The Truth about Temperatures and Voltages - Page 38 - Overclock.net

randomizer · 2 Weeks Ago

Quote:

Originally Posted by TwoCables

But I agree that it would be nice to be able to know the exact temperature.

Rip off your IHS, drill a hole into the die hotspots and implant a lab-grade thermocouple. Providing you can get the thing to run, you should know the exact temperature

TwoCables · 2 Weeks Ago

Quote:

Originally Posted by randomizer

Rip off your IHS, drill a hole into the die hotspots and implant a lab-grade thermocouple. Providing you can get the thing to run, you should know the exact temperature

lol

ChickenInferno · 2 Weeks Ago

Quote:

Originally Posted by Clockadile Dundee

Hey...does the 8k-4096k set go for longer before crashing, with the elevated qpi/dram? Does it show any sign of helping?

Have you tried clock skews?

Any theories wrt why it won't pass?

Have you tried with a single 280?

OT: I have a Asus P6T7 WS coming from Newegg. I am going to find out whether my problem is the IMC, or the classified, once and for all.

I think if it is the IMC, Intel will warranty it, and same for the classy, I just want to know which it is. If all this fails, I will detonate with the force of a thousand suns. :P

Now all I need is 500 video cards. I am going to start folding, once I am up and running again. I think I am going to foray into cooling solution creation with this new board.

With the QPI/DRAM at 1.44375v (Bios) 1.402v(Fluke) it will last for 20hrs but it does give me a hard BSOD so I know the issue is still there. I tried it again with the same settings a few days ago but that just yielded a BSOD at only 14hrs in.

Current Clock Skews are 100ps CPU and 100ps on the IOH (Smallest step size available.) I think what I may do is write down my wprime times and see if I find a trend of faster times with different clock skews, whichever yields the fastest time is the one I will test next on prime.

I am completely left scratching my head on this one. It keeps BSOD with 0x000000124 which could mean too little or too much QPI/DRAM. I thinking it is the former as it wouldn't pass LinX with less than 1.40000v (Bios) (1.362v Fluke) but still refuses to pass prime95 with 1.44375v (BIOS) and keeps asking for more. This last little bit of instability just will not seem to go away with voltage and I fear it is a temperature wall (77C max load on Prime95). I am hoping that it is just the clock skews needing to allign and that I can then drop the QPI/DRAM voltage and maybe the vcore a little bit but that may be wishful thinking.

If you RMA the chip to intel, just realize that you will not get back your choice of batch numbers and you may lose a significant portion of your overclock. "A long time ago in a land far far away" when I didn't know what I was doing overclocking, I had a brown out that killed my old XFX 680i SLI LT and thought that my q6600 was dead too. I RMAd both just to find out that the chip was probablly fine :/ I looked at old screenshots with real temp showing my VID as 1.200000. It wasn't a golden VID (and we know that VID doesn't ussually mean that much in terms of overclocking), but chances are that the original chip would have beaten my new (and already sold) Q6600 with a 1.28750v VID that did 3.6ghz at 1.43v, but I could never really stabilize above 3.8ghz with 1.54v.

Hopefully it's just the board and EVGa will send you a nice new board (E762 or higher) with a personal note so that you don't have to explode with the fire of one thousand suns...though that would be kind of interesting to see

Quote:

Originally Posted by almighty15

Any ideas on what the TJMAX is for my 45nm E6300?

Value 100 doesn't seem 'right' in terms of the temperatures it gives me.

Intel never released the TJ Max Target for the Pentium-D 45nm E6300, but I would suspect that it is 100 based on the Thermal Spec and the fact that almost all 45nm CPUs have a TJ Max Target of 100. Sorry for the bad answer but it's the only one Intel has given us.

Clockadile Dundee · 2 Weeks Ago

Quote:

Originally Posted by randomizer

Most 45nm chips float around the 95-105C mark, so yes, I'd say somewhere around 100C would be ok. Just keep the distance to Tj Max down and the absolute temperature has no real importance anyway

Okay, but not 95w.

Alright. It isn't that funny. :P

mav2000 · 2 Weeks Ago

Thanks for the writeup. I have a funny question thoug. You mentioned that the cores would normally be hotter than the cpu temp. I have it the other way around. SO my cores are at 32 and my cpu would be at 34???

Clockadile Dundee · 2 Weeks Ago

Quote:

Originally Posted by mav2000

Thanks for the writeup. I have a funny question thoug. You mentioned that the cores would normally be hotter than the cpu temp. I have it the other way around. SO my cores are at 32 and my cpu would be at 34???

There is a larger margin of error at that specific temp. range. I assume those are idle temps?

FSF-Foxhound · 2 Weeks Ago

Quote:

Originally Posted by TwoCables

lol

you know twocables is amused when he replies with one word

anyway, setting a too high PLL would only damage the motherboard (where the clock generator is)?

ChickenInferno · 2 Weeks Ago

Quote:

Originally Posted by mav2000

Thanks for the writeup. I have a funny question thoug. You mentioned that the cores would normally be hotter than the cpu temp. I have it the other way around. SO my cores are at 32 and my cpu would be at 34???

A.) The core readings at low temperatures are very innaccurate.
B.) The temp monitoring programs that show CPU temp are just making a guesstimate based off of Dist to TJ Max with their original TJ Max value. These temperatures are useless and should be ignored.

--I don't know for sure what overvolting PLL would damage, but I have never seen an increase in stability from it so I ussually leave it on stock.

Side Note: Photobucket was having a wonderful time displaying an old version of the chart. It took someone quoting the chart and me noticing that they were quoting it wrong before I realized this. I deleted off the old version from Photobucket so that this doesn't happen again.

FSF-Foxhound · 2 Weeks Ago

With the i5s/i7s (LGA1156) really benefiting from higher PLLs, there might be a issue with damaging the clock generators.. but I don't care, I've got two!

I'll look some of that stuff up, or ill just ask on XS.

ChickenInferno · 2 Weeks Ago

On the i7 (1366), some people have actually dropped it way down to around 1.30v and are having good results.

		Overclock.net - Overclocking.net > Intel > Intel CPUs
The Truth about Temperatures and Voltages

Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)