Oops... I made a typo it should be "can't". :D
Printable View
The things I like the most about this board are the number of fan connectors, the LCD poster, the IO shield and the layout overall.
Well i found out the volts i can run stable/errorfree 8.5x485fsb @ 1212.5.
My system likes normal skews only. After many many combinations i realized that the most important setting was nbv. In the beginning i started with 1.51nbv and i got a lot of errors and lock ups. I thought it was fsbv, vcore or ramv. I played with skews but nothing. I raise nbv a little @ 1.53 and then @ 1.55 but i keep getting lock ups during memtest without giving any errors before.The reason i raised nbv so little was that @ 8.5x480(only 5mhz lower than 485) i needed 1.49 to have a stable system. I went up @ 1.61v and i run 70-80%memtest without errors before my system freezed again. Finally i managed to find the magic - almost skyhigh - nbv setting...and that is...1.65:eek:. after that i could lower vcore from 1.325 to 1.29..;). Now ill try to lower ramv and see how itll go.
http://img211.imageshack.us/img211/6763/skrmklipp.jpg
http://img211.imageshack.us/img211/9559/skrmklipp2.jpg
Nic one Greg- Today my pc is crashing all the time. same settings as yday both on 8x450 and 9x400 are failing to run prime95 large fft.
i need money now to go and get that new cpu. where is the nearest blod bank?
2 hours later and lots of up and down i have passed large fft with cpu 1.54V real 1.520v NB 1.73 FSB 1.62V
[QUOTE=greg.m;3797497]Well i found out the volts i can run stable/errorfree 8.5x485fsb @ 1213.
My system likes normal skews only. After many many combinations i realized that the most important setting was nbv. In the beginning i started with 1.51nbv and i got a lot of errors and lock ups. I thought it was fsbv, vcore or ramv. I played with skews but nothing. I raised nbv a little @ 1.53 and then @ 1.55 but i kept getting lock ups during memtest without giving any errors before.The reason i raised nbv so little was that @ 8.5x480(only 5mhz lower than 485) i needed 1.49 to have a stable system. I went up @ 1.61v and i run 70-80%memtest without errors before my system freezed again. Finally i managed to find the magic - almost skyhigh - nbv setting...and that is...1.65:eek:. after that i could lower vcore from 1.325 to 1.29..;). Now ill try to lower ramv and see how itll go.
QUOTE]
bios 0803 settings:
Code:Ai Overclock Tuner [Manual]
CPU Ratio Setting [8.5]
FSB Strap to North Bridge [Auto]
FSB Frequency [485MHz]
PCIE Frequency [105MHz]
DRAM Frequency [1216MHz]
DRAM Command Rate [Auto]
DRAM CLK Skew on Channel A/B [Auto]
DRAM TimingControl [Manual]-5-5-5-15
DRAM Static Read Control [Enabled]
Ai Clock Twister [Moderate]
Ai Transaction Booster [Manual]
Common Performance Level [07]
Pull-In of CH A/B all disabled
CPU Voltage [1.29375]
CPU PLL Voltage [1.50V]
North Bridge Voltage [1.65V] - 1.63V real
DRAM Voltage [2.10V]
FSB Termination Voltage [1.46V] - 1.38V real
South Bridge Voltage [Auto]
SB 1.5V Voltage [Auto]
Loadline Calibration [Enabled]
CPU GTL Voltage Reference [0.63X]
NB GTL Voltage Reference [0.67X]
DRAM Controller Voltage REF [Auto]
DRAM Channel A/B Voltage REF [Auto]
CPU Spread Spectrum [Auto]
PCIE Spread Spectrum [Auto]
CPU Clock Skew [Normal]
NB Clock Skew [Normal]
CPU Ratio Setting [8.5]
C1E Support [Disabled]
CPU TM Function [Disabled]
Vanderpool Technology [Disabled]
Execute Disable Bit [Enabled]
Max CPUID Value Limit [Disabled]
Intel(R) SpeedStep (TM) Tech. [Disabled]
CPU Spread Spectrum [Auto]:confused:
PCIE Spread Spectrum [Auto]:confused:
Why aren't these disabled in the BIOS?
Look at this article to know why you should disable them: http://www.anandtech.com/mb/showdoc.aspx?i=3283&p=20
Here is some screenies:
http://img220.imageshack.us/img220/6763/skrmklipp.jpg
http://img220.imageshack.us/img220/9559/skrmklipp2.jpg
and 0803 settings:
as you can see i used a bit high voltages because i wanted to be sure that they wouldnt cause any prob.Code:Ai Overclock Tuner [Manual]
CPU Ratio Setting [8.0]
FSB Strap to North Bridge [Auto]
FSB Frequency [465MHz]
PCIE Frequency [105MHz]
DRAM Frequency [1117MHz]
DRAM Command Rate [Auto]
DRAM CLK Skew on Channel A/B [Auto]
DRAM TimingControl [Manual]-5-5-5-15
DRAM Static Read Control [Enabled]
Ai Clock Twister [Moderate]
Ai Transaction Booster [Manual]
Common Performance Level [07]
Pull-In of CH A/B all disabled
CPU Voltage [1.29375]-> i changed that @ 1.275 and run linx
CPU PLL Voltage [1.50V]
North Bridge Voltage [1.45V] - 1.44V real
DRAM Voltage [2.10V]
FSB Termination Voltage [1.34V] - 1.25V real
South Bridge Voltage [Auto]
SB 1.5V Voltage [Auto]
Loadline Calibration [Enabled]
CPU GTL Voltage Reference [0.63X]
NB GTL Voltage Reference [0.67X]
DRAM Controller Voltage REF [Auto]
DRAM Channel A/B Voltage REF [Auto]
CPU Spread Spectrum [Disabled]
PCIE Spread Spectrum [Disabled]
CPU Clock Skew [Normal]
NB Clock Skew [Normal]
CPU Ratio Setting [8.0]
C1E Support [Disabled]
CPU TM Function [Disabled]
Vanderpool Technology [Disabled]
Execute Disable Bit [Enabled]
Max CPUID Value Limit [Disabled]
Intel(R) SpeedStep (TM) Tech. [Disabled]
hellllllllllllllppppppppppppppppppmeeeeeeeeeeeeeee ee plizzzzzzzzzzzzzzzzzzz
@ghostleader. My blend test crashed after four and a half hours. I used the settings in my first post. It's all right though. I'll try to up the nb voltage to 1.49V from 1.47V and do it again but this time under Mac OS X. Let's see what happens.
____________________
Rampage Formula @0902, 4x2GB G.Skill PC2-8800 (1117 MHz, 5-5-5-15, 2N) at 1.90V (PL7),
Q9650 ES@4.19 GHz (465x9) at 1.4125V (1.38V real, 1.30V load) on water, LLC disabled!,
Asus 4870 TOP 512MB @880/1180 MHz on water, Areca ARC-1210, ATI Theater 550 Pro, Dell 2405 FPW,
PC Power & Cooling Silencer 750W red, Lian-Li PC-V2000 black
I just ran IntelBurnTest 2.0 with the settings you asked but the entire OS (Win 7) froze after like 10 min. OOPS!
____________________
Rampage Formula @0902, 4x2GB G.Skill PC2-8800 (1117 MHz, 5-5-5-15, 2N) at 1.90V (PL7),
Q9650 ES@4.19 GHz (465x9) at 1.4125V (1.38V real, 1.30V load) on water, LLC disabled!,
Asus 4870 TOP 512MB @880/1180 MHz on water, Areca ARC-1210, ATI Theater 550 Pro, Dell 2405 FPW,
PC Power & Cooling Silencer 750W red, Lian-Li PC-V2000 black
If you change PCI-E frequency, you should slightly raise Vsb same time since higher PCIE clock results in higher DMI clock since same clocking source, and believe me you don't want to cause instabilities whether constant or random there. DMI carries all SB inbound traffic, for all devices attached to the root hub or branching off.
Changing PCIE clock on the RF at high FSB is something that must be done to stabilize certain behaviours, such as infrequent DVD-R write errors, unstable/inconistent HDD/DVD transfer speeds on any or all devices connected to SB, etc, hell even fixes some barely audible sound static you get at some particular small frequency bands. anything connected to the NB through PCIE lanes it provides don't suffer these same problems, and changing PCIE frequency fixes them, so it's both the boards insufficient design for the speeds we run and the clocking reference frequencies devices connected to the SB use which use the same synchronous clocking source as the PCIE frequency and while at the same time the SB also has direct clocking reference pins from the CPU which surprise surprise are clocking with the FSB , which when you mix both together on a device that was designed for stepped fixed clock frequencies where there will be never be overlap, overlapping of FSB frequency and either a sub-division or the PCIE frequency itself overlap....you get a lovely result which I've seen some of you guys post regularly and hasn't clicked this might be the culprit or not making it any better at best :)
I'll give you a hint, one is BSOD's with driver filenames, another is display hardlocks, another funny yet rarer one is when randomly a key will get caught in a loop as if it was physically jammed or the mouse will move on its own :D
I'll welcome the day when vendors sell us hardware that is actually properly designed for the broad range of frequency adjustment they give is, and on top of that go out of their way to test for these kinds of problems which as you imagine come up too often, and usually as a result of the freedom to change frequencies with 1mhz steps. It's lovely for end-user and marketing tools to lever, but in practice it's a dead ringer for a darwin award nomination...
"WHY ISNT MY MOTHERBOARD STABLE AT XYZ FREQUENCY, AND WHY DOES AT XX.XX TIME ON YY DAY OF THE MONTH IT CRASH OR BEHAVE WEIRD....WHAT IS WRONG WITH MY SETTINGS........*months later board is RMA'd when its perfectly fine and person has to buy temp board or out of commission the machine until they get a replacement which they RMA immediately because it has the same problem with the first one*....and 9 times out of 10 the problem is simple, some guy said some settings worked great for them, and passed prime95 or linpack...so 100 other guys used them thinking they could avoid learning to do it themselves....maybe 50 guys actually have those settings work, and you can be all 50 have a different error they can't fix, and the first guy who they got them from had it there as well...but due to whatever it never effected him, maybe his PSU or a raid card he ran his drives off or a sound/network pci/pcie card he runs that are designed for maximum compatibility with other hardware, and his onboard stuff is disabled? more reasons than you'd ever want to know, and it's these problems that I see all over forums requesting help, that are fault of the user to a degree, but more fault of the vendor for giving him flexible adjustment with minimal effort/time/knowledge/experience.
DFI are number one when it comes to sink or swim with bios settings. If you don't understand the basics, and you are scared by pages of bios options with no help data, and documentation for what they do available to those who are lucky enough to actually find the damn thing and then translate engineer to english frequently, then don't buy DFI or don't change the settings.
Asus/GB with their one touch overclock, and marketing-gimmick-change-20-registers with our cool new clock twister enhanced feature, it doesn't tell u what it changes hell they even make sure it changes things they give you control over, and when you tell them there is a bug they thats impossible or we can't reproduce it because we didn't expect you to do something as crazy as changing an actual setting that we gave you because we thought you'd never use it, and we didn't mention it either because we are paranoid our competitors will steal our stolen idea, and that is far more important than helping an actual customer solve a problem....lets tell them it's their fault, but we are working on a new beta bios which may fix it if we can find the email I just lost with the information I won't remember. New bios same problem, turns out they didn't fix it because they got carried away adding a new sub menu that doesn't nothing important, but looks cool and can claim bios support, the acutal problem got forgotten because the 1% of ppl who have it only make up 1% sales, the other 99% won't see any changes so they won't believe our claim we make them :) Thats Asus in a nutshell by the way :)
Customer: "Your bios is broken, x doesn't work properly..." Asus Engineer "Try this bios it adds new options" Customer "Ok...." Customer "This bios has same problem, and new features don't work right" Asus "Wait 3 weeks for new beta bios we are working on which will fix up problems" Customer "This new bios I waited for still has the same problem, but new features work right" Asus "we ran 2 tests and we couldnt reproduce it so there is no problem for us to fix...it is probably your cpu or memory or other hardware causing it...replace them at your own expense and if it still happens you can RMA the board under warranty...*while snickering to himself, hopefully by time he does all that he'll give up and we won't have to lose any more sleep worrying about a fix for that issue!*
So, I raised the nb voltage to 1.49V and then I ran that blend test with 7 GB of memory. I used the 64bit version of macprime 259. As you can see, she's still going strong after like 8hr of priming.
http://img.techpowerup.org/090520/8hr prime95-2.jpg
I haven't tried any super pi 32M yet. But I have tried all kinds of other stuff, one of them being able to play Fallout 3 with everything cranked up to the max for long ass hours. Hope you like the pic :)
~talk to ya later
____________________
Rampage Formula @0902, 4x2GB G.Skill PC2-8800 (1117 MHz, 5-5-5-15, 2N) at 1.90V (PL7),
Q9650 ES@4.19 GHz (465x9) at 1.4125V (1.38V real, 1.30V load) on water, LLC disabled!,
Asus 4870 TOP 512MB @880/1180 MHz on water, Areca ARC-1210, ATI Theater 550 Pro, Dell 2405 FPW,
PC Power & Cooling Silencer 750W red, Lian-Li PC-V2000 black
Not quite, a good value is whatever frequency you find you need for the system to behave right. Personally I've used upto 116mhz, with Vsb at 1.125v.
I always found at 481FSB with mine 110-112mhz was the sweet spot where everything worked right, and ICH9R raid0 performance was most consistent. I use HD Tune Pro to graph the Raid0 performance for my array at each 1mhz step, and concluded that 112mhz was the frequency where performance finished increasing. Any higher resulted in the SB choking from bus limitations, basically like it would spike high then suddenly drop off rapidly until the SB could catch up with the amount of data on the DMI link, then it would spike high again and repeat. In my case I think I found the SB's limit.
@greg.m
Nice work keep it up :up: :clap:
@Chemjsal
Very good work, nice indeed, keep it up :up: :clap:
Have not test much with the new RF bios, having some fun with my Rampage Extreme and a Q9650 with 4x1 ram @ 900MHz (1800 effective) @ CAS 7 right now
@mikeyakame
Very nice info about PCI-E frequency, I have learn a lot from you by reading your posts, keep up the good work, your´re the man. :up: :clap:
Pfff... It's a waste of time again.:down:
I don't know what to do anymore.:shrug:
Everything runs perfectly smooth but it can't stay stable. I need a BIOS that let's me fine tune the settings.
I tried with CPU and NB Clock Skew Normal and ended up with a crash again.
My best runs are with CPU Clock Skew Delay 200ps and NB Clock Skew Delay 100ps but I think that the 100ps difference between the CPU and NB Clock Skew is to large. That's what I'm thinking is causing these stability problems.
I'm so disappointed.:(
I bought this G.SKILL F2-9600CL5D-4GBPI memory to have better clocks and instead of having better clocks I'm running with lower clocks.
There's one thing that I learned from all of this. This is probably the last ASUS hardware that I bought because I don't think that there ever will be a decent BIOS for my current setup.