dinos22

hi guys

i decided to crack open my ultraD and give it a shot......i ran the sticks at 1:1 275MHz at rated timings 2.5-4-4-8 through test#5 v1.55 50x loops no errors so i decided to go for gold...... i have no previous experience with DFI so i just copied all the setting from eva's guide to overclocking and set the timings at 268MHz RAM (9:10 ratio)....298MHz HTT at 2.5-3-3-7-1T....i am not even sure if burn in is ok for TCCD but i don't see why not........

i've been Burning is just past 20minutes so far at test#5 ...so far I am averaging 20.71 errors per pass.....

EDIT: on the 30 mark now at 20.29 errors per pass

PRELIMINARY RESULTS:
20minute mark 20.71 erros per pass.......
30minute mark 20.29 erros per pass.......

according to T_M that would be considered just on the border of low error count and acceptable for burn in

can you guys give me some advice in terms of settings...considering they are posted on the overclocking guide on how to improve the overclock/burning results

Hardware:
DFI NF4 UltraD
3000+ 0447 Winchester
Corsair TwinX1024-4400C25 (80mm active cooling above ram)
Xpertvision 6600GT
wavemaster with Antec SL400

set cpu voltage to 1.55volts + 1.04%
RAM vdimm: 2.8volts
Multiplier: 9
HTT: 298MHz
RAM: 268MHz (9:10)

____________________________________________________________ _

BEST RESULT SO FAR:

300FSB(DDR600) 1:1 on Twinx1024-4400C25

dinos22

might be a little late....i'll just let it go overnight and see what happens tomorrow....hopefully the damn errors will start going down

40minute mark 19.89 erros per pass.......
51minute mark 19.57 erros per pass.......
1 hour mark 19.41 erros per pass.......
1.5 hour mark 18.86 erros per pass.......




i've been reading up a bit on burning in....there are mixed feeling about burning in tccd...some think bh5 is only ram that benefits and some like tccd and in particular think timings tighten with burning....which is the whole point i am doing this......here is something very interesting


from burn-in.com

Most ICs consist of ion junctions implanted on a silicon substrate. The junctions are connected with small traces of conductive oxides (aluminum oxide being one). In burn-in, the ambient heat and the heat caused by current flow causes the junction temperature to rise. This causes the ion 'clouds' which surround the junctions to dissipate from their original locations. In addition, all of the elements which comprise the chip experience acceleration as the heat causes their subatomic particles to begin to move more quickly.

WHO SAID BURN-IN IS A BORE

T_M

from 20 to 18 in 1.5 hours.... should only take you about 15 hours burn in (overnight) to get there

dinos22

yeah we'll have take off around 2pm

once it reaches no errors (i hope)....do i stop the burn in or keep going what's the deal

groovetek

lol T_M talk about being abit TOO optimistic there!

I had some fantastic Corsair Twinx1024 3200XL TCCD kit a year ago and I gained at most like 4mhz by burning it in. THAT SAID, it was only at 2.8v... but i left it going for like days on end...

also... number of errors is highly dependent on temperature (in my experience with both TCCD and BH5 and UTT... hynix not-as-much) - and my oh my TCCD are some scarily hot chips... hotter than hynix D43, and thus alot hotter than BH5/UTT... I wonder how G.Skill and Centon get away with no heatspreaders on them... hmmm.

eva2000

heatspreaders won't help TCCD, hence why Gskill and Centon don't use them

jlccarv

bad heatspeaders most likely act as insulation and there for dont help much unless actively cooled...

The heatspeaders that OCZ use are about the only ones that actually show decent heat dispensation, even without active cooling...

dinos22

i just calculated average errors for 2 hours....it jumped to 19.37 whathe