Official GIGABYTE X38-DQ6/X38T-DQ6 Info/BIOS Thread

[GAR]

loadline calibration been fixed?? and where can i get a newer bios, i curently have f7d

[2DIFF]

and another very good PSU
http://www.silverstonetek.com/produc...a1200&area=usa

I would buy one of this babyes but unfortunately I couldn't find in my country ...

How about this one, could get if for 581 SFR: http://www.1000ordi.ch/docs/30161_en.pdf

Or this one, cheaper, 499 SFR: http://www.performance-pcs.com/catal...ducts_id=21124

[Caped Crusader]

loadline calibration been fixed?? and where can i get a newer bios, i curently have f7d

not yet available, I guess...

[Caped Crusader]

How about this one, could get if for 581 SFR: http://www.1000ordi.ch/docs/30161_en.pdf

Or this one, cheaper, 499 SFR: http://www.performance-pcs.com/catal...ducts_id=21124

don't waste good money on that kind of PSUs unless you're gonna put 3 cards inside the case and half a dozen drives and a truckload of USB devices. One 750W unit is more than enough, and cheaper!!

[2DIFF]

don't waste good money on that kind of PSUs unless you're gonna put 3 cards inside the case and half a dozen drives and a truckload of USB devices. One 750W unit is more than enough, and cheaper!!

Don't worry... was just about that single 12V rail up to 90Amp. Question to you please: do you get your HD3870 clocks with the original bios, and how do you cool the card? thanks

[Caped Crusader]

Don't worry... was just about that single 12V rail up to 90Amp. Question to you please: do you get your HD3870 clocks with the original bios, and how do you cool the card? thanks

ok, I get it I have 60A single-rail. You think I'll have problems feeding one of the newest Quads? don't think so

Anyway, about the 3870. Stock clocks are 800/1200 since it's an Extreme/PCS version with Zerotherm cooler thing (this one: http://www.powercolor.com/Global/pro...roductID=1632). It says 1170Mhz for the memory clock but mine's at 1200... don't ask me why...
The zerotherm solution is quite effective. I got 30ºC idle on vista 64 desktop and 45ºC full load.

I didn't flashed the unlocked BIOS, so the vcore is locked at 864Mhz.

[Caped Crusader]

I realised that my sig was not accurate about the 3870 clocks. Fixed now.

[Richard Dower]

F7 surface yet?...if y'all don't know...there is a new version of Prime95 available, 26.6 or something.

[ToyTen]

so i changed over to X38-DQ6 from DS3P on my Single Stage setup and thought i might as well flash to latest bios
so from F3 to F4

my conclusion is DO NOT USE F4 bios if you have this board!!!!!!!!!!!!

F4 is ****ed i could not go past 445Mhz FSB no matter what and even that wasn't too crash hot stability wise >> i've spent a good 2.5 hours trying to figure it all out and since i have a fair bit of experience from the last few months of testing with Gigabyte boards i was a bit surprised that i could not figure out WTF was going on

after trying to figure it all out i decided to flash back to F3 bios

and now all is fine



DO NOT FLASH TO F4!!!

it's a shame because F4 has some excellent voltage and RAM strap change options

Hi
Am I concerned if I have a Q6600 and the X38T-DQ6 rev1.1 ?

Thx

[Hornet331]

the X38T is a whole other ballpark as the X38 (ddr3 vs ddr2)

The ddr2 version gets a lot more updates then the ddr3 version...

[ToyTen]

oki, thx Hornet331, I'll see a bit later if its necessary or not to flash it...
What is to think about 480 fsb on air (9500led) with that Q6600? I have 518(same air) with an e6600, but it's maxscreen.

Something else, maybe you all noticed: I try to find the right spot to pencil-mod my mobo 'cuz it has such a Vdroop !! 0.08.... too bad :/

[Hornet331]

Found what is the problem that we have to increase the cpu voltage ...
This mobos take the CPU current from all PSU rails 3,3V, +5V and +12 V.
The power neded to overclock the CPU is more then 135 W. So the rails must give minimum 100 A ( I=P/U= 135/100=1,35V (The CPU voltage ), otherwise we have to increase the voltage to compensate the current neded by CPU .

just plain wrong...

the only voltages that a mobo supply to the cpu are 12V and 3.3V, but the 3.3V is only needed during the power up of the processor...

so the only relevant voltage/rail for the cpu is 12V.

if you dont belive me:

VRD 11:
http://download.intel.com/design/pro...s/31321402.pdf

oki, thx Hornet331, I'll see a bit later if its necessary or not to flash it...
What is to think about 480 fsb on air (9500led) with that Q6600? I have 518(same air) with an e6600, but it's maxscreen.

Something else, maybe you all noticed: I try to find the right spot to pencil-mod my mobo 'cuz it has such a Vdroop !! 0.08.... too bad :/

480 should be reachabel (450 is possible with no voltage increase for MCH or fsb)

yeah i also dislike the vdroop on this board, theres an option in the bios that would solve the problem if it would work (loadline calibration). I hope the next bios provides a working solution.

[ChrissTi]

3.3V not enough current => cold boot problems when overclock ... 3.3 also for memo + and mobo. Not all mobos are made by Intel specs ... Intel fanboy . Also you should know that on some mainboards +5V also powers the CPU.
Intel sucks at mobos, once as I remember I had a lot of intel 965 mobos all with lots of problems, had to RMA the whole lot.
I think that X38-DQ6 has a different design and power managemant than Intel's X38 mobos specifications.
On this mobo +12 V rail is supplementary for the CPU, to achive the required power for the extreme CPU's, so the main power for the CPU is still +3,3 and +5V (I do not know the percentage). Read the manual if you don't belive me.

[Hornet331]

3.3V not enough current => cold boot problems when overclock ... 3.3 also for memo + and mobo. Not all mobos are made by Intel specs ... Intel fanboy . Also you should know that on some mainboards +5V also powers the CPU.
Intel sucks at mobos, once as I remember I had a lot of intel 965 mobos all with lots of problems, had to RMA the whole lot.
I think that X38-DQ6 has a different design and power managemant than Intel's X38 mobos specifications.
On this mobo +12 V rail is supplementary for the CPU, to achive the required power for the extreme CPU's, so the main power for the CPU is still +3,3 and +5V (I do not know the percentage). Read the manual if you don't belive me.

wtf... do even you know that the VRD 11.x specification is?
Its universal specification for every s775 mobo... not just intel mobos...

if intel wouldn't release this spec noone could build motherboards with s775, cause noone would know what voltages needed to supplied... it might be true that for some old mobos the 5V line has deliverd the power for the cpu, but for every s775 thats not the case. (see VDR11.x)
The "design" of the DQ6 may be different, but the basis on what this design is build is not different from any other X38/P35 or even any P965 Board that is build according to the VDR 11.x spec. Do you really think it makes any difference if you use 12 phases or 8 phases... the basics are the same.... also the power managemant implementation is exactly the same on every board, or how the hell you think the board knows what voltage the cpu requiers if the board cant interpret the signals coming from the cpu...

Also in the Manual theres nothing stated that the cpu uses 3,3V or 5,0V, the only thing that is mentioned (page 24) that it is recommended to use a PSU with a 8pin (2x4 pin) 12V EATX connector (ops.. 12V, not 3,3V or 5V) if you use an extreme edition processor.

But belive what you want, you seem to know everything better, even better than intel themself.

But i have a problem when you spread misinformation and tell people to buy stuff that doesnt change one thing. Voltage drops dont get better just from buying a psu with 1kW or more.

Its true that when you oc you need a good an stable psu and a PSU that is on its limit can cause more serious drops, but 1kW is beyoned usefullness, if you dont run a system with a highend (oced) quad and quad sli/ quad fire.

I have a 600W PSU (Zalman ZM600 HP-600) and the psu has hardly any drops under load... even with an oced quad... and my PSU only has 12A per 12V rail (quadrail).

if you really want to get solid informations about PSUs visit: http://www.jonnyguru.com/ page. He also posts here on the board and his reviews are first class.

I have bought my zalman psu after doing some research and after i have read jonnys review i pulled the trigger and till now i havent regreted it.

[ChrissTi]

I think that you must shut up and understand how power in a computer works and do some calculations with this little overclock PSU calculator:

http://www.extreme.outervision.com/p...ulatorlite.jsp

To overclock a quad QX9650, at 3600 (and the CPU voltage to remain constant not vdrops just 1.25V), with an 8800GTX, a tv tunner, 3hdd, .... you need at least 1000W to run the system at 90% load.
More overclock juise, more power ...

And you can laugh as long as you want ... I'm not gonna argue with you anymore, believe what ever you want.

[Hornet331]

I think that you must shut up and understand how power in a computer works and do some calculations with this little overclock PSU calculator:

http://www.extreme.outervision.com/p...ulatorlite.jsp

To overclock a quad QX9650, at 3600 (and the CPU voltage to remain constant not vdrops just 1.25V), with an 8800GTX, a tv tunner, 3hdd, .... you need at least 1000W to run the system at 90% load.
More overclock juise, more power ...

And you can laugh as long as you want ... I'm not gonna argue with you anymore, believe what ever you want.

god your spread so much FUD, it isnt even funny any more... are you working for gigabyte or a PSU company?

http://www.techreport.com/articles.x/13375/15

here you go Quad extreme X6850 with 8800gtx less then 290W at socket! Which means that if you assume that the PSU has an effenciy of 80% the actual consumption of the system would be ~230W....

and i laught even more at you... you pretend to know stuff from what you have no clue and mock yourself...

to end this once and for all with your own link to the PSU calculator with better sepc than you have provided:

DUN DUN DUN 584W




Sorry guys for crapping this thread, but i cant stand it when somone recommends stuff which is totally unnessary. I hope you can forgive me.

[Isepick]

Speaking of voltages, etc on this board (X38T)...I was pushing mine this morning to get up over 4 Ghz and as I was adjusting the voltages they started getting out of control. I set Vcore to 1.75 and I got it to boot into windows and Everest said Vcore 2.05!!! Everest seems to have been correct all the way up before this, so I'm not questioning its accuracy (too much)...my Vdimm also jumped from 2.00 set in bios to 2.10v...I wonder if this is a bios bug or a limitation on the power supply section of the board. Will verify these numbers with my Fluke if someone can tell me where on this board the test point to measure is....

BTW ignore my previous clock skew posts, since I had some of the memory subtimings way out of line anyway. That Anandtech article helped me understand a lot more about this board, and I finally got my memory to run at its rated speed using the settings in that article. My first OC rig, so it has been a learning experience.

Oh and prime locked up in about 3 seconds at 4 Ghz, but with the huge variations on the voltage I didnt want to increase it anymore in the bios...but I DID get a screenie off of Everest first

[freqy]

hi,

did anyone find a little trick or bios update to increase ram above 1066 on a x38-dq6...i remember someone saying that a fix was on its way to make this easier.

im using dominator pc-8500 at 1066 ...but i bought this ram and motherboard thinking i could over clock it to higher rates with ease...

thanks for any help

[ChrissTi]

Anyway all that I've wanted to say that the best PSU is the one that gives more current for the CPU, and not to be forced to rise the voltage to achive the necessary power when under full load. More voltage means higher temperatures, and risk to damage the CPU.
Also a part of that current goes to the other devices, so the cpu still has less current and needs a rise of the cpu voltage, but is lower on a higher current PSU ...
The problem is that if there is a PSU that gives enough current it gives also more power than necessary (because of that 12V rails, example 12*90Amps=1080Watts)... That's why for better overclock is better higher power PSU.

For a normal user with a little oveclocking skills, not enthusiast overclocker, It will be good an 700-850W PSU, but to have a good +5V and +3,3 amperage and at least 40Amps on 12V single rail.

[GAR]

im running 1066 with ease with a 4gb set......but 1100mhz is a bit tough.

[ChrissTi]

GAR manual or auto timings ?

[freqy]

yeah its tough to get past 1066.


i just downloaded everest to compare memory benchmarks with isepick above who is running ddr3 1600mhz 800 x2 ) 7 7 7 20.

im running lower at 1066mhz ddr2 (dom pc8500 x2 ) standard performance enhanced
at apparently lower timings 5 5 5 10 (static tread 7) timings

everest says:

me=8378 read (10023 for isepick)
me=3226 write (7140 more than double ???? )
me=59.3 latency (53.4 - hey you!! how?? haha)

i would have thought i would at least get lower latencies? being as though my timings are lower?:/

now i just lowered my timings ..cause im very upset. haha >:

5 5 4 9 (static 7 and standard performance)

and i get

me=read 8437
me=write 3087 (weird why did this go lower than my last score :/?)
me=latency 58.5 ( hmm still not better than ddr3)


im using a q6600 at 3.2ghz
fsb 400 with 2.66d and x8
x38 dq6 f7h
dominator pc8500

freqs

[ChrissTi]

yeah its tough to get past 1066.

i just downloaded everest to compare with isepick above who is running ddr3 1600mhz 800 x2 ) 7 7 7 20.

im running 1066mhz ddr2 (dom pc8500) standard performance enhanced
at 5 5 5 10 (static tread 7) timings

everest says:

8378 read
3226 write
59.3 latency

i would have thought i would at least get lower latencies? being as though my timings are lower?:/


now i just lowered my timings too>:

5 5 4 9 (static 7 and standard performance)

and i get

read 8437
write 3087 (weird why did that go lower?)
latency 58.5


im using a q6600 at 3.2ghz
x38 dq6

freqs

and the other timings ? auto or manual ?

[freqy]

latency settings for the ram during everest tests

f7h bios reads from top to bottom in advanced timing control

3 3 6 42 3 7 5 2t

>>>

o.k next test turbo enhance 5 5 5 10

read. 8417
write 3043
latency 58.9

hmm ...not as good as my 5 5 5 4 9 .


why on earth would my 'write' score go down when i reduce the latencies or use turbo??

im not sure which setting to use now :/
*****
i just reset to 5 5 5 10 standard to retest and the write is at 3020 now so everest seems to have a slight fluctuation in readings.
each time i hit report on 'write' the score goes down each time i test in succession - i think i need to purge the ram each time perhaps. it went down to 2997 and back up to 3033.

i am using the free copy everest home edition.

[Hornet331]

Anyway all that I've wanted to say that the best PSU is the one that gives more current for the CPU

uhnn..... so wrong again... the best PSU is not that, that give the highes current... The best PSU is that, that gives a stabel and ripple free voltage at a given current. You cant do any thing with a PSU that can give 60 amps over a line, but that line drops or ripples like mad...

and not to be forced to rise the voltage to achive the necessary power when under full load.

i dont know what you mean by that... but if you allude to the voltagedrop that occours while the cpu is loaded, that has nothing to do with the PSU... you can have a 100amps 12V line and the drop would still be there... i allready have pointed that out several post ago.

Also you dont raise the voltage ,while overclocking, to give the cpu enough power... the raise in voltege, while ocing, is because at higer frequencies the signals inside the cpu get disordered. With higher voltage you can compensate that effect, you artificially increase the peaks of the signals (0 and 1). But this increased voltage also produces more heat and has other side effects, like faster ageing of the processor or electron migration. (sudden death northwood syndrome)

More voltage means higher temperatures, and risk to damage the CPU.

correct

Also a part of that current goes to the other devices, so the cpu still has less current and needs a rise of the cpu voltage, but is lower on a higher current PSU ...

uhn... if other devices affect the the voltage of your cpu than either your mobo is going byebye or your PSU is to weak.
I agree that a to small PSU can have problems with a fully load system, especialy with highend graphic cards. But if your PSU is not powerfull enough to power your system no voltage rise will make your system stabel. So i dont really understand what you want to say with that statement.
Also nowaday we have multirail PSUs, usually they are dual rails, better PSU have more rails(4 or even more). Usually the CPU has its own rail and the gfx has the other line so they cant affect each other.

The problem is that if there is a PSU that gives enough current it gives also more power than necessary (because of that 12V rails, example 12*90Amps=1080Watts)... That's why for better overclock is better higher power PSU.

wrong, even a 2kW PSU wont give more current to the cpu, as the cpu needs. The only effect you have is that the PSU is running in the partial load zone. In this zone PSUs are very inefficent and you waste a lot of power... CO2 comes to mind. :p

For a normal user with a little oveclocking skills, not enthusiast overclocker, It will be good an 700-850W PSU, but to have a good +5V and +3,3 amperage and at least 40Amps on 12V single rail.

oh, i can prove you otherwise. There are enthusiast ocers out there that oc with a 550W power supply (but its more out of misery than fun :p)

But i agree. An avarage ocer dosen't need more than a 850W supply even with SLI.