Abit IX38 Quad GT - Living Review II

[supershanks]

@glowball that looks great what fans you using ??

@Hawklan That looks a decent bit of kit certainly an improvement on the original., though i must admit i'm more of a fan of copper blocks.

@clunk

I did 420x9 on the DFI for 14 hours and it failed on one core, re-running here as well

Yep life's a bitch, I say it's a software/driver issue I'm taking pics as it progresses this time

[glowball]

Originally Posted by supershanks

@glowball that looks great what fans you using ??

These: Akasa AK-161 BL-S 40mm Blue LED Fan

Must say, I'm quite liking this board, few quirks... but that's how it is these days with 'anything' to do with a PC.
I can't find a supplier of 3mm thick copper sheet, only need 10cm x 10cm anyone help?

[supershanks]

Cheers have seen them . Might order a couple.
Akasa DFS401012M Hope you didn't get these

There are a few Ebay suppliers, i've had aluminum sheeting from them.
BRAND NEW COPPER SHEET 200mm x 150mm x 0.3mm I think he may have got the decimal place wrong not sure.

I'm quite liking this board, few quirks.

Yep know what you mean

[supershanks]

Just ordered 2 of these Genuine Papst quality 40mm fan Typ 412 F/2H 3 pin con

Technical specifications
Patch cord 3 Pin Board
AC/DC DC
Dimension A 40 mm
Dimension B 10 mm
Dimension C 32 mm
Tachosignal Yes
Bearings Sintec
Nominal voltage 12 VDC
Noise 29 dB(A)
Power input 0,80 W
Nominal speed 6000 min-1
Air flow 9,0 m3/h
Weight 17 gr.

[glowball]

Originally Posted by supershanks

Akasa DFS401012M Hope you didn't get these

I needed 2 but couldn't afford them!

There are a few Ebay suppliers, i've had aluminum sheeting from them.
BRAND NEW COPPER SHEET 200mm x 150mm x 0.3mm I think he may have got the decimal place wrong not sure.

Looking at the other stock, similar thickness's. Thanks.

Your fans won't have pretty lights

[supershanks]

Your fans won't have pretty lights

True I'm jut the man into black

Prime is at 8hrs now, have taken pics just to make sure

An interesting thing re those PWM figures, I have the mCubed T-Balancer bigNG The digital sensor is fastened on the top side of the PWM , my mobo is mounted inverted. The temps are shown as 41C whilst guru is showing 64C I wonder if we are getting hot under the collar because we simply have more information from guru than we normally have.

1) 41C is pretty close to the mossfet figures i quoted earlier, using the same sensor on the gigabyte & p5k, ok the abit has active cooling.
2) We don't generally have access to Mossfett/pwm core temps. Perhaps we are getting over e4xited.
3) I know Clunk had very high shutdown level PWM temps, perhaps this was a dodgy board;Umm:

I'm keeping an open mind atm about these temps & hope to do a few more benches to check things out.

[Clunk]

Originally Posted by supershanks

3) I know Clunk had very high shutdown level PWM temps, perhaps this was a dodgy board;Umm:

Yours were higher than mine at one point 112c was the highest I saw, and then it shut down. I seem to remember a screenie of yours where they were 117c or something?

[BUFF]

Originally Posted by supershanks

TThe temps are shown as 41C whilst guru is showing 64C I wonder if we are getting hot under the collar because we simply have more information from guru than we normally have.

1) 41C is pretty close to the mossfet figures i quoted earlier, using the same sensor on the gigabyte & p5k, ok the abit has active cooling.
2) We don't generally have access to Mossfett/pwm core temps. Perhaps we are getting over e4xited.

I've wondered about this before - in my (limited) experience with Asus & Gigabyte's I don't recall having indiviual temps for each phase as abit does & iirc on at least 1 board the PWM temp sensor was actually quite far from the MOSFETs etc.

Originally Posted by supershanks

Kicked off last night. got back lunchtime to see it still running with 16hrs up. played with skins in guru, went to open cpu-z ready for the pics.............. It Froze.:shrug

you may not be aware but there is a known conflict between uGuru & CPU-Z 1.41 which results in what you have just described.
it doesn't happen with uGuru & CPU-Z 1.40 & earlier.

[supershanks]

Sorry that's right it was this :-

9x356, PWM Temps & the search for lower Vcore
Was at 356x9 I was that low a fsb as i was trying to see how far my 1.40v vcore (9x400 rock solid stable on Gigabyte X38) would take me.
Settings were :-

ABIT IX38 QuadGT

OC GURU:
 
CPU OPERATING SPEED................ - USER DEFINE
EXTERNAL CLOCK..................... -  356   MHz 
MULTIPLIER FACTOR.................. - x 9 
# ESTIMATED NEW CPU CLOCK.......... - 3204  MHz 
DRAM SPEED (CPU:DRAM) ....... - 1:1.15 DDR2- 1068
PCI EXPRESS CLOCK.................. - 105MHz 
 
VOLTAGE CONTROL:
 
CPU CORE VOLTAGE................... - 1.40 V 
CPU Core Voltage Offset............ - 0 V
Target CPU Core Voltage............ - 1.40 V
DDR2 VOLTAGE....................... - 2.175V 
CPU VTT VOLTAGE.................... - 1.50 V 
MCH 1.25V VOLTAGE.................. - 1.35 V 
ICH 1.05V VOLTAGE.................. - 1.12 V 
ICHIO 1.5V VOLTAGE................. - 1.50V 
DDR2 REFERENCE VOLTAGE............. - Default 
CPU GTLREF 0+2..................... - 63%
CPU GTLREF 1+4..................... - 63%
 
ADVANCED CHIPSET FEATURES:
 
DRAM TIMING SELECTABLE............. - By SPD

So looking back with the knowledge we have know :-
1) 1.5v VCC is 25% (0.20v) too much for this FSB see Approximate VTT values and FSB ranges
2) GTLRef is at auto at least 3% below the "sweet spot" of

2) GTLs set to 66%-70% of VTT seem to be at the most effective point for "stable overclocking" of quad-core CPUs

the same link.
3) The 112C was passive cooling , i added the fan after seeing those temps.

Maybe we have a situation that surplus vcc power is dissapated as heat??

As I say i'm interested to see what the pwm temps show when i have a wider range of clocks to look at.

[supershanks]

The PWM is interesting , i'm trying to keep an open mind on it till i have a few more samples. Certainly when i ran my 9x400 12hr prime i saw no noticeable temps, using 1.33 Vtt & 68% 67% Gtlref

As you see above I have over egged the VCC in the 356x9 shot.

you may not be aware but there is a known conflict between uGuru & CPU-Z 1.41 which results in what you have just described.
it doesn't happen with uGuru & CPU-Z 1.40 & earlie

No I wasn't
. It did happen exactly as i said opening 2nd cpu-z to show memory tab, might just use memset in future It's odd it's just happened once.

Wasn't getting stung again took this earlier without issue. Have left them open, if it's a memory conflict odds are it will only happen on opening


Cheers for that