**Official ABIT IX38 QuadGT Review/Overclock/Guide Thread**

[R*r]

@ust

how much you give th cpu vtt?my q66 also sometime act like that.1 core failed raising cpu vtt help my stability

[ust]

Hello R*r

As said I just followed this template of Supershanks:
http://www.clunk.org.uk/forums/3914-post30.htm

CPU VTT VOLTAGE.................... - 1.33V

[R*r]

ups so sory miss the link,i can stable at 400x9 with 1,41 vcore and 1,42 cpu vtt

maybe you want to try my setting?ive been using the setting 24/7

[Supershanks]

R*r is right Ust tweak vtt up a noitch & repeat run , sounds like your fairly close to stability.

luck

[jcool]

Buzzing sound when the system under load, bro? My previous Commando was, but this board is just running silent without any sound like that.

No, crackling and buzzing sounds are persistent but much louder when the system is NOT under load but in idle mode - that's what I can't comprehend.
I also don't know why it happend so suddenly.
Speaking of suddenly, right now i'm in idle and it's dead silent - maybe it heard something about "RMA"

@UST: You might also wanna try to increase ICH voltage, I only use 1,32V Vcore and 1,3V VTT for 400x9 but 1,2V ICH. Supershank's 1,13V haven't been enough for me.

[AMDfreak]

Well, I just put in my new HD 3870, and the problem is most definitely NOT the vid card. My PWM's are screaming even at idle right now and my next stop is Newegg's RMA site. Would anyone like to talk me out of buying a DFI board and just selling the new IX38 I get?

[jcool]

Damn... that makes two of us.

[mark0326]

I've crunched the mosfets cause I've mounted my waterblock on AS instead of thermal pad.

I thought that the PWM waterblock should have integral spacer's either end to stop the block ever coming into contact with the PWM's and thus damaging them - hence the need for the thicker tape.
Hope you get RMA sorted OK.

[ust]

'Bout talking you out of buying DFI's X38...
DFI's BIOS' are massive in an overwhelming way. I used to run an Athlon 4000+ on the Lanparty NF4-SLI DR Expert and it's propably one of the motherboards I've owned with highest production value - BUT unless you're an overclock master their BIOS' are gonna make you dizzy

@UST: You might also wanna try to increase ICH voltage, I only use 1,32V Vcore and 1,3V VTT for 400x9 but 1,2V ICH. Supershank's 1,13V haven't been enough for me.

Okay I'll try it out mate, thanks

[situman]

Well, I just put in my new HD 3870, and the problem is most definitely NOT the vid card. My PWM's are screaming even at idle right now and my next stop is Newegg's RMA site. Would anyone like to talk me out of buying a DFI board and just selling the new IX38 I get?

Why not get the Asus?

[ust]

Okay so I wanted to show you what problems I'm running in to here.

9*400 should be quite attainable for most Q6600's on water and with highend components.

So I cranked the FSB frq. up and did'nt touch my other settings at all:
ABIT IX38 QuadGT

OC GURU:

CPU OPERATING SPEED................ - USER DEFINE
EXTERNAL CLOCK..................... - 400 MHz
MULTIPLIER FACTOR.................. - x 9
ESTIMATED NEW CPU CLOCK.......... - 3600 MHz
DRAM SPEED (CPU/DRAM)........... - DDR2-1:1.25 DDR2- 1000 Mhz
PCI EXPRESS CLOCK.................. - 100MHz

VOLTAGE CONTROL:

CPU CORE VOLTAGE................... - 1.4125V
CPU Core Voltage Offset............ - 0.00V
Target CPU Core Voltage............ - 1.4125V
DDR2 VOLTAGE....................... - 2.175V
CPU VTT VOLTAGE.................... - 1.42V
MCH 1.25V VOLTAGE.................. - 1.45V
ICH 1.05V VOLTAGE.................. - 1.20V
ICHIO 1.5V VOLTAGE................. - 1.50 V
DDR2 REFERENCE VOLTAGE............. Default
CPU GTLREF 0+2..................... - 68%
CPU GTLREF 1+4..................... - 67%

ADVANCED CHIPSET FEATURES:

DRAM TIMING SELECTABLE............. - MANUAL
CAS LATENCY TIME (TCL)............... - 5
RAS TO CAS DELAY (TRCD)............. - 5
RAS PRECHARGE (TRP)......... .......... - 5
PRECHARGE DELAY (TRAS).............. - 15
REFRESH CYCLE TIME (TRFC)........... - 30
WRITE RECOVERY TIME (TWR)......... - 3
WRITE TO READ DELAY (TWTR)........ - 3
ACT TO ACT TIME (TRRD)............... - 3
READ TO PRECHARGE (TRTP)............ - 12
COMMAND RATE....................... - 2T
PCIE Root Port Function............ - V1.0A
PEG FORCE X1....................... - DISABLED
INIT DISPLAY FIRST................. - PCIEX

ADVANCED BIOS FEATURES / CPU FEATURES:
THERMAL CONTROL.................. - Disabled
C1E FUNCTION....................... - DISABLED
EXECUTE DISABLE BIT................ - DISABLED
VIRTUALISATION TECHNOLOGY.......... - DISABLED
EIST FUNCTION...................... - DISABLED

Fired up Prime and two of the cores failed right away.......

In my oppinion all of my voltages are fairly high and my settings fail safe.
Any suggestions?

[Supershanks]

Ust i don' quiet understand what you did.
from appearing to be close to being stable here

I returned core3 had errored after 1hour and 83 minutes while the other three cores were stil running - and they had been running for 53 hours without errors ( ! ) when I stopped the test.

It being suggested that you increase vtt & ich
you now say

So I cranked the FSB frq. up and did'nt touch my other settings at all

?? we thought you were already on that??

Can't read your settings or zoom in on them.

it might be worth you copying a template from Here pasting it into a post here & filling in the settings you used.

we can perhaps help you best this way.

luck

[ust]

Hi again matey

Sorry I can see what you are talking about. I'll try to express myself more thoroughly now...

First I tried to follow your 400*9 template, but Prime failed right away with those settings. So i bumped up my Vcore to 1.4250 and lowered the FSB Frq. to 389 from 400. With these settings, as said in my previous post, one core failed after 1 hour and 80 mins. while the other three went on with no errors for 52 hours.

Okay, back to present time. I'm advices by Supershanks and R*r to bump up my VTT and MCH, but decided to try out 9*400 again since it seem to be a sweetspot for the CPU judging from the countless OC threads here.

So I tried with a higher VTT and MCH to reach 9*400 and this is where we are now

With these settings 2 of the cores fail immidiately:

ABIT IX38 QuadGT

OC GURU:

CPU OPERATING SPEED................ - USER DEFINE
EXTERNAL CLOCK..................... - 400 MHz
MULTIPLIER FACTOR.................. - x 9
ESTIMATED NEW CPU CLOCK.......... - 3600 MHz
DRAM SPEED (CPU/DRAM)........... - DDR2-1:1.25 DDR2- 1000 Mhz
PCI EXPRESS CLOCK.................. - 100MHz

VOLTAGE CONTROL:

CPU CORE VOLTAGE................... - 1.4125V
CPU Core Voltage Offset............ - 0.00V
Target CPU Core Voltage............ - 1.4125V
DDR2 VOLTAGE....................... - 2.175V
CPU VTT VOLTAGE.................... - 1.42V
MCH 1.25V VOLTAGE.................. - 1.45V
ICH 1.05V VOLTAGE.................. - 1.20V
ICHIO 1.5V VOLTAGE................. - 1.50 V
DDR2 REFERENCE VOLTAGE............. Default
CPU GTLREF 0+2..................... - 68%
CPU GTLREF 1+4..................... - 67%

ADVANCED CHIPSET FEATURES:

DRAM TIMING SELECTABLE............. - MANUAL
CAS LATENCY TIME (TCL)............... - 5
RAS TO CAS DELAY (TRCD)............. - 5
RAS PRECHARGE (TRP)......... .......... - 5
PRECHARGE DELAY (TRAS).............. - 15
REFRESH CYCLE TIME (TRFC)........... - 30
WRITE RECOVERY TIME (TWR)......... - 3
WRITE TO READ DELAY (TWTR)........ - 3
ACT TO ACT TIME (TRRD)............... - 3
READ TO PRECHARGE (TRTP)............ - 12
COMMAND RATE....................... - 2T
PCIE Root Port Function............ - V1.0A
PEG FORCE X1....................... - DISABLED
INIT DISPLAY FIRST................. - PCIEX

ADVANCED BIOS FEATURES / CPU FEATURES:
THERMAL CONTROL.................. - Disabled
C1E FUNCTION....................... - DISABLED
EXECUTE DISABLE BIT................ - DISABLED
VIRTUALISATION TECHNOLOGY.......... - DISABLED
EIST FUNCTION...................... - DISABLED

[jcool]

I thought that the PWM waterblock should have integral spacer's either end to stop the block ever coming into contact with the PWM's and thus damaging them - hence the need for the thicker tape.
Hope you get RMA sorted OK.

Naw, that's only true for the Mips PWM block, not for the watercool...

@ust: Looks like your CPU doesn't like high VTT. You might want to try to lower it (it's not one of these settings where more is always better).
Another thing, how about your CPU temp? You can check them with Coretemp 0.94 (do NOT use 0.95 or it will crash your system at the very least!)
If they are too high (over 65) your CPU might also be unstable @3,6Ghz.
Mine gehts unstable at about ~69C...

Edit: The problem here is that your GTL Ref and VTT aren't corresponding any more. If you want to run 1,4V VTT you also need to increase GTL settings. See here for a detailed explanation.
I know it's a lot but hey, if you want easy plug and play overclocking, don't set your expectations that high or get a Dualie

[ust]

@ust: Looks like your CPU doesn't like high VTT. You might want to try to lower it (it's not one of these settings where more is always better).
Another thing, how about your CPU temp? You can check them with Coretemp 0.94 (do NOT use 0.95 or it will crash your system at the very least!)
If they are too high (over 65) your CPU might also be unstable @3,6Ghz.
Mine gehts unstable at about ~69C...

Edit: The problem here is that your GTL Ref and VTT aren't corresponding any more. If you want to run 1,4V VTT you also need to increase GTL settings. See here for a detailed explanation.
I know it's a lot but hey, if you want easy plug and play overclocking, don't set your expectations that high or get a Dualie

Hello

My CPU temps are not the problem, the two hottest cores only get 51c when stressed. This is with 1.4250 Vcore.
Using Coretemp V.0.95.4 and I have experienced no crashes what so ever.

Heh I know OC'ing is not a piece of cake, hence my posts in this thread.
Though I must admit that I don't quite understand what you're trying to say 'bout the GTL settings and VTT. I thought a GTL value of 67%/68% fitted my current VTT value?

[jcool]

According to the article here it doesn't
Of course this has been written for a DFI board, but it should hold true with the Abit, too.
What cooling setup do you use anyways? I get 65C max with my Apogee GTX, kinda sucks
I might need to reapply the cooler, apparently that liquid metal pad isn't half as good as I thought.

[Supershanks]

See here for a detailed explanation.

i went there all exited thinking i was going to learn something

I thought a GTL value of 67%/68%

If you read the anentech article you'll see that gtl is pretty mucg cpu specific, it vcan vary from cpu . mark0326 has similar problems. I had a real bad day at the office when i 1st got my board until i hit what seemed for me a stable place at

CPU GTLREF 0+2..................... - 68%
CPU GTLREF 1+4..................... - 67%

It's worth lowering vcore almost to failure point & using Prime & guru as described. yours might be stable at say 69/68 or whatever,

jcool is right re the temps as it says here

Our tests revealed that CPU load testing failures generally occurred when CPU temperatures exceeded 65C or so. Over 65C it becomes ever more difficult to tune GTL voltages as there is always some uncertainty pertaining to the cause of a CPU core failure: Was it temperature or VTT/GTL related?

ust
What fft does prime fail on large oir small??
your sub timings

WRITE RECOVERY TIME (TWR)......... - 3
WRITE TO READ DELAY (TWTR)........ - 3
ACT TO ACT TIME (TRRD)............... - 3

may be a little tight try running those on 4 not 3.

luck

[ust]

What cooling setup do you use anyways? I get 65C max with my Apogee GTX, kinda sucks
I might need to reapply the cooler, apparently that liquid metal pad isn't half as good as I thought.

Triple Rad with three antec tricool on medium, Laing D5 pump, distilled water, Antec P182 with no intake fan at the moment and have applied Arctic Silver one the CPU. My CPU is the only component that is being cooled by the water loop.

SUPERSHANKS:

I'm running the "Small FFTs test" in Prime.

Okay I'll try lowering the timmings a bit.

So there isn't really a way to determine which GTL settings work the best?

Thanks guys, keep it coming

[Supershanks]

just b4 i go & fall into my pit
As jcoiol sdays the anandtech aarticle is for the dfi board, absolute values like the graph shown here VTT & FSB Limits won't necessarily apply to ther Abit, BUT the methodology does.

My vtt settings have been (all a minimum of 12hts prime stable)

400x9 1.33v probably overcooked
450x8 1.30v probably low to blob on
465x8 1.45v again is over the top
470x8 1.36v This is my current 24/7 seems very comfy.

so i think 450 & 470 are about blob on. 465 should probably be 1.34v
with ~1.30 as jcool says for 400. Throughout my gtlref has been at 68/67 but as stated you need to verify what is best for your quad.

[jcool]

i went there all exited thinking i was going to learn something

Sorry mate just using your nice page

About the subtimings, I think they're ok. I'm using 3-3-3-8

@ust:
There you go... I also cool my 8800, the PWMs and my HDDs - especially the 8800 heats up the water.

[Supershanks]

So there isn't really a way to determine which GTL settings work the best?

yes there is it's a laborious job i've tried to descriibe the process here Finding Optimum GTLRef

Luck

About the subtimings, I think they're ok. I'm using 3-3-3-8

Dang maybe i should tighten mine

[Humble]

I myself, also have problems with divider 1.6+ and 2.0+ in BIOS 12B06.
Divider 2.0 even has two version, one is 2.0 only and the other is 2.0+ (with plus sign). Anybody know details about the difference between the plus and non plus version?
When I use divider 1.6+ and 2.0+, my system is locked up and can't even POST, so I have to clear CMOS. Meanwhile the other dividers are working just fine.

Can anyone here explain this issue? I still wait 4 the answer.

And more issue to say, at my setting now (which may be just a low OCing state compared to yours), the PWM's temperature still get ridiculously hot at maximum about 72C. I want to add a 4 cm fan above the PWM heatsink, but the clips is missing. Any idea how to mount a fan without the clips?

Thanks.

[Supershanks]

I myself, also have problems with divider 1.6+ and 2.0+ in BIOS 12B06.

Need more info what fsb are you running with Ram timings & voltage??

the PWM's temperature still get ridiculously hot at maximum about 72C

what vtt have you set??
either try making some retension wire from bending paperclips, or find somwhere to wedge a fan that fits. I'm running atm with a60mm fan that wedges in between my heatsink retension bolt & the back of the case.
Rubbish picture but you get the idea

[ust]

Hmm what should maximum PWM temps be?
At 1.42 Vcore my temps reach 90c +

[jcool]

I'm reaching a max of 70C on water, which is perfectly fine for 24/7 use. With stock PWM cooling and no airflow in the area I got close to 100C, which was still stable. Officially they are rated for 135C but anything over ~85C will shorten their lifetime considerably.