overclocking x48t-dq6 q9550

[Classicman]

Yes, that's my RAM. Are you sure about the overvolting, because it provides guarantee up to 1.95V with stability and mines work at 1.96. I'll try the small FFT moder after the test i've been doing is done. I've modified my first bios setting into the following. It's still performing the prime95 test for one hour and no error came up. Let me give you the bios settings below.

[Classicman]

Robust Graphics Booster___________ [Turbo]
CPU Clock Ratio ____________ [8]
Fine Clock Ratio ____________[0.5]
CPU Host Clock Control_ [Enabled]
CPU Host Frequency (MHz)__________ [412]
PCI Express Frequency (Mhz)_______ [100]
C.I.A. 2__________________________ [Disabled]
Performance Enhance________________[Standart]
System Memory Multiplier____(SPD)__[4.00D]
Memory Frequency (Mhz)_____________1648
DRAM Timing Selectable_______ SPD __ [Manual]

CAS Latency Time________________ 8
Dram RAS# to CAS# Delay_________ 8
Dram RAS# Precharge Delay_______8
Precharge Delay (tRAS)__________ 24

ACT to ACT Delay (tRRD)_________ 4
Rank Write to READ Delay________ 8
Write to Precharge Delay_________ 12
Refresh to ACT Delay______________ 60
Read to Precharge Delay__________ 6
Static tRead Value_______________8
Static tRead Phase Adjust________ [Auto]
Command Rate(CMD)_________________[2T]

CPU/PCIEX Clock Driving Control [800mV]
CPU Clock Skew Control [normal]
GMCH Clock Skew Control [Normal]

System Voltage Control____ [Manual]
DDR3 OverVoltage Control__ [+0.45V]
PCI-E OverVoltage Control_ [+0.10V]
FSB OverVoltage Control___ [+0.35V]
(G)MCH OverVoltage Control [Normal]
Loadline Calibration_______[Enable]
CPU Voltage Control_______ [1.350]

[Nickel020]

Most boards overvolt a bit compared to the value the voltage is actually set to, so you may be at over 2.0V. Also i remember reading somethin that right at the time you press the power button the voltages might be a little higher than they're supposed to be overcolting the memory a little more.
But this was basically just a guess, something to try that i thought might have helped.

What did you change exactly to get Prime to run again?

[Classicman]

Your information will help, thank you, because i reduced many voltages. If you're asking what i changed for the 3.7Ghz system, i changed nothing, but strange thing it did give error. That's why i feel insecure about the system:). Now I reduced many voltages and system is at 3.5ghz, stable this ****ing time. I think you're right about the RAM thing, because small fft doesn't give error. That's also why i sent an e-mail to a guy who is said to be very good with OCZ rams. He is a tech in OCZ i guess whose mail is eryder@ocztechnology.com

[Nickel020]

Yeah, Ryder knows his stuff, he manages the OCZ forum but also hangs out a Xtreme Systems quite a bit.
He can help you much better than I can witht he RAM issue, he knows the stuff, for me it's still a lot of trial and error (like it is every normal person who doesn't work in that field I guess). I also have no experience with DDR3 and although the timings themselves are the same they need different values that I can't tell you.

GL with that and let us know how it goes!

[Nickel020]

Glad to hear it works! Motherboards sometimes have problems that are peculiar problems where a certain FSB will not run well but a higher one will (as is the case with your board). Well there's mothing like actually having first hand experience with the board.

If you try to go higher you may need a higher VCore and maybe the D multiplier will work better again than the B one, try both to be sure.

[Classicman]

There is a strange thing with B and D multiplier in my mobo. Although it says B is 333 and D is 400, both of them shows the same memory frequency. Could that be a possible BIOS bug? I hope giga-byte fixes it by F6 (still in beta). First i thought to lower the memory multiplier; but as long as 4.00B and 4.00D shows the same, i couldn't do it, because the other multipliers are too low, in order to raise the memory frequency to 1600mhz, i had to clock the cpu by more than 460 and that, as you can guess was ****ty.

[Nickel020]

B and D should show the same frequency. Think of it as something like memory timings for the northbridge itself.
Try 3.2B/D though and see if you can overclock the CPU further at settings that you're comfortable with. If you try to do it without lowering the multi you won't know whether it's the RAM or your CPU settings holding you back.
Also try lowering the CPU multiplicator to 6 or 7 and stay with the 4.0B/D multiplier and increase your FSB to see how high your RAM can go.

I wouldn't worry though about using a lower multi for the RAM though since it doesn't actually provide any noticeably performance difference. Here's an explanation I worte for someone else:

Now the bandwith of the RAM to NB is the same if the 2.0 system memory multiplier is used, i.e. DDR2 667 for a 333 (1333 quad-pumped) FSB CPU, DDR2 800 for a 400 FSB CPU. If you increase the RAM speed without increasing the FSB, data is transferred faster to the NB from RAM, but the NB can't pass it on any faster to the CPU than before though resulting in no performance benefit (there's a litte benefit actually, as latency is improved if RAM speed goes up, but it's small ~5% or so).
If you're not overclocking get DDR2 800 RAM, if you plan on overclocking in the future get 1066 DDR2 (runs at 533 MHz so you can increase your FSB to 533 (2133 quad pumped) without having to overrclock the RAM). The reason why faster DDR2 RAM exists is solely for overclocking purposes, as slower will limit the maximum overclock of your CPU.

[Nickel020]

Looks good, all except FSB overvoltage maybe. I have not epxperience running +0.3V 24/7 on a 45nm quad and I don't know if it's a good idea.
Head over here and ask those folks, they should be able to help you:
Gigabyte X48T DQ6 Living Review - Clunk's Forums.
They can also help you much better with finetuning your settings than I possibly could.

[Classicman]

Nickel020, i have good news man. I have adjusted the voltages again, exclusively considered the fsb voltage you mentioned and established a 4 hour 17 minutes prime95 blend test stability with no crash. After that, one of the workers failed. My trials showed me that, my cpu can't handle 3.65GHz less than 1.3125V, yet there are people reach higher clocks. if the VVt is less, the stability reduces to less than one hour. Here are the settings:

Robust Graphics Booster__________[Turbo]
CPU Clock Ratio _________________[8]
Fine Clock Ratio ________________[0.5]
CPU Frequency____________________3.65GHz(430x8.5)
CPU Host Clock Control___________[Enabled]
CPU Host Frequency (MHz)_________[430]
PCI Express Frequency (Mhz)______[100]
C.I.A. 2_________________________[Disabled]
Performance Enhance______________[Standart]
System Memory Multiplier_(SPD)___[4.00D]
Memory Frequency (Mhz)___________1720
DRAM Timing Selectable___(SPD)___[Manual]

CAS Latency Time_________________8
Dram RAS# to CAS# Delay__________8
Dram RAS# Precharge Delay________8
Precharge Delay (tRAS)___________24

ACT to ACT Delay (tRRD)__________4
Rank Write to READ Delay_________8
Write to Precharge Delay_________12
Refresh to ACT Delay_____________72
Read to Precharge Delay__________6
Static tRead Value_______________8
Static tRead Phase Adjust________[Auto]
Command Rate(CMD)________________[2T]

CPU/PCIEX Clock Driving Control [800mV]
CPU Clock Skew Control [Normal]
GMCH Clock Skew Control [Normal]

System Voltage Control___________[Manual]
DDR3 OverVoltage Control_________[+0.40V]
PCI-E OverVoltage Control________[+0.10V]
FSB OverVoltage Control__________[+0.20V]
(G)MCH OverVoltage Control_______[+0.20V]
CPU GTLREF1 Volrage Ratio________[Auto]
CPU GTLREF2 Voltage Ratio________[Normal]
Loadline Calibration_____________[Enable]
CPU Voltage Control______________[1.3125]

I really don't have the desire to stabilize it more as i spend enough time for this i guess.