Abit IC7-G AGP voltage issue

[osage]

System:

P42.8C
Abit IC7-G
1 GB corsair XMS 3200 RAM
Enermax EG465P-FMA 460 Watt PS
ATI Radeon 9700 Pro
Catalyst 3.9 drivers
DX9.0b
NOTHING OVERCLOCKED

I did a lot of research on the web for this issue, but the posts I found about my particular issue in various forums were not responded to.

I was wondering if any of you experienced system builders might have experience with this issue.

Here's the deal:

About 3 times a week, when I am doing non-gaming Windows stuff on my PC, I get a video error message that jacks my resolution down to VGA and asks me to reboot.  I usually reboot with no loss of data.

But when playing AH2 beta 1995 at full detail, I sometimes get suspicious hard-boot-required crashes.

ABIT provided Winbond Hardware Doctor Software to monitor temps and voltages.

Take a look at this:


I understand VDDQ is the AGP voltage.  I have it set to 1.5v in ABit's Softmenu, but it only shows as 1.45V in pre-booot System Health bios menu.

When I actually boot, I see the abovfe in Winbond.  If I try to beef up the AGP voltage, it only increases the DDR VTT voltage to the outer limits.

This IC7-G has been kind of a lemon.  The onboard mic input doesn't work right and the fan/heatsink for the Northbridge chipset has had to be replaced.

Is the voltage reported by Winbond accurate?  Even the BIOS report is .5 volts low for VDDQ.  I paid for a good power supply, so it's not a likely suspect, especially since all the other voltages are OK.

Any thoughts?

Thanks in advance!

[bloom25]

The 1.5V is not supplied by the power supply, but rather obtained on the motherboard by regulating the 5V and 3.3V supply.  (That goes for every motherboard.  ATX computer power supplies provide -12V, -5V, 3.3V, 5V, 5Vstandby, and 12V.  The negative voltages are rarely used and not very critical.  (The serial port is about the only thing that may make use of the -12V supply.  I can't think of anything that uses -5V in new systems.)  The 3.3V, 5V, and 12V outputs are by far the most critical.  The CPU core voltage is obtained through switching mode voltage regulators on the motherboard itself, as is the 2.5V used for DDR memory, and the AGP signalling voltage.

First off I can see that your CPU is running significantly too hot.  If that reading is accurate (64C) and is the idle, rather than full load, temperature you are certainly going to run into serious stability issues when taxing the CPU at all.

Vddq is not the supply voltage to the AGP card, but rather the signaling voltage between the card and the motherboard Northbridge.  AGP 1x/2x cards use 3.3V (and a significantly different connector).  AGP 4x signaling voltage is 1.5V and AGP 8x is 0.8V.  If that reading is accurate, 1.25V, that is too low for reliable AGP 4x signaling and too high for AGP 8x.  That voltage is provided by voltage regulators on the motherboard itself.  Vtt (at least in my experience) is a reference voltage used to compare against.  This voltage reference is probably a typical silicon bandgap reference, which should output around 1.3V.

Personally, I'd RMA the motherboard and get a new one.  You also have a CPU heating problem to deal with.

[osage]

Dangit Bloom, I was hoping you weren't gonna say "RMA."

The 9700 is 8x AGP and is set that way by default in the BIOS.

Do I have to clean reinstall the O/S and everything else if I swap out the board for another IC7-G?

I spent many hours tweaking this ****.

--Osage

P.S. that temp is almost normal for an idle 2.8 C chip (58-60 deg C).  I run setiathome and had just turned it off.  It starts to throttle down at 90 automatically, anyway.

[214thCavalier]

That version of winbond reports the Abit VDDQ and VTT incorrectly.

The VDDQ is as you say the vid cards voltage BUT they got it wrong.

In that version of winbond the VDDQ is actually showing the VTT voltage and of course the VTT display is really showing the VDDQ voltage.

Keeping that in mind you will see when you check the readings in bios it will agree with the above statement.

[DAVENRINO]

Originally posted by osage
Do I have to clean reinstall the O/S and everything else if I swap out the board for another IC7-G?

P.S. that temp is almost normal for an idle 2.8 C chip (58-60 deg C).  I run setiathome and had just turned it off.  It starts to throttle down at 90 automatically, anyway.

You shouldn't have to reset anything except the BIOS settings and jumpers.

I don't know much bout Intel s but I would be panicked if my Barton ran that hot.

DJ229 - AIR MAFIA

[bloom25]

If Vddq and Vtt are reversed, those voltages are fine for a video card running at an AGP 4x transfer rate.  Since Osage said Vtt changed with bios adjustments to AGP signaling voltage, it would agree with the software having the two reversed.

As for the replacement board, no you won't have to do anything other than possibly change a few things in the bios.

I still think 64C idle is too hot for that chip.  The last 2.6C P4 I did ran 45C idle with the stock heatsink.  By chance, did you not install an exhaust casefan below the powersupply?

Edit:  I figured I'd look up what Vtt is used for.  It looks like Vtt is a DDR voltage reference and should be 1/2 of the DDR supply voltage.  I.e. If the DDR voltage is 2.5, Vtt should be 1.25.

[WhiteHawk]

Looks like the vcore is too low???  I have an Ic7-max3 and had stability issues until I upped the vcore to 1.55V.  I overclock now and have it set to 1.65V.
  But you may want to find out a solution to your heat problerm (if its a problem) before you up the vcore.
  Ive read anything from, 'over 60C is too hot', to 'p4's are fine up to 100C.'  I have my warning set at 67.5C.  
  I run 42C idle and push 65C under AH load.
Here is a previous discussion...


Personally i think Abit are measuring incorrectly as temp probes seem to confirm its high by approx 10 degrees C.

On abit ic7-max3 39.5°
same set-up (fans, ram, cpu) on gigabyte 8KNXP board was 29° so im reluctant to believe how accurate this board reads ...


2.4 oc to 3.0 IDLE
Abit Max3 39.6 Case 27
Asus p4c800 26 Case 27

Stop panicking get a better heatsink fan combo than standard, and overclock the SOB.

I know banana had problems with his install but most people don't and those who do get past it, MrBlack was happily overclocking the crap outa his succesfully, (same as many others are) until his cooling killed it.
Point is its not all bad and some of the boards have no problems anyway.

[osage]

It's been reported that that ABIT's CPU temperature calculation algorithm for the IC7-G gives a high reading.

I have the stock CPU fan and a top of the line Lian-Li case that should do a good job.

One problem is that the new landlord has cranked up the heat in my radiator-heated building so my ambient temperature is hotter than it is in the summer (with air conditioning).  I'm not complaining, the old landlord was a cheapskate and it sometimes got pretty cold.I'm trying to figure out a system of cracking windows to keep it cooler in here.

Recently the damn chipset replacement fan just pulled a mounting hoop and fell over onto the vidcard, effectively killing it.  It only runs at about 1800 rpm now.  Another one is on order.

The problem could well be that the GPU is getting too hot.

[osage]

Originally posted by 214thCavalier
That version of winbond reports the Abit VDDQ and VTT incorrectly.

The VDDQ is as you say the vid cards voltage BUT they got it wrong.

In that version of winbond the VDDQ is actually showing the VTT voltage and of course the VTT display is really showing the VDDQ voltage.

Keeping that in mind you will see when you check the readings in bios it will agree with the above statement.

Yep, looks like that is correct.  Thanks!