Winnie topping out at 2250?!?!

So yeah I'm decidedly not happy about my preliminary OCing attempts on my Winnie and here to ask you guys if there's anything I've done wrong thus-far. Cause from most reports I'm reading, 2250 is on the very low end for a Winchester. Anyway, here's what I've done:

Dropped the HTT to 3x
Set the RAM Ratio to 1:2
Set startup VID to 1.3v
Set VID to +126% (effectively 1.63v)
Set DRAM to 3v
Set CPU/FSB Frequency ratio to 9x (multiplier)
Raised FSB to 250

So if I've done everything correctly my RAM is running at DDR 250 (even though its DDR 400) with a little extra juice just in case, and my chip is running 9x250 (2250mhz) without intereference from too high an HTT with what should be enough vcore to OC it.

At this speed I BSOD in windoze and get 83 errors in memtest test #5. Temps are 27c idle, 35c after a full memtest.

So I ask, is my Winnie just a rampant POS? Do I have a bad memory controller? Is there something I'm not doing right? Can a burnin save me? Or should I just go back to drinking more beer? As always your help is greatly appreciated.

Edit: Could it be my TTGI thats holding me back?

I'd drop the ram voltage if I were U. Some ram chips just don;t like anything more than 3v.

Have you tried different RAM dividers? I haven't used them much but running your ram at half-speed (ultimatly below spec) then trying to greatly overvolt them doesn't seem that reasonable to me... If they're running below spec, they should run fine at stock voltage right? Did you memtest them at stock to make sure the memory was good?

Regardless, 2250 seems somewhat low... why don't you try setting your multi down to 8 and try to raise the FSB some more in an attempt to see if your chip is just speed binded at 2250. I can only get up to about 2600 before my chip craps out... what stepping and week is your Winchester? If it's a brand new CBBID 0508+ chip that may help explain things (as in, AMD is even further binding their chips for whatever reason...)

Yeah, its a CBBID, I think 0503 knew I shoulda written that down before I buttoned her up. I'll try lowering the DRAM voltage and then try dropping the multi just for ****s and giggles. Will report back soon

I've had three CBBID cpu's. All did right around 2450mhz. I'd lower that DRAM voltage if I were you. Should be fine at 2.6-2.7. Too much voltage almost always causes instability and errors in memtest.

-Collin-

Set a memory divider & continue overclocking. Most likely your memory is not able to keep pace if you're getting errors in memory test.

drop the vcore down.... I have mine at 1.52 running 2340MHz..... and as said already drop the vdimm down to 2.7 - 2.8..... how fast can that memory go btw?

well, I dropped the vDIMM back to 2.6 stcok, I set the multi down to 8x and it booted fine 8x250 with a 3/4 mem divider (DDR 187).

Then I kept the multi at 8x and raised the fsb to 281 for 2248mhz cpu speed and DDR 210 and I got 4234 errors on memtest #5 yikes.

So maybe thats too much speed for the crappy value ram so I kept 8x281 with a 2/3 divider for DDR 187 and the whole test might have well have failed, I cut it off while still counting up from 200000 errors and climbing.

Anyone wanna offer speculation there? Is my RAM not liking dividers? Maybe the mem controller on the CBBID was binned down? perhaps my PS is lagging? Gonna keep trying some other configurations

Try testing for stability at stock speeds. Something seems to be wrong here if you're getting memory errors at that low of a clock speed, with that core voltage, water cooling on the CPU, & your memory running below stock. If a CPU wont run stable at stock then theres no hope for overclocking it obviously.

Ok, update: Went out to test if the RAM just doesnt like dividers so I created a testing matrix with these given settings for all tests

HTT 3x
Multi 9x
FSB 250
vDIMM 2.6
Vcore 1.68
2251mhz cpu

Then I booted with every mem divider and here are the memtest results with DDR speed as reported in memtest

1/02 divider DDR250 - 4212 errors
3/05 divider DDR500 - 4538 errors
2/03 divider DDR333 - 4310 errors
7/10 divider DDR500 - 4232 errors
3/04 divider DDR374 - 4724 errors
5/06 divider DDR416 - 4234 errors
9/10 divider DDR500 - 4382 errors
1/01 divider DDR??? - no memtest
auto divider DDR??? - no memtest

So this seems odd because dropping the vDIMM decreased stability across the board. I went back to the settings that almoast worked before with the raised vDIMM to 3.0v and the errors dropped to 24 (can we say burn-in).

Regardless, it seems my dividers are NOT functioning properly... 3/05, 7/10, 9/10 just default to a DDR500 setting for some odd reason in memtest. Not sure what to make of that, any of you guys got suggestions?

Bieng that as it may with the raised vDIMM I tested the "working" dividers as per the previous matrix and got:

1/02 divider DDR250 - 24 errors
3/04 divider untested
2/03 divider DDR333 - 31 errors
7/10 divider untested
3/04 divider DDR375 - 57 errors
5/06 divider DDR416 - 31 errors
9/10 divider untested
1/01 divider untested

So increasing the vDIMM helped a LOT. definitely wondering a lot now about whats going on here. it seems like bad memory or memory incompatibility, but I'm not sure. Gonna try upping the vDIMM some more see if that gives me more headroom. If you guys could keep tuned in here for updates and advice I'd greatly appreciate it.

oh and to clarify, at stock everything she runs just fine, no memtest errors, no instability. just not responding well to my attempts to push her. i really hope I didnt get a binned chip

Rma the ram either that or the CPU doesn;t like the high htt speed. U have to remember that the cpu now plays a very important role in this. at 250 you really should still be fine but at 281 I would not be surprised to see a bunch of errors because of the cpu.

your ondie memory controller may not like much past 200, if you wanna test that out, get some tccd ram or just grab a venice on Monday!

Mother of God, now my frikkin system wont even power up... Mobo standby light is on but she wont turn over

I'm thinking your CPU was marginal to begin with if you're having so many problems.

Yeah, it might be, but I'm wondering if its the processor... My problems seem more wide-spread than that, more like my motherboard or power supply are damaged. ugh, driving me nuts. I got her to boot again but my raid array is destroyed and I cant seem to recognize one of my hdds, even though they both pass all diagnostics with flying colors, even the long ones. wtf is going on here

man, I have the same problem.. max oc on mine is something like 2230.. cbbid 0501.. ohh well.. maybe someone will buy this so i can get a venice

I know someone with the exact same problem.
cbbid 0501 cpmw 3200 90nm
we tested a 3000 90nm cbbhd 0440 spmw
posted and screenshot @ 2.7 1.6v core. not stable yet, but 2.25 to 2.7= crazy.
my cbbid clocks to 2.5 at 1.57-1.60vcore
-kkat

Try posting this over at the DFI forum www.dfi-street.com or searching their forums... I think I remember seeing a few threads about corsair value ram and value ram in general. The main point their support guys try to make is "why are you putting value ram in a system that is meant for overclocking to high frequencies? it's like putting a chevy engine in a nascar car..." But as I have noticed, many people DO have problems with their value ram as it wasn't made to be OC'd :-\

Yeah, I'm ditching the Value RAM in favor of some of the Muskin Blue UTT stuff that everyone has been clamoring about lately. If that doesnt help maybe I'll sell my Winnie for a Venice

For now she's sleeping so that I dont get angry and find my foot in places it shouldnt be