help me i think i know why i can't oc much

[exodusprime1337]

so i was going on for a while about my core temp not reading right, well i was playing with it today and found that the g2 stepping of the 5000 be should have ok core temp readings. so i found out that my proccy is not idling in the 30's but more like the 20's like 24-25 which seems more on par with the 6400 black editions i'm seeing, but the odd thing is is that when i run orthos core 1 goes to 36 and core 2 goes up to 55, and core 2 is always the core that is failing when i give it a bit more voltage. omg what do i do??

[SirKeldon]

It's strange, second core is always the most "fresh" (also in the intel's, i own a Core2Duo laptop) ... in my 6400+ i noticed on full tasks, my Core #0 is about always 7-8ºC hotter than #1.

It seems a false or strange reading of the temps or excess of thermal compound near core #1, i've seen your lapping work and it's damn flat ... so we can blow out the flatness of the equation. Have you tried another programs like SpeedFan or Everest ??

[exodusprime1337]

maybe it's time to lap the ihs on the chip?? taking a vote here guys,

yes: lap that baby flat like th hsf i did yesterday

no: don't risk it??

you vote please help

[trog100]

do it if u like it will achieve nothing usefull thow.. two possibilities for one core reading different than the other.. the heat sensors in the chip aint the same.. one core aint making good contact with underside of the heat spreader..

whatever u do will have no effect on either of the two possibilities..

plus 55 C aint hot either.. least not hot enough to stop your clocking..

sorry dude..:o

trog

[SirKeldon]

Before lapping IHS i'd see if it's really needed ... to replace a HSF is cheap, but replace a 5000+ BE will be the double or almost the triple. Make a test: after lapping my HSF i took it and placed it backwards, then i got the CPU and tested if it was making good contact with the IHS, one nice experiment it's to see if they "attract" to each other or looking with a zoom-glass if the contact it's the appropriate. Make sure your CPU is flat with a metal piece or a "level". If it's right i think something is going wrong with your sensors software or the sensors theirself (maybe you need a BIOS update?)

I won't risk that nice 5000+ B.E. with that huge and nice overclock you have, my temps with high voltages were dramatically high also on my 6400+, but since i've adjusted Processor-NB voltage to a higher value, things went better and temps balanced again. Give it a try to that three things before lapping cause usually AMD's IHS are good flatted, mine it's not polished of course but it's totally flat so i decided not to lap it at the moment cause the IHS and the HSF seemed to have a special kind of "magnetic attraction".

But the decission is yours :cool: ... though if you decide finally to lap it, then share with us how the temps downgraded, maybe we should give it a try :laugh:

[unsmart]

I think the different core readings is normal[ my e6300 is around 4c apart]and do to a got spot in the die not anything with heatsink contact. I believe the hot spot in a AMD64 CPU is over the MEM controller and C2D it's by the cash interface? I don't remember where I read it but there was and article some ware about hot spots being a big road block in CPU speed increases even though over all temps may be low.

[trog100]

deep down inside the chip.. a "hotspot" could be very very small and very very hot.. he he.. they would make the temps we see look silly..

trog

[exodusprime1337]

yeah you may be right, it's just i'm not sure what to do about this. the 5000 is know to have crappy temp sensors, and the evga board.... well it's temp sensors are even worse. i'm just not sure what to do.

to the degree of raising the fsb voltage? how does this help in lowering cpu temps??

it's always one core that fails as well which is the real problem. at idle both cores are about the same 23/24 for core 1 and 22/24 for core two. when i oc though, one core skyrockets beyond the other one, core 2 to be exact. and after a few minutes it's hitting 60. and core 1 is at about 38/40. this is horrible i tested with a lazer thermometer and was able to confirm that the core is most likely sitting at idle in the 23-27c range but at load it's hard to measure, evga reports from it's sensor about 55 full load but that's just from the chip under the die. is there a possibility that i could pull the ihs off and reapply the thermal compound to it and then reaseat the ihs?? is this suggessted or even feasable??

[trog100]

people used to take the spreader off and sit the cooler right on top of the chip.. which is how Tbird chips used to be done..

but the term "obsessive" springs to mind..:)

trog

[exodusprime1337]

well i wouldn't say i'd be obsessing about it, but i would have to say that the severe weight of the ultra120xtreme and it's mounting hardware will probably shatter the core or at least chip it. i may have to keep it on, but i'm just wondering if maybe removing it and replacing the tim, then reapplying it would help??