Let me just say that trying to debug bios settings that cause your computer to crash prime95 after an hour SUX!Quote:
Originally Posted by jwfitt
Let me just say that trying to debug bios settings that cause your computer to crash prime95 after an hour SUX!Quote:
Originally Posted by jwfitt
The funny thing is that on my QX9650, when I'm fine tuning the OC, it's always Core #4 that fails and, amazingly enough, Cores #3 and #4 always reports the lowest temps by 1 or 2 degs.
I guess the temps could be off... but still odd that the supposedly coolest core(s) would fail and not the hotter ones.
Quote:
Originally Posted by dimsdale
Hahaha! No kidding!! I'm glad there's an emoticon that sums up my feelings...
:horse:
But yeah running blend and having to wait an hour every time you run a test...ugh.
I'll be more than happy at 3.8 if I can get that to work. The real puzzler to me is that 4ghz passes small fft testing with no issues. The manual RAM settings are the area I am least comfortable with, but I've gotta think 4 ghz can be stabilized if the right settings were applied there... sigh...
Anyone know if the FSB Strap setting actually affects performance? Is running the Strap at 400 MHz better than running the strap at 333 MHz?
all other things equal, the lower strap will increase performance, but it is hard to get this board stable at high fsb clocks on the 266 strap. 333 seems to be the best of both worlds and the 400 strap seems to be the easiest to stabilize.
Not sure if you've seen this, but damn...
http://www.hardforum.com/showthread.php?t=1138241
4ghz stable at 1.3625 volts on air?? Must be one lucky chip! i see a few others with similar g.skill ram at 4ghz too...
Here are my currently testing settings for FSB 450 / CPU 3.825 GHz / DDR2 1081 on a QX9650. (Update: currently 45 mins prime95 blend stable). Currently I'm seeing load temps of 70, 70, 69, 69.
I think the problem getting higher CPU clocks on the Maximus Formula and at least the QX9650 might be the FSB Term Voltage. In connection with the P5E3, Anandtech said "FSB Termination Voltage: Maximum of 1.50V. Undeniably the Achilles' heel of this fine board, 1.5V (VTT) is simply not enough to carry most 65nm quad-cores above ~500MHz FSB. Most CPUs/MCHs require this voltage to quickly ramp up from the default (1.20V) to near maximum when overclocking from about 450MHz FSB and higher." (See http://www.anandtech.com/cpuchipsets...spx?i=3156&p=6)
Extreme Tweaker
Ai Overclock Tuner : Manual
Ratio CMOS Setting : 8.5
FSB Frequency : 450
FSB Strap to North Bridge : 333
PCI-E Frequency: 100
DRAM Frequency: DDR2-1081
DRAM Command Rate : 2T
DRAM Timing Control: Auto
DRAM Static Read Control: Enabled
Ai Clock Twister : Moderate
Transaction Booster : Diabled
Relax: 0
CPU Voltage : 1.36875 (1.3520 in Everest) [UPDATED]
CPU PLL Voltage : Auto
North Bridge Voltage : Auto (1.6160 in Everest)
DRAM Voltage : 2.04 (2.0960 in Everest)
FSB Termination Voltage : 1.50 (1.4240 in Everest)
South Bridge Voltage : Auto
Loadline Calibration : enabled
CPU GTL Reference : .63
North Bridge GTL Reference : .67
DDR2 Channel A REF Voltage : Auto
DDR2 Channel B REF Voltage : Auto
DDR2 Controller REF Voltage : Auto
SB 1.5V Voltage : Auto
CPU Spread Spectrum : Auto
PCIE Spread Spectrum : auto
________________________________
QX9650
Asus Maximus Formula (1004)
G.SKILL 8GB (4 x 2GB) DDR2-1066
Thermaltake eXtreme 120+AS5+Scythe
WD 150GB Raptors in RAID 0
EVGA 8800 GT
My current ~air settings 24/7 on full load (Rampage 0219@bios)
http://www.imageshack.gr/files/maai8nmbpa6ibccy4t2x.jpg
LMFAO did you see how the Nvidia chipset needed 1.68v to clock 4ghz.Quote:
Originally Posted by jwfitt
OUCH !
Hmm.... If you are passing small fft at 4ghz, but not blend, that (supposedly) suggests it's more of a RAM problem than a CPU problem directly, right?Quote:
Originally Posted by jwfitt
If so, maybe it is your FSB Termination Voltage then. You are on AUTO for that, right? What is the bios (and everest as "FSB VTT", I think) reporting as your voltage there under the AUTO setting?
I thought that's what it suggested, since small fft was focused on testing your cpu and blend seems more like a RAm and overall system test. Most people I've seen seem to tweak their cpu settings first (using fft as feedback), then move onto RAM and blend mode...Quote:
Originally Posted by dimsdale
Hmm...well at the time of success with small FFTs I was manually set at 1.4 in the BIOS for the FSB VTT, not sure what everest reported it as.Quote:
Originally Posted by dimsdale
I never did try pumping the FSB VTT up into the yellow numbers. Maybe I'll give that a go when I get home in 2 hours...weee!
so you CAN flash back to MF bios? I just need some confirmation
there seems to be some mixed reports over the internet and the search feature doesn't seem to work for me..
Yes you can flash back, already did................
Why? The RF bios is far better than any of the MF bios'...far more stable.Quote:
Originally Posted by nordien
To be sure the flashback can be done !Quote:
Originally Posted by theonlybabyface
its been confirmed many times over :wiggle:Quote:
Originally Posted by nordien
You guys with the QX9650, are you shooting for 4GHz only on highest multi and then using a divider to make results? Cause if a 1.5yo Q6600 can do 450~505 on air on even my P35, I find it odd that a new Penryn cant keep up if not pass it by by just a tad. I use the 8x on just about all my systems even if they offer 1 or 2 higher.
What happened to overclocking the old reliable way, where we check for max clock v/s max FSB but isolated from the rest of the system? are you guys disabling ALL performance options related to chipset and RAM? Do you keep RAM in 1:1 w/ relaxed timings and adiquite volts to insure stability?
Keep in mind when aiming for FSB, it is not just the CPU, but the mobo as well. creative use of FSBT, GTL and PLL can all play an important role in higher CPU FSB improvements. On our mobo, Static Read Control helps smooth out the curve on higher FSB climbs. This mobo tested better then most in this area since a lot dont offer this setting.
What is the box labeld max for each QX9650 and Q9450 as opposed to Intel MAX?
My QX9650 is not the greatest. I started with loose RAM and lots of AUTO and went after the FSB.
I've got it at 421x9.5 right now for 4GHz and I left it priming this morning, so we'll see how it does when I get home.
I'd like to run higher FSB but it doesn't seem to like to go much above 450. I'll post my settings later when I get home.
yo mates. i know this is premature but i am just exhilarated that i got this far that i just got to tell u guys. based on the input from nuckin futs and so many other personal from this site, i have just got to 3.96GHz. going to start Prime 95 Blend test shortly. If everything is ok, will post more info tomorrow. for now find the screenie and bios settings.
btw, everyone, please comment if my volts are acceptable based on the screenie coz this IS my weak point. also please comment my bios settings in case i crash. ;)
alright, here are some updates with readings from BIOS
please take note that the figures in bold are actual readings from BIOSPHP Code:Ai Overclock: [Manual]
OC From CPU Level Up: [Auto]
CPU Ratio Control: [Manual]
Ratio CMOS Setting: [9]
FSB Frequency: [440]
FSB Strap to North Bridge: [400]
PCI-E Frequency: [118]
DRAM Frequency: [1173]
DRAM Command Rate: [2T]
DRAM Timing Control: [Manual]
Primary Info : 5-5-5-15-3-42-6-3
Secondary Info : 8-3-5-4-6-4-6
DRAM Static Read Control: [Enabled]
Ai Clock Twister: [Light]
Transaction Booster: [Disabled]
Relax Level [0]
CPU Voltage------------------1.475 [[B]1.456v[/B]]
CPU PLL Voltage ------------- 1.68 [[B]1.792-1.808v[/B]]
North Bridge Voltage -------- 1.51 [[B]1.552v[/B]]
DRAM Voltage ---------------- 2.20 [[B]2.304-2.320v[/B]]
FSB Termination Voltage ----- 1.48 [[B]1.552v[/B]]
South Bridge Voltage -------- 1.100 [[B]1.120v[/B]]
Loadline Calibration -------- [Enabled]
CPU GTL Refference ---------- [0.63%]
North Bridge GTL Reference -- [0.67%]
DDR2 Channel A REF Voltage -- [Auto]
DDR2 Channel B REF Voltage -- [Auto]
DDR2 Controller REF Voltage - AUTO [[B]1.152v[/B]]
SB 1.5v Voltage ------------- 1.55 [[B]1.616v[/B]]
CPU + PCI-E Spread Spectrum --------- [Disabled]
I'd be perfectly happy at just 400 FSB with my 9650, I've gotten to 450 on the FSB and stayed fine on small FFT testing. It all comes crashing down when I get into Blend testing though.
However, I've just bumped a few things up to see if blend would stabilize at 10x400, and it's looking good. Well over an hour going at the following settings:
Gonna start scaling back some settings and see where it goes...Code:Extreme Tweaker
Ai Overclock Tuner : Manual
Ratio CMOS Setting : 10
FSB Frequency : 400
FSB Strap to North Bridge : 333
PCI-E Frequency: 100
DRAM Frequency: DDR2-801
DRAM Command Rate : 2T
DRAM Timing Control: Auto
DRAM Static Read Control: Disabled
Ai Clock Twister : Moderate
Transaction Booster : Disabled
Relax: 7
CPU Voltage : 1.4875
CPU PLL Voltage : 1.60
North Bridge Voltage : 1.67
DRAM Voltage : 2.00
FSB Termination Voltage : 1.70
South Bridge Voltage : Auto
Loadline Calibration : Enabled
CPU GTL Reference : .63x
North Bridge GTL Reference : .67x
DDR2 Channel A REF Voltage : Auto
DDR2 Channel B REF Voltage : Auto
DDR2 Controller REF Voltage : DDR2_REF
SB 1.5V Voltage : Auto
CPU Spread Spectrum : Disabled
PCIE Spread Spectrum : Disabled
last update for the night before bed. system boots well into windows without any glitch. however, prime95 is no go. fatal error the moment the the blend test is run.
anyone with ideas on this? i am guessing perhaps there is shortage of Vcore here but hope someone can give me some ideas more. will try again tomorrow after work.
@ cyrixMII300, Great job on that so far.
What happened to all those RAM tweakers on here? Never even made a comment on my RAM sub timings POST a week ago. I thought this had become a heavy RAM sub timing tweakers tread. Oh well!
So any other RAM tweakers left here? I was wondering if any of you coming from previous single phase DDR voltages have noticed an OC performance either in stability or temps w/ same 4 stick configurations when moving to this w/ new 2 phase DDR? Hows overclocking on the 2x2g looking for top end OCZ offerings compared to the 2x1g? Is 2x2g @ DDR1200 realistic for everyday use and good temps?
I haven't been able to get my 2x2 sticks to even DDR1100.
I've been through ReaperX, G.Skill and now these Mushkins (all PC28000)
I dont run 4GB so I cannot say, Dominators 2x1GB PC8500 will do 1275 on this board with a little extra coolingQuote:
Originally Posted by Nuckin_Futs
While the timing is rather relaxed, A high FSB will compensate for any bandwidth loss, this board really likes 500FSB and 1200+ ram
DRAM Command Rate - 2N
Dram Timing Control - Manual
*In Brackets are timings that can be optimized later, outside of the brackets is a good starting point*
CAS# Latency - 5
RAS# to CAS# Delay - 5
RAS# PRE Time - 5
RAS# ACT time - 18
RAS# to RAS# Delay - 4(3-4)
Ref Cycle Time - 45(25-55)
Write Recovery Time - 6(3-6)
READ to PRE Time - 4(2-5)
READ to Write Delay (S/D) - 9 (7-10)
WRITE to READ Delay (S) - 5 (3-6)
WRITE to READ Delay (D) - 5 (4-6)
READ to READ Delay (S) - 4 (3-5)
READ to READ Delay (S) - 6 (5-7)
WRITE to WRITE Delay (S) - 4 (3-5)
WRITE to WRITE Delay (D) - 6 (5-7)
DRAM Static Read Control - Disabled (Enabled)
Ai Clock Twister - Light (Strong)
Transaction Booster - Disabled
Relax Level - 2 (0-3)
A few numbers changed for improved bandwidth but this setup got my Dominators stable at 1274MHz 3 hours OCCT and still running
jwfitt,
CPU Voltage at 1.4875 for 4ghz with a Thermalright eXtreme 120!
What kind of temps is that generating??