I had an Asus P2B and I found that power hungry AGP cards wouldn't work in it. I think the best card I tried that worked was my Quadro 2 (same as a Geforce 2 Ti). It wouldn't be too surprising if that was the issue. Do you have another motherboard you can try the video card with?

I am using a MSI Geforce4 Ti4200 on a P2B. I had a lot of stability issues during benchmarking application runs that were resolved by installing ForceWare 45.23 drivers.

Out of curiosity, which power supply are you using with that system

Antec SmartPower SL350 350W.

Asus is one of the brands "wisely" feeding 3.3V to AGP from its internal DCDC converter (L16 inductor), same one powering RAM 🙁 Its good for period correct graphic cards only, so up to ~2001. Higher end might need modding to bypass onboard 3.3V for ATX source.

Hey folks.

Thanks for all the feedback, much appreciated!
Apologies for the brain fart, but the mobo I have is a Gigabyte GA-6BXC rev2.0. Not sure why I wrote Asus, probably because the card is Asus and it has a brown-gold-ish colour, like my motherboard... mental associations... 🤣

Something I forgot to mention, the fan actually spins, so I don't think it's an overheating problem. Also, the black screen issue happens always at the same time; I can see the first helicopter test in 3Dmark2000, it fails almost at the end of the second helicopter test.
The highest driver release I tried was version 30.82 I believe, never went to the 4x .xx series.
My power supply is a Corsair semi-modular 450W.

I actually bought two Geforce4 Ti 4200, waiting for the second one. I will test again as soon as I get this other card.
In the meantime, if I have the time, I will give this one another try in a Pentium 4 system. The P4 has an FX5900 running the 45.23 drivers in Win98 as well.

Thanks!

https://theretroweb.com/motherboards/s/gigaby … ga-6bxc-rev-2-0 is even worse, as seen on teh picture there is LDO right next to AGP slot
LX8384 3.4V at 5A = ~17W max power to AGP slot. You might need to solder a wire between ATX 3.3V and AGP 3.3V, lifting output leg of LX8384 is optional.

rasz_pl wrote on 2024-04-24, 06:14:

https://theretroweb.com/motherboards/s/gigaby … ga-6bxc-rev-2-0 is even worse, as seen on teh picture there is LDO right next to AGP slot
LX8384 3.4V at 5A = ~17W max power to AGP slot. You might need to solder a wire between ATX 3.3V and AGP 3.3V, lifting output leg of LX8384 is optional.

Is this a wise mod?

rasz_pl wrote on 2024-04-24, 06:14:

https://theretroweb.com/motherboards/s/gigaby … ga-6bxc-rev-2-0 is even worse, as seen on teh picture there is LDO right next to AGP slot
LX8384 3.4V at 5A = ~17W max power to AGP slot. You might need to solder a wire between ATX 3.3V and AGP 3.3V, lifting output leg of LX8384 is optional.

The 6BXC 2.0 revision is actually OK.

Gigabyte statement regarding 6BXE, but it's the same for 6BXC, which also has the "Voodoo 3" jumpers:

"On our old version M/Bs, for example GA-6BXE PCB 1.9, we used LVD regulator to generate 3.3V power for AGP card. Because Voodoo3 card drains too much power from 3.3V, the LVD regulator can't stand too much current and offer enough current to Voodoo3. On GA-6BXE PCB 2.0, we implement JP18 and JP19 connectors which directly connect the 3.3V from power-supply to AGP device. At beginning, the system uses 3.3V power from LVD regulator. When the regulator can't offer enough current for AGP card, the power-supply will provide the 3.3V power to AGP card immediately."

bloodem wrote on 2024-04-24, 08:44:

The 6BXC 2.0 revision is actually OK. […]
Show full quote

rasz_pl wrote on 2024-04-24, 06:14:

https://theretroweb.com/motherboards/s/gigaby … ga-6bxc-rev-2-0 is even worse, as seen on teh picture there is LDO right next to AGP slot
LX8384 3.4V at 5A = ~17W max power to AGP slot. You might need to solder a wire between ATX 3.3V and AGP 3.3V, lifting output leg of LX8384 is optional.

The 6BXC 2.0 revision is actually OK.

Gigabyte statement regarding 6BXE, but it's the same for 6BXC, which also has the "Voodoo 3" jumpers:

"On our old version M/Bs, for example GA-6BXE PCB 1.9, we used LVD regulator to generate 3.3V power for AGP card. Because Voodoo3 card drains too much power from 3.3V, the LVD regulator can't stand too much current and offer enough current to Voodoo3. On GA-6BXE PCB 2.0, we implement JP18 and JP19 connectors which directly connect the 3.3V from power-supply to AGP device. At beginning, the system uses 3.3V power from LVD regulator. When the regulator can't offer enough current for AGP card, the power-supply will provide the 3.3V power to AGP card immediately."

I have the same board, and I run it with a Ti4200. It runs fine. However, a Voodoo 3 does not.

appiah4 wrote on 2024-04-24, 10:30:

I have the same board, and I run it with a Ti4200. It runs fine. However, a Voodoo 3 does not.

That is quite strange. I have 4 x 6BXC rev 2.0 boards and also 2 x older rev 1.9 boards (which don't have the jumpers), and, still, Voodoo 3 3000 cards work fine with all of them.
What issue are you facing, more specifically? AGP power delivery problems usually result in hard reboots or freezes in 3D apps/games.

Either way, the Voodoo 3 has at least 2 times lower power consumption than the Ti 4200, so if the latter works, the former should also work (from a power delivery standpoint).

bloodem wrote on 2024-04-24, 08:44:

The 6BXC 2.0 revision is actually OK.

implement JP18 and JP19 connectors which directly connect the 3.3V from power-supply to AGP device.

Hah its right there on top of the board, JP 12 JP 13 Voodoo jumpers 😀
I also didnt look too hard at the board the first time, it actually has three LX8384 LDOs 😮 , that would put it at ~50W of 3.3V, weird that they still needed those jumpers.

appiah4 wrote on 2024-04-24, 08:07:

Is this a wise mod?

as bloodem pointed out this board already has this mod build-in :] silkscreen text describing it is located on the upper edge

First of all, just wanted to clarify I'm quite happy with the Geforce3 I have in this build. Using the Geforce4 is just for science and entertainment 😁

Oh yeah, the Voodoo jumpers on the motherboard, forgot about those.
Now that I think about it, it's possible that I didn't do a comprehensive test with the Voodoo3 when I had it in this machine.
At any rate, I will add enabling the motherboard's 'voodoo3 mode' to the testing, and perhaps try again with the Voodoo3 as well.

On the power front, it seems the Geforce3 has a TDP of 40W: http://hw-museum.cz/vga/172/gainward-geforce3
While for the Geforce4 ti4200 is 33W: http://hw-museum.cz/vga/179/msi-geforce4-ti-4200

I'm also thinking that perhaps the Geforce4 doesn't like the bus speed. The CPU I have uses a 133MHz FSB, which in this mobo means that the AGP is bus overclocked.

bloodem wrote on 2024-04-24, 12:31:

That is quite strange. I have 4 x 6BXC rev 2.0 boards and also 2 x older rev 1.9 boards (which don't have the jumpers), and, sti […]
Show full quote

appiah4 wrote on 2024-04-24, 10:30:

I have the same board, and I run it with a Ti4200. It runs fine. However, a Voodoo 3 does not.

That is quite strange. I have 4 x 6BXC rev 2.0 boards and also 2 x older rev 1.9 boards (which don't have the jumpers), and, still, Voodoo 3 3000 cards work fine with all of them.
What issue are you facing, more specifically? AGP power delivery problems usually result in hard reboots or freezes in 3D apps/games.

Either way, the Voodoo 3 has at least 2 times lower power consumption than the Ti 4200, so if the latter works, the former should also work (from a power delivery standpoint).

I have doublechecked and my prblematic board is actually a 6OXT. Oops.

kahuna wrote on 2024-04-24, 23:17:

I'm also thinking that perhaps the Geforce4 doesn't like the bus speed. The CPU I have uses a 133MHz FSB, which in this mobo means that the AGP is bus overclocked.

I suspect that the VBIOS of your GeForce4 Ti card supports sideband addressing and it is being enabled. That could be the cause of the instabilities with the AGP @ 89 MHz.

Thanks for the suggestions!

I don't know what it is, but the card is perfectly fine.
I plugged this Geforce4 Ti4200 in a Pentium 4 build: P4 3.2GHz Northwood with a Gigabyte motherboard based on the SIS 655FX + 964 chipset, it works great here.

I also tried with yet another Gerforce4, in this case a Ti4600 (same chipset as the Ti4200, AGP 4x).
The i440BX/C3 based system didn't even show anything on screen, nor over DVI, neither over VGA.
No problems whatsoever on the Pentium 4 build above.

Glad the Geforce3 works great in the i440BX/C3 machine!