Poor Overclocking Newer (ie:0503) Winchester Owners - Check This Out!

[rise4310]

thanks AG i was just gonna start a new thread and re-read this for the umpteenth time and got an idea or 2 to try. seems like my volts really are a problem for me, that problem being i'm a wimp with volts, lol. i gotta remember to just set the damn things where i neeed to be originally and then back them down. i finally got 2475 stable at 9x275 1:1, well still priming but memtest, occt, superpi (29.38 32m ) and 3d's all pass no problem.

 

volts and timings, the damn tref is a killer. are these volts ok for 24/7 ?

 

vid 1.425 + 1.13= 1.59 bios and 1.56 ITE 34c idle 48ish load

ldt 1.4 ITE 1.36

chip 1.6 ITE 1.58

dram 2.9 ITE 2.92

 

the chipset and pwmic are around 41 when priming/occt and 45 with rthdribl running. looks like the AS ceramique helped me out ~5c

[custom_coms]

I'd say yo should be fine...I have to pump 1.64v into my winnie to get 2.35ghz, load max at 49-50 with POOR case cooling (only two fans plus psu 2 fans, can support 8+2 psu fans....)

 

Sam

[spdiscus]

I haven't read through this thread in awhile, but from my CBBID experience, staying at 1.6V, 1.3x1.23 is the best for 259x10, but I have to go with 1.45x1.1 for 236x11.

[SaFrOuT]

the best i can get out of this cp is

 

280x9

Vcore = 1.375 x 123%

 

Load temp is 42C, using Tt Tower112

 

i can't get anywhere more than that or i'll need total Vcore of 1.7 just to be able to hit 2.6GHz

 

and i am stable also @ 2.4GHz @ Vcore 1.53

[rise4310]

well i errored out last night twice around 3 hours running large ffts in prime. i tried lowering the ram volts because i heard it may help some tccd at 2.8, no go. put it back up to 2.9 and found my problem was the ldt ratio, i had it on auto and in a64 it was reading 1100 (4x). i had been told on AT to keep that under 1000 so i dropped the multi to 3x and i'm 8 hours prime now.

[mucker]

Good job Rise! How many volts do you have on the CPU? I am about nearly the same as you 2.5-4-3-10 @ 2.8v. I have the CPU at 1.4v (110%) which has it about 1.49v. Also now testing 275x9. I've got 247x10 stable but want the higher fsb. My CBBID 0505 is pretty flakey around 2.50 GHz on all multipliers. The ram seems to be very good though...

 

m

[Loxguard]

I'm running my Winchester 3200+ @ 2.5 ghz Prime stable for 20+ hours with my vcore at 1.67. Temps stay below 50 C....I like to live dangerously.

[mucker]

If there was a better long term history of max voltage for Winchesters, I'd probably live dangerously too. I just don't have $250 in spare cash laying around for a replacement. I'm actually quite happy to run 2.48 GHz @ 1.44v in my hot climate.

 

m

[rise4310]

mucker, yes its a b!tch around 2500. i really think it has alot more legs though with the ldt down at 3 now. i'm still at 1.425 (1.13) which gives me 1.59 bios and 1.56 in ITE. temp right now is 44c under the large fft's for 10 hours.

 

i think i may have some volts to trim off at this speed though. kinda like i was forcing a square peg thru a round hole if that analogy works. now that i'm getting a better handle on timings and such i might be over volting, we'll see.

 

i figure i'll push upward for a bit first obviously, lol but i'd be happy with this speed 24/7 as it seems these chips need a ton of volts to get much higher. i'd like to touch 290x9 anyway.

 

just looking at your timings, nice, about the same bench as mine on everest (7440/2864/39.2). that ram is treating you well. i wish i could get that done at your volts, nice work.

 

see you here or there bro.

[mucker]

You're doing great my brutha! I've been kind of noticing that with these Winchesters, the higher the speed rating, the lower the voltage needed, however, alot of them are a b!tch to get over 2.5 GHz. AMD should change their Winchester naming scheme, call them all A64 3800 and then purchase a 1.44v, 1.52v, 1.6v, 1.7v variety JK of course...

 

m

[SaFrOuT]

i have noticed very strange thing

my proc 5 hours ago wasn't stable 301~306 FSB with multi = 8x @ current voltage , but i was stable @ 300x8

 

during those 5 hours ran prime for 2 hours on 300x8 and then increased the fsb to 303x8 and ran prime for hour and then increased it for 304 and ran prime for sometime and now i am running stable @ 305x8 with same voltage settings and prime is stable for 28 min now which wasn't posible 5 hours ago

 

could it be that those new winnies like to be increased step by step or what ?

 

it is a CBBID 0503 anyway

 

EDIT: i couldn't resist and i raised the FSB to 308x8 @ same voltage and it is running prime now and stable so far for 5 min <== got an error after 7 min

 

EDIT2: back to 305x8 and i'll let it burn for a while

 

 

[Vista]

Yes, I call this effect "Winnie burn-in". Or maybe it's user burn-in. Or maybe it's a combination of you getting better at tuning and your RAM and CPU and TIM and dust breaking in. Whatever, I'm seeing that too with a CBBHD. It wouldn't go stable with looser timing at 2475 until I threatened it with 1.56v, then I backed off to my sig specs and it seems tamed

 

Edit: I'm beginning to suspect that the AMD 64 has a rookie mode.. if you try to OC with less than 1.5v it assumes you just have a bad PSU and aren't serious. If you give it a kick in the pants it wakes up and resets into Enthusiast Mode. I can see in a few years, we will have CPU's that report back to home base, "Yes, Vista overclocked the CPU by 25.3% in a 20% zone, and look at those peak temperatures! No warranty granted." Sort of like the new BMW's with the snoopy on-board engine monitors.