Temperature too high? or keep going?

[ShikiRoa]

Here's a screenshot of my current OC, been running 12 hours orthos w/o any hiccups yet... I'm just wondering if I should keep pushing some more? Or should i leave it at 3.2 for a 24/7 machine? Also should I lower the multiplier to 8 and go for higher fsb and get 3.2? So I can lower my voltage and temp as well?

System Specs:
E4300 @ 356x9=3.2ghz
DFI Infinity P965-S
OCZ Platinum rev.2 4-4-4-15
Tuniq Tower 120 @ 2k RPM

Vcore in bios i think i have it set to +.125

 

[BillParrish]

if you have PC6400 (800MHz) memory you should absolutley try to do 8 x 400 with a 1:1 memory divider and go from there. This will maximize you memory bandwidth. Your temps should stay about the same as 3.2 GHz is 3.2GHz no matter how you get there as far as the cpu is concerned. You cannot burn up this cpu, it will shutdown if it gets too hot and you cannot disable this function. Around 70C under full load is my personal limit and what I set as an alarm temp in the bios. At 85C the cpu will thermal trip. The speedfan coretemp values are accurate and should be used as a guide. You are just getting her warmed up at 53C plenty of headroom to go, 65C under load for 24x7 operation is not unreasonable

Your current Vcore is 1.4 ish volts as reported by speedfan and that is fairly accurate.

I get concerned (personally) at anything over 1.45 and like to stay around 1.4 or less (but I keep machines forever and worry about really long term effects) Per Intel data sheet 1.55V is absolute max Vcore. Lots of posts on this, its your stuff.

 

[unclewebb]

Your running really stable with low temps at 3200 MHz so why not add another 10 or 11 MHz to the FSB and keep your multi the same and go for 3300 MHz. As long as you keep the FSB at 400 MHz or less you will stay on the faster strap and your performance will continue to improve.

With my revision B2 processor I'm Orthos stable up to 3400 MHz with the core voltage you're using. It would be great to see you equal that overclock so you can get rid of the myth about the E4300 isn't a good overclocker. Nice work.

 

[ShikiRoa]

Thanks for the replys!

One more question, speedfan is correct assuming e4300's tjunction is at 85 right? So is it better to just work with DTS instead of the actual core temp? DTS is just the temperature left until a forced cpu shutdown right? And the DTS is grabbed from the probe built into the C2D chip, so won't it be better to use that as the standard instead of the actual core temperature? If that's the case.... then both Speedfan and Core Temp 0.95 are correct since they both show a DTS of 30°C (Speedfan 53+30=85, Core Temp 67+32=99) So what would be the optimal DTS to keep your cpu at? As Bill suggested about 15°C from Critical? And that's a temp that should make your cpu last for at least 4-5 years?

Sorry for all the questions, and thanks again!

 

[BillParrish]

Thanks for the replys!

One more question, speedfan is correct assuming e4300's tjunction is at 85 right? So is it better to just work with DTS instead of the actual core temp? DTS is just the temperature left until a forced cpu shutdown right? And the DTS is grabbed from the probe built into the C2D chip, so won't it be better to use that as the standard instead of the actual core temperature? If that's the case.... then both Speedfan and Core Temp 0.95 are correct since they both show a DTS of 30°C (Speedfan 53+30=85, Core Temp 67+32=99) So what would be the optimal DTS to keep your cpu at? As Bill suggested about 15°C from Critical? And that's a temp that should make your cpu last for at least 4-5 years?

Sorry for all the questions, and thanks again!

wow you hit the nail on the head, there is a big discussion right now, we thought we knew but now are reinvestigating. The DTS which is inside the core is taking reading from several places in the core and presenting the highest to the software. Speedfan and coretemp and others ARE reading that DTS value and ... woah speedfand and coretemp dont agreee.... ??? oh I see you clicked on the coretemp option (maybe the default but no one uses it "temp to Tjmax" instead of just having it display the core them.

speed fan says 53C

coretemp says 32c till 85C 85-32 = 53C


(sorry had to think it through myself)

I have the intel datasheet open now trying to find if the Tj(max) is listed for each procesor. What we do know is that some have a Tj(max) (called Tjunction in coretemp but that verbage is not really accurate now) of 85C and some have a Tj(max) of 100C and we are fairly certain the dual cores are 85C and the quad cores are 100C but no one is willing to bet their life on it atm. We have also discovered that until recently we thought you could read a register and determine if it was 85C or 100C but that has been found to not be true.

Bottom line The internal trip does work and you cannot burn one up, but the interupt that is susposed to signal that the processor has gotten hotter than settings set by EIST or other BIOS related thermal utilities may not be working. weeeeeeee !!!!

Just keep the coretemp (change that option back ) under 75C and you are good, forever. I dont even let mine run that hot but I watercool.

I am fairly certain your Tj(max) is 85C but am still investigating.


Heat is one thing, high CPU voltages (Vcore) can also degrade long term reliability. IMO keep Vcore under 1.45 volts and I like around 1.4V even better.

 

[unclewebb]

BillParrish: I have the intel datasheet open now trying to find if the Tj(max) is listed for each procesor.

Let us know if you ever find that info. I've gone backwards and forward through the Intel documentation and talked with others and to the best of my knowledge that simple piece of data is not documented by Intel. They also don't document how you can read that info from a processor. Without it, all temperature monitoring software that reports absolute temperatures based on the DTS can not be guaranteed to be 100% accurate.

As you said it is best to watch the DTS value and leave yourself 15C of head room to be on the safe side.