Announcement

**Rags** · 26 December 2001, 19:32

Actually it is both. The G400 will run with the 1.5 AGP spec (AGP 2.0). It was the first out that did this all automagically.

Rags

**KevinST** · 27 December 2001, 02:40

Hi all,

I don't doubt that the G400 will work at 1.5v - I have a friend that has it working like that.
However, pin 2A is NOT connected to 0v (GND) as it is supposed to be according to te Intel AGP spec. So... and motherboard (such as the P4B266) that tries to detect if the AGP card is 1.5v compatible will fail because the pin is not tied low.

**Rags** · 27 December 2001, 06:19

Originally posted by KevinST
Hi all,

I don't doubt that the G400 will work at 1.5v - I have a friend that has it working like that.
However, pin 2A is NOT connected to 0v (GND) as it is supposed to be according to te Intel AGP spec. So... and motherboard (such as the P4B266) that tries to detect if the AGP card is 1.5v compatible will fail because the pin is not tied low.

Wrong. It's an auto switching card. That means it will change the state of the pin depending on the motherboard it's in. It works, and works very well. According to the Intel spec, this perfectly fine, because of when the poll to the card occurs is what matters, not what you read with your DVOM sitting outside the box. The G400 will not fail to be detected as a 1.5v card. It should be detected as such everytime.

Rags

**KevinST** · 27 December 2001, 07:34

Originally posted by Rags

Wrong. It's an auto switching card. That means it will change the state of the pin depending on the motherboard it's in. It works, and works very well. According to the Intel spec, this perfectly fine, because of when the poll to the card occurs is what matters, not what you read with your DVOM sitting outside the box. The G400 will not fail to be detected as a 1.5v card. It should be detected as such everytime.

Rags

Unfortunatly, on the P4B266 it is failing to be detected as a 1.5v compatible card, whereas my GeForce2GTS card is detected as a 1.5v compatible card!

On the P4B266, there is detection circuitry that detects the 1.5v compatibility BEFORE the chipset is powered up - and probably before the BIOS is booted... and so before any commands are sent to the AGP card. If the MoBo in this dormant state detects that the AGP card is not 1.5v compatible (and the only way it can do this from what I understand of Intels spec, is to reference the TYPEDET# pin) then the MoBo isn't even allowed to boot up - thus protecting the chipset and the AGP card from damage.

I can't see any reason why Matrox would want to change the state of the TYPEDET# pin - the spec states that 1.5v compatible cards should have this tied low - non 1.5v cards should have this floating.

**Wombat** · 27 December 2001, 08:51

I can't see any reason why Matrox would want to change the state of the TYPEDET# pin - the spec states that 1.5v compatible cards should have this tied low - non 1.5v cards should have this floating.

You already quoted the answer yourself:

"In the Special interface group, TYPEDET# indicates whether the interface is 1.5v or 3.3v. If TYPEDET# is open (not connected to any power rail or signal), the interface is 3.3v signalling. If TYPEDET# is shorted to ground, the interface is 1.5volt."

but it's also said again in Section 5.5: "All 3.3 volt cards leave the TYPEDET signal open. All 1.5 volt cards tie this signal hard to ground."

So the AGP spec does not allow a card to inticate that it is 1.5V and 3.3V capable. That would have required two pins. A card can indicate that it is either a 1.5V or a 3.3V card. The G400 detects what it needs to be, and sets the signal accordingly. I think it is the only card to do so. Some other cards were capable of running at both specs, but you had to set a jumper on the video card.

Check sections 4.3.8, 5.4.4 and 5.4.6
Just how do you know how your MB's circuitry works?

**Greebe** · 27 December 2001, 08:53

I can't see any reason why Matrox would want to change the state of...

Compatability reasons?

**KevinST** · 27 December 2001, 09:11

Nah - the TYPEDET# pin is used to tell the MoBo if the AGP card is 1.5v compatible.

It's not a data line that is used to set anything on the AGP card... it's purely a "flag" that's either connected to GND or left floating. If this pin is shorted to GND then it's capable of working at 1.5V, if it's open then it will not work at 1.5V - that's what the Intel AGP spec says!

Nowhere in the spec does it say that the TYPEDET# should not be tied low if the card is working at 3.3v... it's just there to show that the AGP card can work at 1.5v if required.

**Wombat** · 27 December 2001, 09:50

Originally posted by KevinST
Nah - the TYPEDET# pin is used to tell the MoBo if the AGP card is 1.5v compatible.

It's not a data line that is used to set anything on the AGP card... it's purely a "flag" that's either connected to GND or left floating. If this pin is shorted to GND then it's capable of working at 1.5V, if it's open then it will not work at 1.5V - that's what the Intel AGP spec says!

Nowhere in the spec does it say that the TYPEDET# should not be tied low if the card is working at 3.3v... it's just there to show that the AGP card can work at 1.5v if required.

What is wrong with you? It says right here: Section 5.5: "All 3.3 volt cards leave the TYPEDET signal open. All 1.5 volt cards tie this signal hard to ground."

What do you mean "work at 1.5v if required?" Required by what? If you were right, then running at 3.3V should be an option for all 1.5V cards. It's not. It wouldn't even make sense.

**Rags** · 27 December 2001, 12:35

Kevin,

You really need to quit being so argumentative. First of all, the spec allows for different types of AGP 2.0 interfaces. One is an autodetecting motherboard that detects whether or not a card is AGP 1.0 or 2.0 (3.3 or 1.5). Another is a pure AGP 2.0 compliant card requirment, where the signal has to be 1.5 V. --this one has a different slot so older cards will not fit.

At any rate, the G400 runs at 1.5 when in a 2.0 slot whether auto or continuous interface. The pin that is pulled to ground is done so automatically depending on the slot it's in.

Rags

**KevinST** · 28 December 2001, 04:18

All - sorry if I'm coming accross argumentative - it's not intentional. I'm not trying to gain anything by partaking in this conversation - just trying to put across some ideas that may be of help to other people.

As far as I know the P4B266 is the only MoBo that tries to detect if the AGP card is 1.5v capable before applying power to the chipset and AGP card, thus avoiding damage to the chipset and AGP card. Apparently, there are several AGP cards out on the market that have the 1.5v cut out on the edge connector, but are not 1.5v compatible... and so the MoBo and AGP card have both been destroyed.

It would be interesting to learn how the G400 learns what slot it's in "The pin that is pulled to ground is done so automatically depending on the slot it's in. " the only way it could do that is detect the voltage... but by that time, the MoBo would be destroyed if the AGP card wasn't actually 1.5v capable.

Anyway, if this is going to turn into a flame war then I won't continue contributing to the discussion.
As a parting word though - it appears that currently the G400 is the only AGP card that doesn't appear to work in the P4B266. Maybe as more of these boards are sold other AGP cards will fail to work proving that Asus have read the spec wrong and I've also been wrong. But maybe the G400 will be the only card that doesn't work.

Announcement

G400 and 1.5v AGP spec

G400 and 1.5v AGP spec

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment