Printable View
The reason that I'm thinking that it's the NB Voltage is because it fails Prime95 Blend in Ubuntu 64 bit with the updated option ROMs and it doesn't do that with the original one but the original one gives me a BSOD in Windows 7 Ultimate X64.Quote:
Originally Posted by Trust13
Over 4 hours Prime95 Blend stable with Wine in Ubuntu 64 bit. :)
http://i431.photobucket.com/albums/q...ODBIOS0001.jpg
So it was vNB then??
Ubuntu is happy with vNB 1.28V and the MOD BIOS 0001. I'm going to try Windows 7 Ultimate X64 and hope I can pass the 4 hours also without problems.Quote:
Originally Posted by leo27
It looks like the updated option ROMs are going to fix the BSOD. No need to bump the vCore higher than necessary to make it go away. That was the solution to fix the MACHINE_CHECK_EXCEPTION BSOD in Windows XP X64 and it would probably be the same with the original 0001 BIOS and Windows 7 Ultimate X64 to fix the 0x00000124 BSOD.
Maybe now with the updated option ROMs it's going to be possible to use lower vCore. :shrug:
Succesfully passed the 4 hours with MOD BIOS 0001.
Forget about lower vCore. Windows 7 Ultimate X64 needs more. :D
http://i431.photobucket.com/albums/q...RD6-tREF-7.jpg
Oh you got my attention, as I am running Win7 X64 as well.
so A-Grey did any other voltages need to be raised other than vCore??
No, only the NB Voltage and CPU Voltage.Quote:
Originally Posted by leo27
I already knew that it would need a little more NB Voltage than with with the original 0001 BIOS when I saw the change in the DRAM Clock Fine Delays.
The CPU Voltage also needed a little bit more to push it over the 4 hours without making an error in Prime95.
My BIOS settings:
Code:Ai Overclock Tuner [Manual]
CPU Ratio Setting [9.0]
FSB Strap to North Bridge [400MHz]
FSB Frequency [450MHz]
PCIE Frequency [100MHz]
DRAM Frequency [1199MHz]
DRAM Command Rate [2N]
DRAM CLK Skew on Channel A/B [Auto]
DRAM Timing Control [Manual]
Refresh Period [16120]
DRAM Static Read Control [Disabled]
Ai Clock Twister [Lighter]
Ai Transaction Booster [Manual]
Common Performance Level [06]
Pull-In of CH A/B all disabled
CPU Voltage [1.31250V]
CPU PLL Voltage [1.50V]
North Bridge Voltage [1.31V]
DRAM Voltage [1.80V]
FSB Termination Voltage [1.20V]
South Bridge Voltage [1.05V]
SB 1.5V Voltage [1.50V]
Loadline Calibration [Disabled]
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 [Delay 300ps]
NB Clock Skew [Delay 200ps]
Any chance you could post your 0001 modded bios? :)
http://solor.windowsdream.com/Asus.R...mula/BIOS/MOD/Quote:
Originally Posted by Darkstormz
1002bMOD = 0001 MOD, just changed some strings to get some version consistancy and IF 0001 ever becomes offical im sure it will be in 11XX range because of new feature. So calling this 1002 does not hurt anything.
ahh thanks alot :up:Quote:
Originally Posted by SoLoR
@SoLoR
Euhmm... did you just fix the most annoying bug in the BIOS for ASUS. It's still early to call it 100% stable but I'm making progress being MemTest For Windows stable. :rofl:
http://i431.photobucket.com/albums/q...D6-tREF161.jpg
if i did something i did not do intentionaly.... i can try to break it again ;)Quote:
Originally Posted by Alien Grey
I think it's going to be possible to have it stable running at 9 X 470MHz and DDR 1253MHz with tRD 6 and tREF 16120T.
SoLoR fixed it for us without knowing it. :rofl: We really need the updated option ROMs. This is the key to have it running stable with DDR +1200MHz. If ASUS took the time to try them out they would have found this a long time a go but they didn't so they couldn't fix it for us.
An other thing that I've found thanks to Trust13 and OLDcomer is to use CPU GTL Voltage Reference 0.67X when using Ai Clock Twister Lighter. It isn't the FSB termination Voltage that let you run it stable with lower CPU Voltage it's the CPU GTL Voltage Reference. At 0.63X it was impossible to have the CPU stable at 4.23GHz and using more FSB Termination Voltage and CPU Voltage couldn't make it more stable. At 4.23GHz it likes to run with low FSB Termination Voltage and CPU GTL Voltage Reference 0.67X with lower CPU Voltage than I was expecting to have to use.
The downside is that you need more NB Voltage to run it stable with the updated option ROMs but I don't think that this is a big problem if we can run stable at higher speeds. :yepp:
The higher NB voltage is probably because the DRAM Clock Fine Delays increase with 1T with the higher NB Voltage. Maybe delaying the DRAM CLK Skew on Channel A/B is going to fix that but I didn't try it yet. It has to run stable before I can try that.
sounds like good progress, i wish we new the import of newer option roms months ago though :D
Well... im kinda moding bios since i learned how to do it more then a year ago around bios 0803. I however dont want to promote them, because i dont want to be responsible if something breaks and/or dont work... Until i break only my mobo because of bad bios flash (happened once), its ok. If i would break someone elses, then it wouldnt be that ok :) Also main issue was i never tought it could improve OC ability, because honestly it doesnt do anything for me... im "locked" on 8x451mhz (3608) and cant get even mhz more 100% stable (stoped trying last summer and just accepted this fact), why... beats me, but i guess its just bad cpu.Quote:
Originally Posted by leo27
There's definitely progress. It's like a thought you've got to delay the DRAM CLK Skew on Channel A/B to lower the NB Voltage and stabilize the board at these speeds. :up:
SoLoR do not know if previous 6590 or original 6685 had the same problem:Quote:
Originally Posted by Spyrus
When moving scroll bar of Firefox i see the CPU Usage to increase for firefox.exe.
That is ok, have seen it also on my laptop,
but i can see also some crackling noise coming from a game
that is running in background (alt+tab in firefox while playing and
immediately start scrolling).
Can you confirm this behaviour? I would try with Media player for example
and see if there occurs too, but i am at work again (all thoughts come up here :rofl: )
didnt notice anything like this... this might be dpc latency issues people randomly reports while playing sound... For example you have some sound playing (in your case game) and then you use some program that causes cpu usage spikes (using lots of interrupts) and sound starts to crackle. This sound crackling issue is (almost) always because of dpc latency spikes and cant be caused by lots of things like drivers, applications etc etc. I personaly have absolutly no issues with onboard soundcard, but its also true i dont play games much, only BC2 and WoW once in a while but didnt notice anything... Im still wating on some new drivers... slack soundmax and drivers form a year ago :< i couldnt make THX TruSound work with it yet as well, but in all honesty i gave up quickly...Quote:
Originally Posted by Spyrus
'dpc latency spikes' i googled it to investigate further. I will do some test and report back.
Edit: Used the dpc checker ... no spikes found.
I noticed, that some of you(especially Alien Grey) are using CPU/NB Skews with 100ps difference, but with values like 300ps/200ps. Since 100ps/none was the key reason to be able to do more than 460FSB, does higher value skews, but with 100ps difference, do any better compared to 100ps/none?
Yes but it depends on what voltage you need to keep it stable.Quote:
Originally Posted by spoof
CPU Clock Skew Delay 300ps and NB Clock Skew Delay 200ps for really low voltage. NB Voltage of 1.25V to +/-1.35V.
CPU Clock Skew Delay 200ps and NB Clock Skew Delay 100ps for NB Voltage +/-1.35V to +/-1.45V.
It isn't always a voltage issue like what I'm experiencing now trying to run stable at 9 X 470MHz and DDR 1253MHz. I've tried really hard but it keeps failing Prime95 Blend. The longest I can keep it stable is an hour so now I'm going to try to find out how much lower I have to go to keep it stable.
I think I'm not going to try to run LinX for a very long time. It doesn't make sense at high CPU speed.
I'm trying to run LinX for 25 passes stable at 4.22GHz and everything is just fine but it's BSOD 0X00000124 again so I add more CPU Voltage and everything is fine but at some point it's BSOD 0X00000124 again so I add more CPU voltage and more and more... What the :banana::banana::banana::banana: is wrong. :mad:
But then I realize that if you add more voltage you have more heat so the water temperature starts to climb up and the CPU becomes unstable and needs more voltage but then the temperature starts to climb again and the CPU becomes unstable again. Whatever I do it's always BSOD 0X00000124 after some time. :confused:
So I'm thinking that running LinX for a very long time at very high CPU speed isn't a good stability check and now I'm running Prime95 Small FFT's again to check the stability of the CPU.
What do you guys think of that?
Well i always did only 15-20 passes of lynx for "quick" check and then i was runing prime for hours (15+) if it passed that i considered it stable :) Runing lynx for very long time is also probably not healthy for CPU :) for me it reaches INSANE temperatures (~85-90 so maximum of maximum that its still considered "safe").
and on another note, i updated x-fi software pack today. You dont really need to update it, since it doesnt change any functionality but it does have updated host openal and audiocp... but you might want to download it to install it on next format :) Also i still didnt have any luck with geting EAX5 to work :(
Thanks I'll try the new X-Fi pack later. I'm trying to get my Q9650 stable at 4.22GHz with DDR 1250MHz. :D