I have a 440GX board which seems to run a Quadro 2 Pro (Essentially a GeForce 2 GTS/Ultra) with the FSB at 133MHz just fine, also ran a Radeon VE before that. XP won't boot though (Stays on the splash screen infinitely) but I have a suspicion this is the onboard SCSI crapping out when HAL kicks in and not down to the GPU as 9X starts fine along with other things - Linux for example. XP never liked the system anyway and as I now don't need it for serious Vegas work anymore I plan to go back to 98SE/2K instead of 98SE/XP eventually.

FYI, the 440GX is essentially an SMP capable 440BX, so the AGP implementation should be about the same.

I'm kinda interested in learning which graphics cards (or at least which graphics card families) will probably work fine at 89MHz AGP. Only thing I can kinda remember is that newer cards had an overall better chance of having no problems with the overclocked AGP port.

http://www.anandtech.com/show/574

GF3, GF4 - worked
I switched off sideband by RivaTurner

Tertz wrote:

GF3, GF4 - worked
I switched off sideband by RivaTurner

How about the MX's?

This is only my experience:

6800Ultra (cut for AGP 3.3V) Ok
5900 Ultra/XT Ok
5700 Ok
Ti4200 Ok

Radeon hit and miss.

You can find FX5900 cheap.

Tetrium wrote:

How about the MX's?

I did not try them. Most probably will work.

Tertz wrote:

Tetrium wrote:

How about the MX's?

I did not try them. Most probably will work.

That would be good news, as I think the MX's are more performance correct.

Tetrium wrote:

Tertz wrote:

Tetrium wrote:

How about the MX's?

I did not try them. Most probably will work.

That would be good news, as I think the MX's are more performance correct.

The MX's are the low end Geforce 4s. The Ti's are higher end.

Rhuwyn wrote:

The MX's are the low end Geforce 4s. The Ti's are higher end.

The GF4MX's were filling the budget spots in the market when the GF4 (and GF3 kinda too) was for the more performance oriented segment.
The GF4MX's weren't actually just cut-down GF4 cards though.

And I was also referring to the GF2MX's, which are somewhat slower.

The reason I asked specifically about the MX's (both the GF4MX and GF2MX variants), is because I have tons of those, they produce little heat compared to their higher-end counterparts and are much less likely to be running at their (thermal) limits. It's why I often prefer mid-range over high-end in certain builds.

Tetrium wrote:

That would be good news, as I think the MX's are more performance correct.

This needs facts is GF4 MX compatibility close to real GF2 or GF3. Also there were several MX models. One of things to try is unpatched UT99, where normal GF4 have some issues (at least on some of drivers), which GF3 and earlier have not.

Tertz wrote:

Tetrium wrote:

That would be good news, as I think the MX's are more performance correct.

This needs facts is GF4 MX compatibility close to real GF2 or GF3. Also there were several MX models. One of things to try is unpatched UT99, where normal GF4 have some issues (at least on some of drivers), which GF3 and earlier have not.

I don't know how compatible the GF4MX is compared to the GF3 and GF4. The GF2MX is of course quite a bit slower and older.
There were actually quite a few different varieties of both cards, but this mostly impacted performance.
From what I've read, GF4MX was actually more closely related to GF2 than the GF3 and GF4.

Tetrium wrote:

From what I've read, GF4MX was actually more closely related to GF2 than the GF3 and GF4.

In theory. If UT99 test will give no issue specific to GF4 on 45.23, this will be a proof. If you want to do the testing - pm me.

My Voodoo 5 5500 seemed to work at 89MHz AGP; although sometimes it wouldn't POST. I turned it back down to stock as I think the card is too valuable to risk like that.

Tertz wrote:

Tetrium wrote:

From what I've read, GF4MX was actually more closely related to GF2 than the GF3 and GF4.

In theory. If UT99 test will give no issue specific to GF4 on 45.23, this will be a proof. If you want to do the testing - pm me.

I'm not sure where you're trying to go here?

The Geforce4 MX was NV17, it was a Geforce 2 MX die shrunk with the memory controller from the Geforce 4 Ti ported over. An MX440 is neck and Neck with a Geforce 2 Ultra, while an MX420 will compete with a GTS

Tetrium wrote:

Tertz wrote:

Tetrium wrote:

From what I've read, GF4MX was actually more closely related to GF2 than the GF3 and GF4.

In theory. If UT99 test will give no issue specific to GF4 on 45.23, this will be a proof. If you want to do the testing - pm me.

I'm not sure where you're trying to go here?

You may say about your doubts more concretely.

Tertz wrote:

You may say about your doubts more concretely.

Which doubts?

candle_86 wrote:

The Geforce4 MX was NV17, it was a Geforce 2 MX die shrunk with the memory controller from the Geforce 4 Ti ported over. An MX440 is neck and Neck with a Geforce 2 Ultra, while an MX420 will compete with a GTS

And this with less power consumption 😁

Tetrium wrote:

Tertz wrote:

You may say about your doubts more concretely.

Which doubts?

You said "I'm not sure". If that was what PM, - I may help to get the game for testing, to check is GF4 MX behavior there is same as normal GF4 and hence to conclude about possibly higher MX compatibility.
Would be useful to get GF2 compatibility level, its max performance and DVI connector (missing mostly on GF2).