product will be free from design defects or errors known as “errata.”
Current characterized errata are available upon request. Further, this Limited Warranty does NOT cover:

...
any Product which has been modified or operated outside of Intel’s publicly available
specifications...

Intel's warranty makes it pretty clear what they do and do not cover. Explicitly listed under "non-covered" are any processors run outside of specifications. This leads to a "you break it, you buy it" situation. When you turn up the clocks/voltage, you're taking responsibility for the consequences. A lot of things are generally quite safe, but exceptions exist and plenty of things are decidedly unsafe, especially when you start manipulating voltage. The bargain made by the ethical overclocker is that increased performance comes attached at the hip with increased responsibility. Sometimes that responsibility includes voluntarily eating the cost of something that you broke, even when you could bamboozle your way into a free replacement. The alternative, obtaining free product under false pretenses from a company that most of us buy things from, raises prices for everyone.

[Glorious]

After reviewing this thread, I've realized that one of the reasons for clone believing TaT to be a flawed application is likely based on a misunderstanding of where thermal measurements come from in a PC. This is actually a very, very common problem that needs further explanation.

you also mentioned and issue with cooling, the only issue was with TaT, I've been running the latest Core Temp app showing a 20c lower temp for the cpu's and they've yet to exceed 51c even under synthetic load since.

The most likely explanation for this observation isn't technically a difference in measurement. It's simply a difference in where that temperature is being read from, or what stored value that measurement is being computed against.

You can get CPU temperatures from two places these days. You can either get them from analog diode on the motherboard near the CPU that's intepreted by the BIOS, or you can get them from the DTS (digital thermal sensor) that's on the die of the processor.

If these programs got the CPU temperature from differing places than obviously the values could be different without anything being wrong with either program. Not only is the diode on the motherboard not actually on the CPU, but the temperature provided comes from differing circuitry measured by differing algorithms in the BIOS. In your particular case, however, both of those programs get it from the DTS by reading a MSR (machine-specific register) on the processor.

But this means we run into a different problem, which is what stored value that measurement is being computed against. What you read from the MSR in Intel's DTS isn't actually a temperature. It's a counter that counts down to TjMax (or Tjunction). This value isn't explicitly stored. The published TjMax for mobile Core2s is 85c. The TjMax for desktop processors is not published and differs between cores and maybe even steppings, but for conroes it's widely speculated (and confirmed by experiment) to be 105c.

This immediately explains the 20c difference. You calculate temperature by subtracting the DTS counter from TjMax. Intel's TaT (thermal analysis tool) gives you the 20c higher core temp because it's using the 85c TjMax value. This is because TaT is intended for use on mobile processors even though the processor in this case is really a conroe with a real TjMax of 105. Core Temp gives you the lower (and correct) core temp because it's using the appropriate TjMax for your processor.



So TaT really isn't at fault. It's reading the exact same MSR as CoreTemp, the only issue is that it is applying the wrong TjMax to it. The lesson here is to try and understand what your tools are doing before dismissing them as flawed because their results appear to be inaccurate. TaT is designed for mobile processors, not desktop ones. This is why it will identify any desktop processor as a Pentium M. Using the right tool for the job is essential, and when using a tool that's not specifically designed for the job it's essentially that you understand what it's doing so that the result is correctly undertood.