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

[Richard Dower]

Good Gigabyte resource guys, has alot of beta bios before anyone else...bios settings and O/C advice:

http://www.jzelectronic.de/jz2/index.php

F8d beta bios:

http://rapidshare.com/files/95459820/x38dq6f8d.exe.html

Few more updates...new version of Gigabyte Dynamic Energy Saver, version B8.0128.1:

http://www.giga-byte.com.tw/Support/...lity&FileID=80

@BIOS, version 1.23:

http://www.giga-byte.com.tw/FileList...ech_a_bios.htm

http://www.giga-byte.com.tw/Support/...lity&FileID=52

[6_6_6]

F8D is a little buggy, Try F8C

None of the features is removed or not working ... If you are missing them then there is a problem in your system.

Thanks. I tried F8C and i still needed more voltage to get the same thing and EIST line was not there either with 8 x 400.

I can confirm that:

1.36500v to run 8 x 400 stable in F8C-F8D.
1.33750v to run 8 x 400 stabile in F7.

I am back to F7. That seems to work well except the EIST line at 8 x 400. Same issue apparently with all bioses after too.

[OM DUST]

Richard - thanks for the links - Much appreciated. Just curious - why has everyone fled from the F3 BIOS - early testing on Q6600 seemed to show it was stable at lower voltages and allowed significantly higher FSB. DINOS seemed to get excellent results with stock BIOS. Was it just a search for a vdroop fix? Or was there a great deal of instability with quad cores that I missed?

[ChrissTi]

Board has finally arrived. Is there a download page (other than Gigabyte's) with all of the BIOS for X38 DQ6?

Appreciate the link. If vdroop isn't terrible and I can get stable 9X400 I will likely stay with F3 - but just in case I'd love to have F8d standing by.

http://www.jzelectronic.de/jz2/index.php

[Richard Dower]

got my new Raptors!!...both drives are 00NLR5 and were both made January 2nd 2008!!

Picie included, sorry for the crappy image quality, cheesy camera-phone using Bluetooth:

[OM DUST]

Put Board together last night, with my shiny new Q6600 G0. Love the back plates, they worked great on the 790fx board - here they don't work with my water block . Oh well, otherwise all went well.

I loaded XP Pro and did all my updates and such. Started o/c early in the AM. I am now up to 9X375 at 1.3525, FSB and MCH at Normal. Temps are great. CPU is 2-3C above ambient, chipset is 5-6C above ambient. Memory is underclocked (DDR2 750 as oppossed to 800 - will do 855) - at 2.1V 4-4-4-12 at the 2.0A strapping. No turbo or extreme.

But now I can not go any higher. I tried raising CPU voltage step by step all the way to 1.4, while raising FSB and MCH voltages each step by step. Clearly it is not heat that is holding me back - but it will not even load into XP. I'm sure this being my first time away from AMD I am missing something obvious.

Oh - yeah there is some voltage fluctuation. But I'd like to get to 3.6 stable before I flash to F8d, as it seems like there may be more tweaking there.

[CyberDruid]

Am I the only person that finds this board challenging to OC stable? I can pull off high clocks for a SS but if I want Folding stablility I have to dial down...

I've worked a lot of different boards...this one is still fuill of mysteries for me...

[Dakkahun]

@OM DUST: I'd try upping the fsb and (g)MCH voltage a little, 2 or 3 steps from stock. Also, try changing the memory strap to 2.00B, or D.

[OM DUST]

@OM DUST: I'd try upping the fsb and (g)MCH voltage a little, 2 or 3 steps from stock. Also, try changing the memory strap to 2.00B, or D.

I have tried upping both. I took MCH all the way up to +.20 and FSB up to 1.5. Could I need to jump the CPU voltage up past 1.4 or higher to pass this threshhold?

Just to be sure I'm clear - the strap is the chipset timings? I thought strap A was the loosest? If not, then I will flip to the loosest and work my way up.

[OM DUST]

Good Gigabyte resource guys, has alot of beta bios before anyone else...bios settings and O/C advice:

http://www.jzelectronic.de/jz2/index.php

F8d beta bios:

http://rapidshare.com/files/95459820/x38dq6f8d.exe.html

Few more updates...new version of Gigabyte Dynamic Energy Saver, version B8.0128.1:

http://www.giga-byte.com.tw/Support/...lity&FileID=80

@BIOS, version 1.23:

http://www.giga-byte.com.tw/FileList...ech_a_bios.htm

http://www.giga-byte.com.tw/Support/...lity&FileID=52

Wish I read German. Is 'C' strap looser timing than 'A' strap. That is, I want to run 400 FSB - Ram at DDR2-800 at its EPP settings. Would I run it at 2.0 A or 2.0 C?

[Dakkahun]

@OMDUST: The 2.00B strap seems to do the trick for me at FSB 400.

[REVHEAD]

@OMDUST: The 2.00B strap seems to do the trick for me at FSB 400.

thats what I use, gives nice results in everest as well.

[OM DUST]

Thank you both - really appreciated. I will try as soon as I get home. Rev you seem to have a very similar system - what are your FSB/ CPU/ MCH/ voltages?

Dak - you are running 8 X 400?

[Dakkahun]

Yes, I run 8*400, with the memory at ECC timings. I've run for a while with 8*450, but the latencies were worse.

[REVHEAD]

Thank you both - really appreciated. I will try as soon as I get home. Rev you seem to have a very similar system - what are your FSB/ CPU/ MCH/ voltages?

Dak - you are running 8 X 400?

Because my ram doesnt overclock well I run that @ 800mhz

my fsb is 400mhz

My Vcore voltage is 3.475 in bios

Loadline calibration on just for the heck of it ,it doesnt really do a thing for me on or off.

My cpu is a b3 stepping quadcore and my MCH voltages are on default.

Memory performance is on extreme and my PIE performance is on turbo I think..

I have a link to my bios pics if you want to take a peek ,but you need to register to see them ,sorry if thats a inconveniance

[Nuckin_Futs]

I also need now 1.36500v to run 8 x 400 stable in F8D.
I needed 1.33750 to do the same in F7.
Is this a BIOS issue or is the CPU gone?
Why are you folks so unresponsive here?
Maria

Not too sure. Can you flash back to F7 to confirm?

Am I the only person that finds this board challenging to OC stable? I can pull off high clocks for a SS but if I want Folding stablility I have to dial down...

I've worked a lot of different boards...this one is still fuill of mysteries for me...

I can't help to notice either that not too many people utilizing a good chipset to higher OC. Especially a chipset easily known for 500FSB/2000HT.

For those asking about EIST, do know the differences 1st and then decide wich feels better.

EIST and TM1/2 are both thermal monitoring functions to save power by reducing both vcore and multi.

EIST is the Intel branded method that controls the sped steps from the mobo chipset so it of corse is the least favorable and slowest since it also uses system resorces. EIST must see VID and FID from TM table on CPU die and work in supportied BIOS'. EIST will control what is known as the CPU P state in steps of 1 at a time (ie: your CPU highest of 10, then 9, 8, 7, and finally 6). User will feel this change a bit more since it is using RAM, FSB and CPU as the info is constantly sent back n forth the PC system bus, so it has a slight lag sometimes, and thus will make higher overclocks become much more unstable.

TM2+C1E is another Intel feature but it is on die so the transitions between stepoing is much faster and almost unnoticed by user. TM2 uses the on die VID and FID (vcore + multi ratio) to instantly switch from high to low. In most cases, an average user will not have time to see, let alone use an inbetween multi like 7 or 8 and this change just takes up more time. So TM2 drops to the lowest every time CPU calls for a P state. Now normally that is the moment you do anything in OS, adn is no longer energy saving if it is alwayy on high, so C1E uilizes those tables from TM and allows CPU to function up to certain % usages before calling for a P stare (hault) so a light duty app like media player or surfing the net can work on the low VID/FID. However, at above a certain stress level, C state calls for loading and CPU goes to full multi/and vcore for max power. This is what is called C State + P state.

Now not all motherboards fully function in this C=P State and I do believe Gigabyte is one, wel mine never did. You can however, Endble TM2 = C1E and at least get multi control for reduced clock and save on a bit of heat for high OC. vocre will remain full as set because this mobo is fixed to older tech TM1 not 2, so it doesnt read the VID table to reduce vores along multi. It may be a good thing on some not so stable mobos and or CPUs' since the low VID may be too low for certain OCs'. This was a big issues with old E6600's.

So, even when you have the choice of both at such the mild overclocks I see here a lot, use one or the other, but not both of EIST and TM2+C1E. I'd higly suggest TM2+C1E instead for better stability when overclocking.

Hope This Helps!

[Kondik]

whoah .. my X38 DQ6 got Back from RMA ... CHanged ! they gave me a New one ... now's the time to ask .. what MOBO to keep I had less problems with Gigabyte however the GB can;t be Thermalrighted in way i would like ... god i hate this planet

[OM DUST]

Not too sure. Can you flash back to F7 to confirm?



I can't help to notice either that not too many people utilizing a good chipset to higher OC. Especially a chipset easily known for 500FSB/2000HT.

For those asking about EIST, do know the differences 1st and then decide wich feels better.

EIST and TM1/2 are both thermal monitoring functions to save power by reducing both vcore and multi.

EIST is the Intel branded method that controls the sped steps from the mobo chipset so it of corse is the least favorable and slowest since it also uses system resorces. EIST must see VID and FID from TM table on CPU die and work in supportied BIOS'. EIST will control what is known as the CPU P state in steps of 1 at a time (ie: your CPU highest of 10, then 9, 8, 7, and finally 6). User will feel this change a bit more since it is using RAM, FSB and CPU as the info is constantly sent back n forth the PC system bus, so it has a slight lag sometimes, and thus will make higher overclocks become much more unstable.

TM2+C1E is another Intel feature but it is on die so the transitions between stepoing is much faster and almost unnoticed by user. TM2 uses the on die VID and FID (vcore + multi ratio) to instantly switch from high to low. In most cases, an average user will not have time to see, let alone use an inbetween multi like 7 or 8 and this change just takes up more time. So TM2 drops to the lowest every time CPU calls for a P state. Now normally that is the moment you do anything in OS, adn is no longer energy saving if it is alwayy on high, so C1E uilizes those tables from TM and allows CPU to function up to certain % usages before calling for a P stare (hault) so a light duty app like media player or surfing the net can work on the low VID/FID. However, at above a certain stress level, C state calls for loading and CPU goes to full multi/and vcore for max power. This is what is called C State + P state.

Now not all motherboards fully function in this C=P State and I do believe Gigabyte is one, wel mine never did. You can however, Endble TM2 = C1E and at least get multi control for reduced clock and save on a bit of heat for high OC. vocre will remain full as set because this mobo is fixed to older tech TM1 not 2, so it doesnt read the VID table to reduce vores along multi. It may be a good thing on some not so stable mobos and or CPUs' since the low VID may be too low for certain OCs'. This was a big issues with old E6600's.

So, even when you have the choice of both at such the mild overclocks I see here a lot, use one or the other, but not both of EIST and TM2+C1E. I'd higly suggest TM2+C1E instead for better stability when overclocking.

Hope This Helps!

I think it helps. Its allot to absorb. Especally given I was up all night trying to figure out why CPU is so unstable over 9 X 350.

I can do 9 X 350 all on stock 2.0B Strap. P95 stable for 2 hours. Idle temps 19-21C (using CoreTemp) Load temps in the 40sC. Above 350 I need to add +1. to both FSB and MCH and take CPU to 1.35 (BTW CPU VID 1.325 ) These settings are ok until 9X375, at this point I need to go to +2 and +2.5 and CPU all the way up to 1.41. (again as Ram is underclocked I have ram at 2B strap at 2.2V). Temps 25-27C Idle, 47-50C load. I can not get stable beyond this. I have EIST disabled, and T-1 disabled, but load line calibration is enabled - vdrop about the same but vdroop much better at load 1.4. After 1 hour 20min P95 failed on core 4 at 9X380. Last night I tried flashing F6, F7 and F8d - all with the same results. I am a little frustrated as my RAM is still well underclocked at DDR2 750, and I appear to be using a great deal more voltage on CPU, MCH and FSB than others and still not hitting what virtually every other Q6600 G0 is getting on this board.

Do you think I should disable load line calibration? Should I enable EIST. I am still using f8d.

PS - I checked PSU rails and they are very solid even during second hour P95.

[koc]

I use bioc F8d

400X9 stable all the way

Vcore 1.46 , all other voltage set to auto

Idle temp 35 - 40
load temp 55 - 60

I might lapping the HS

[OM DUST]

Wow Rev - 1.475V - how is your temp? I was concerned about 1.41V. I see you also have memory to extreme, I have it at normal and Graphics Booster at Normal as well. I see at 400 FSB 'D strap' you did not need to up your FSB or MCH. Of course I have a SLACR G0 which should provide better stepping, though a 1.325 VID is not making me happy. Maybe I just need to keep pumping volts and then I can try tweaking back on FSB and MCH. I tried 'D' strap last night - I didn't seem to get better results. Funny there is no 2C strap.

[OM DUST]

KOC you say 400X9 also stable using stock FSB and MCH volts. Am I giving unnecessary volts to the board at a FSB at or below 400??

KOC - what strap A, B, C, or D?

Also your sig says 3780- - was that speed not stable?

It looks to me that the biggest indicator of high o/c stability is nothing you can set, but is in fact the CPU Vid. From what I see even a B3 stepping with a low Vid (1.225) beats a G0 with a high Vid (1.325). Seems that may be more a factor than stepping, strap, or any settable Bios tweak.

Maybe I will try 8X450 - though not sure if my memory even at 5-5-5-15 can do it.

[Nuckin_Futs]

OM DUST; If all you need is DDR800, then use your 8x multi and up the FSB. This CPU (even the later GOs') can handle 3.5GHz so you should be gooas as 500x8 in 1:1 for DDR800. Might need 1.42vand +.2v on allmobo volts? I dont know if this mobo offers any kind of descrete CPU PLL, GLT Ref and FSBT tweak. If so high FSB quads need at least PLL: 1.6v, SFBT: 1.45v - 1.5v, and CPU/NB GTL Ref of .67x % if available (some only offer 63% on CPU). CPU PLL, and mobo FSBT and very important for higher stable CPU FSB overclocks.

DO NOT enable EIST during an OC trial. It will only make less stable and get in the way. It's is perfectly fine to enable TM but C1E can be left off for test and enabled when done for heat reduction.

Who can provid a BIOS CPU and RAM overclocking section template for this mobo?

ThanX!

[OM DUST]

Hmmm. I have been p95 blend test stable at 8X425 for a few hours and it seems ok at 1.420. Not sure I can push the memory much further.

I tried pushing voltage all the way up to cpu 1.5 and NB +2.0 and MCH +2.5 and at the 9X multi it was just unstable at 400. Running small FPUs would cause crash.

I would have liked 3.6, but with a CPU Vid of 1.324 I just don't think it will happen.

Weird thing is, Before I flashed, I thought the CPU Vid was WAAAY Lower . But the Vid is hard wired right? The M/B BIOS can't effect it, can it?

[koc]

KOC you say 400X9 also stable using stock FSB and MCH volts. Am I giving unnecessary volts to the board at a FSB at or below 400??

KOC - what strap A, B, C, or D?

Also your sig says 3780- - was that speed not stable?

It looks to me that the biggest indicator of high o/c stability is nothing you can set, but is in fact the CPU Vid. From what I see even a B3 stepping with a low Vid (1.225) beats a G0 with a high Vid (1.325). Seems that may be more a factor than stepping, strap, or any settable Bios tweak.

Maybe I will try 8X450 - though not sure if my memory even at 5-5-5-15 can do it.

Even with 420X9 i increase the Vcore and the DDR Volt only , and all other volts set to auto and the system very stable also , but i lower the FSB to 400 coz i want keep the load temp below 60C

I use strap D2.66

[6_6_6]

Nuckin_Futs, hey plenty of thanks! This answered a plethora of questions for me.


I cannot confirm voltage issue. However, F8D needed 3v more than F7 to pass 2 hours of prime95 small FFTS along with many other tests (3dmark, OTTC, everest-sandra stress, etc). But I have not tested F7 with large FFTs before. It fails. I did not test F8D with large FFTs so I do not know if it would fail or pass at its stable voltage for other tests.


As for EIST, I can confirm that EIST is disabled on 4 x 800. Its setting is no longer in BIOS and software reports it as disabled. F7, F8C, F8D all same results. EIST with 9 x 400 works in BIOS though... Also works for 9 x 333, 9 x 380... This brings me to a question:

Is EIST implemented only with the highest multiplier of a cpu (9 for Q6600) and not available with the intermediary steppings?

9 x 333 with no extra voltage (VID 1.30)... C1E, TM2, EIST enabled... I can see my voltage alternate between 1.16v and 1.29v when the speed changes from 2 GHz to 3 GHz. However, it is constant 1.35v with 8 x 400 even though speed alternates between 2.4 GHz and 3.2 GHz. And like I said, software is reporting EIST disabled and it is not available in BIOS. So is this C1E with TM2 kicking in and using constant TM1 voltages?

What do these options by themselves do? As far as I recall, EIST would not work if C1E was not enabled. Or something like that. Could you please be kind enough to let me know what each one of these will do when other 2 are disabled?

Is there a reliable way of finding out which one of these options are enabled or disabled? Sandra / Everest / PC Wizard / Crystal CPUID... all seem to be giving conflicting results. PC Wizard says Hyperthreading is enabled on my system for example. Which is the best way to find out the current state of each option?

When all C1E, TM2, EIST enabled, which one is being used in windows to throttle the clock?

How can I also make my CPU not go over 71C. Previously, the system would crash at those temperatures but now Thermal Management does not seem to work at all as I am even going sometimes to 80C and 10-15 more for the cores. This is not good.

Thanks in advance!

Not too sure. Can you flash back to F7 to confirm?



I can't help to notice either that not too many people utilizing a good chipset to higher OC. Especially a chipset easily known for 500FSB/2000HT.

For those asking about EIST, do know the differences 1st and then decide wich feels better.

EIST and TM1/2 are both thermal monitoring functions to save power by reducing both vcore and multi.

EIST is the Intel branded method that controls the sped steps from the mobo chipset so it of corse is the least favorable and slowest since it also uses system resorces. EIST must see VID and FID from TM table on CPU die and work in supportied BIOS'. EIST will control what is known as the CPU P state in steps of 1 at a time (ie: your CPU highest of 10, then 9, 8, 7, and finally 6). User will feel this change a bit more since it is using RAM, FSB and CPU as the info is constantly sent back n forth the PC system bus, so it has a slight lag sometimes, and thus will make higher overclocks become much more unstable.

TM2+C1E is another Intel feature but it is on die so the transitions between stepoing is much faster and almost unnoticed by user. TM2 uses the on die VID and FID (vcore + multi ratio) to instantly switch from high to low. In most cases, an average user will not have time to see, let alone use an inbetween multi like 7 or 8 and this change just takes up more time. So TM2 drops to the lowest every time CPU calls for a P state. Now normally that is the moment you do anything in OS, adn is no longer energy saving if it is alwayy on high, so C1E uilizes those tables from TM and allows CPU to function up to certain % usages before calling for a P stare (hault) so a light duty app like media player or surfing the net can work on the low VID/FID. However, at above a certain stress level, C state calls for loading and CPU goes to full multi/and vcore for max power. This is what is called C State + P state.

Now not all motherboards fully function in this C=P State and I do believe Gigabyte is one, wel mine never did. You can however, Endble TM2 = C1E and at least get multi control for reduced clock and save on a bit of heat for high OC. vocre will remain full as set because this mobo is fixed to older tech TM1 not 2, so it doesnt read the VID table to reduce vores along multi. It may be a good thing on some not so stable mobos and or CPUs' since the low VID may be too low for certain OCs'. This was a big issues with old E6600's.

So, even when you have the choice of both at such the mild overclocks I see here a lot, use one or the other, but not both of EIST and TM2+C1E. I'd higly suggest TM2+C1E instead for better stability when overclocking.

Hope This Helps!