x48 + QX9650 = extreme anger. please help

[Fitseries3]

i have this OCZ freeze, shin etsu and AS ceramique

 

[Solaris17]

leave the pads if the chips dont make good contact if they make good contact leave the pads off.

 

[Fitseries3]

which TIM should i use? i need to know asap!

 

[Fitseries3]

i used the OCZ freeze that i got with the QX9650.

oh... ebay guy/lady... if your reading this... i'll get back to you. i think i will just RMA this chip with your help. no hard feelings. things happen. i know its not your fault. i was not blaming you. you are very honest and a great seller. thats why i contacted you before doing anything stupid like leaving negative feedback. the sensors are still stuck but i want to make 100% sure the chip is flawed before i send it out for RMA.

this qx9650 stock cooler is actually PRETTY good. the fan is insane! it has a high/low switch on it but im guessing somewhere around 140cfm on high and about 95cfm on low.

 

[Fitseries3]

well if you looky here.... you can see i fixed something. dont know what, but one thing is for sure....

1.3v for 4ghz is pretty nice.

 

[Solaris17]

well if you looky here.... you can see i fixed something. dont know what, but one thing is for sure....

1.3v for 4ghz is pretty nice.

ahh cooling the bridges congrats fit

 

[Fitseries3]

POS 7 drivers from ATI wont install at all.

 

[Solaris17]

POS 7 drivers from ATI wont install at all.

funny considering their supposedto be more compatible on your OS.....

 

[Fitseries3]

this is crazy.... first it gave me hell and now this...

OH... and this is with the stock cooler from the QX9650...

 

[Solaris17]

................................hmmmm i have a sneaking suspition.....congrats im glad its working...but oc the cards i want to confirm a theory..

 

[Fitseries3]

i went from 4.1ghz -> 4.5ghz on that voltage with setfsb in windows. 4.6ghz makes 3dm stop in the cpu test. i esc out and everything is fine. im testing 4.5ghz with 4850 OCed.

 

[Fitseries3]

is this what you expected?

 

[Solaris17]

i went from 4.1ghz -> 4.5ghz on that voltage with setfsb in windows. 4.6ghz makes 3dm stop in the cpu test. i esc out and everything is fine. im testing 4.5ghz with 4850 OCed.

hmmm but it wont do it with the X2's thats whats getting me

my first 3 resolutions are as follows

this is what i think could be the problem

drivers (but we tried tried that)

then i though wait a sec it might be a bios option/driver issue

or PSU i know your using dual and its a long shot but loaded and OC'd you might have an amprage issue on the secondary PSU

im thinking

if mobo issue with 2 cards their might be (if their is) a bios tweak you can do..for example for SLI i had problems at first but in my bios theirs an option called sli appeture size (or similar) and i found going from auto to disabled fixes some stability issues and it affects scalability for example when disabled i loose some sli peformance but i can more stabiltiy perhaps their is a similar xfire option you should try adjusting...

for drivers im thinking that while they seem to work amazing with W7 maybe you should try vista and see if the same things happen

for PSU some times when the card is underpowered you get a black screen artifacts slow performance etc...try switching you 850 to the cards and 750 to the rig or vis versa...also...instead of plugging in the cables (the 6 and 8pin) try bumming off the rails and split the rails up if you have adapters instead of running them off of 1 molex string try a diffirent molex string and split up the power draw.

these are just some ideas i had..

 

[Fitseries3]

im using the 1200watt pc power now.

 

[Solaris17]

im using the 1200watt pc power now.

ahh try hooking up your 750 as a secondary to the cards....when the system loads the rails might drop maybe even something as little as .0x volts and that may cause a fluctuation large enogh to trip the cards up..as the cards are probably drawing quite a bit of wattage as well as a proc the general rule of thumb is that the more wattae they are drawing the smaller the disturbance can make them glitch...i dont mean to be giving ideas that waste your time im just trying to solve a puzzle i mean i have no doubt in my mind that psu can handle it but the mobo chokes and voltage regulators etc may be whats causing the flux if any...say your cards spool up and start drawing wattage for that split second inbetween tests were the cards arent laoded when they load again their may be a flow delay between the cicuits starving the cards know what im saying?

 

[Wile E]

i feel like im totally relearning everything on this board.

i have p45 down and i could school anyone on p45 but this x48 is heckling me like MAD.

all the settings seem to do nothing to make things better.

i have already identified that this qx9650 has a few problems...

1. stuck sensors: reads 20, 20, 9, 4 at all times on air , water, and TEC
2. FSB wall at 462mhz: tried all multis from 6 - 14(obviously the higher ones did not boot)
3. needs TONS of voltage to boot into windows although it can post at low volts

Mine walls at 449Mhz. Don't feel bad. Not even clockgen or SETFSB will get me past it. My Q6600 would go all the way up to 475fsb, so it's the chip.

im using the 1200watt pc power now.

When benching, people like Hipro and Kingpin are using both a PCP&C 1200w and a Corsair HX1000 at the same time. Might be a good idea to try the 2 psu solution like Sol said.

 

[fps_dean]

i have tried all auto

vcore = 1.35 - 1.6v
cpu gtl = auto, .63, .65, .67
pll = auto - 1.74v
fsb term - auto - 1.5v
dram - 1.7 - 2.1v (ddr3)
vNB = auto - 1.71v (fan on NB, temps fine, 35c max)
nb gtl = auto, .61, .67
vSB = auto - 1.2v

tried different FSB straps and memspeeds, loosened timings, everything man.

Try this.

vcore = 1.32v
LLC - enabled
cpu gtl = 67%, try 65% if 67% fails but 67% should do better
cpu pll = 1.55v (no need to go higher than this)
fsb term - 1.37v
dram - stock for your RAM
NB = auto should be good here
nb gtl = most people report the NB on the X48s do better at 67%, but try 61 too
SB = leave this on auto, leave your PCI-E frequency to 100

I'd also suggest lowering the multiplier for your RAM to run it at a nice and low speed at stock timings, let it pass memtest for 20 minutes and leave it at a lower-than-rated setting so we can be 100% sure that's not the problem. When you get the CPU stable bring this back up.

If this fails the chip is probably a dud.

 

[fps_dean]

is this what you expected?

I get the same score with my e8600 stock, is there that much of a difference between XP/2003 and Vista?

 

[Fitseries3]

usually i get several different errors upon BSOD but with this chip i always, 100% of the time, get 0x000000124. there is never a different error code. it just leads me to think that the chip has something wrong with it.

what i did was run the e8600 all the way to 4.8ghz on the board and found its settings. today i popped the qx in and left the settings the same but raised the vcore.

it's currently running 4.2ghz at 1.4v even BUT its set 1.4675 in the bios. there is no vdroop fix for this board and loadline cannot be disabled. there are 2 settings for loadline, one being normal and performance is the other. when on normal the board runs fine as long as you have everything right. however, performance mode, which helps with the vdroop, causes BSOD all the time.

 

[Fitseries3]

nope.... total fail. i can run windows but no benching at all. dies instantly with the 124 error code.

 

[Fitseries3]

tell me what you make of this....

higher speed... lower score.

previous score....

 

[Solaris17]

bus speed/multi background programs....or a hint of instability

 

[Fitseries3]

so what should i do here? send the board back? RMA the cpu? sell it all and get nehalem?

 

[Wile E]

so what should i do here? send the board back? RMA the cpu? sell it all and get nehalem?

I say try to get a RMA on the QX first. Who knows, you might get a C1 (tho I personally haven't seen any around ). If you're worried about them not accepting it, crank the pll up to 2V and fry it. Ask DaMulta, it works like a charm.

 

[Fitseries3]

it seems like both my e8600 and the qx9650 need to be rmaed. they both need .12v more on this board than the p5q3. im thinking to RMA both board and CPU at the same time. that way i'll have everything new when i get both back.