Printable View
Hi.
I have an Asus P5Q-E and XEON X3350.
I am currently at 3.6GHz 1.28V, 450fsb stable. But when I go to 480fsb, which is 3.86GHz, I am not stable at all, not even at 1.4V!
I know this is something to do with GTL Refs.
I had the following voltage settings:
1.3V VTT
480fsb
1.28V NB Voltage
1.45V Vcore
I tried setting GTL Refs but had no idea what to set them to.
Any ideas?
Thanks
Will.
More increases the value of VTT (1.34~1.38), NB (1.32) and GTL Reference 0.63x to try both. Then do the stability test.
Hi.
I tried 1.38V, and even 1.4V on the NB, and the GTL Refs, it just ends up crashing shortly in Prime 95 :(
Best I've had is 4mins before crashing in prime 95, got about 20mins so far, not long I know, but I'm making progress, I hope the progress is not for nothing though!
Got Vcore at 1.45, I'm hoping it'll be fine at 1.4V since it's stable at 1.28V at 3.6.
what are you doing with the ram,
how much, any increase in volts for that,
I found that once you start getting up there in fsb, the nb needs it's volts...
Well the ram is below stock speed, and on stock voltage.
NB is at 1.4v and stock is 1.2V.
I've lowered the fsb to 475 and voltage to 1.4V, hopefully at 3.8GHz I might have some hope. I will feel happy when it gets to 1 hour, I know it really needs 8-12 hours.
Don't know if it's the mobo not capable of 480MHz or if it is just settings.
Can anyone help please?
I've got the following volts on it at the moment:
VTT: 1.4V
NB: 1.4V
CPU: 1.38V
CPU GTL 0/2: 0.650
CPU GTL 1/3: 0.655
NB GTL: 0.650
FSB is at 470 and ram is underclocked at 930.
It passed IBT 20 runs with 2gb ram fine, over and over again, it passed hours of blend, but froze in BFBC2.
P5Q series with quads prefer other ram divider than 1:1...
Disable static read control, TRFC to 55, Clock twister Light/Lighter
Which prime does crash ? small, large blend ? Try small first custom 20-20K set to test CPU and Vcore... then move onto FSB and NB volt tweaking, these need to be properly set (more isn't always better !)
Oh, that's surprised me, I was thought all mobo's and overclocks 1:1 was the one to go with!
So it does not matter what divider it is as long as it's not 1:1?
It was crashing on blend, I changed GTL's, it went longer on blend, and Small FTT to test CPU. I really don't have 8 hours to test the PC though, well, when I go on my bike rides, or go to college I could, but I'm scared of leaving it on, if it like crashes and is in a crashed state for hours on end, or just a fire hazard.
I tried the settings Leeghoofd recommended and it crashes, 1 hour into prime blend :(
have you tried 20-20K custom first ? to set up the NB and FSB GTl's please use a bootable Memtest cd and see if it passes at least 2-3 full runs
I wouldn't expect a lot more than 480fsb stability with a P5Q-E. It's not the top end of the P5Q series and even they have trouble getting 480+. If you want 500FSB+ I'd try a Gigabyte UD3 series, extreme, or USB3. You could also be nearing your CPU limit so buying a new mobo could be a bit of a waste if it doesn't improve things - it did for me though!
I didan't have issues with the P5Q Dlx to get 500FSB, but the rmas needed to do 1200Mhz, on another divider it was hit and miss. It requires, as with all boards, correct setup of the voltages and GTl's... it takes some time to get there, and that counts too for UD3P and co...
I think the main issue here are the 4 dimms and I have no clue how high they can get stable...
I couldn't manage it on my P5Q-D or -E at all, and I tried with 1296+ capable reapers, and a pair of Gskill PI9600's which were good for 1250+. Even the 2 MIIF boards with their very fine GTL controls got me a barely benchable 506x9. I only know some with older Q6600's got stability with 500fsb, but less so on 45nm quads. Maybe I didn't look into the P5Q enough, particularly after hearing it clocks better than the MIIF, but I spent enough time (almost a year) trying with the Asus range and moved on.
All 3 of the EP45 range I now own do 500FSB with ease and good enough cooling, and have benched beyond 540. I admit that a change of mobo would be a bit drastic for that 100-200MHz increase though!
4 dimms won't make it any easier for sure, you might want to try 2 sticks for now and see if that improves things. If it does, maybe try increasing MCH volts with all 4. Try a higher TRFC for 4 sticks too, something like 75-80 to be certain. You could also raise your performance level to ease NB stress.Good luck.
Well I had it at 475fsb a couple weeks back, not stable, but I notice how it does Small FTT's in Prime95 for over 8 hours, yet fails Blend within 1 hour or maybe 2.
The ram is underclocked from 1080 I think it is for PC8500 to 940MHz and slightly overvolted.
The NB and VTT is high at 1.40v
At 3.7GHz it's running fine off 1.38V, yet for 3.8GHz I can't even get it stable at 1.45V, so something must be set wrong.
Bumpty bump. Anyone got any ideas why it fails Blend after 1 hour but not Small FTT's after 8 hours?
I have p5q-e with e8400, the board has no problem with high fsb. It could be your CPU that can't handle the high fsb. I've been on 550fsb for 2years now. If you are stable during small fft but fails during blend, it's definitely your ram or NB. Give your ram and NB way more voltage than the rated spec. I know it sucks but it gives you stability. Also, I know that Lee is like the master of P5Q series boards but I haven't noticed that my board gives me better stability if I use other than 1:1 ratio. Give your NB more than 1.4 to something like 1.5
try to mess with CPU FSB term voltage and the NB GTL... sometimes a lower value works better... plz post pictures of your Bios screens...
Is 1.5v safe on the NB?
Ok, when I've finished my College work I'll try get some pictures of the screen. Thanks for the help guys.
totally unnessecary, you are pumping volts to solve instability. it's more about finesse... correct settings and you can easily go with under 1.4NB... with 1.5 plz actively cool it as it will get too hot and P45 is very sensitive to heat !