Great progress Gary

A nForce2 based board is really worth a try, those were the best socket a board in that time and very very stable. Nvidia did a great job with that chipset series.

About the AsRock BX test... this looks like memory problems and might be related by the VGA bios. 0.30c is still very early..

Quote:


They got it stable, otherwise the card wouldn't run that well, but memory problem is still present if you ask me. This would describe some of the errors we've seen from your reports.

Had a few minutes to fool about today and did some DirectX version swapping... Mostly the reason I dived into this was the DXdiag test had stopped working (would freeze the computer up on launch)

I had gone from DX7 (spinning colored cube) to DX8 (spinning cube with DX logo) and the machine would no longer start the DX8 test, curious. Cleary one crash had messed things up.

So I uninstalled all traces of DirectX and went right to DX 8.0, same thing. Odd. Installed DX 8.1, still no go. So for the first time I installed DX9b. Test 1 (DX7 interface) failed but instead of a crash I got an error message about not being able to open a DC? Test 2 (DX8 cube with new green X logo) ran just fine and then Test 3 (DX9 cube with green X logo) also ran fine.

Most odd indeed, passed a DX9 rendering test.

The VA6 platform has been the best yet- I'm running it with a P3-800 in a slocket adapter with 3 sticks of 128mb PC-133 ram, Win98SE with no patches or chipset drivers and a phat 400 watt Antec power supply.

I'm at an impasse at the moment, nothing to do till the drivers show some improvement or a new platform comes in. I have a Abit KT7A (KT-133A chipset) with an Athlon XP 1700+ and a 950 mhz Thunderbird. Worth a shot and a member on another forum is letting me have it for the cost of shipping.

Quote:


Yes.

Question- If running a DirectX test, why would it fall back to software? Would that not be an automatic test failure?

Running 8.1 here-

https://www.youtube.com/watch?v=0Gxf6CUa1wI

One reason I have a 4500 running next to Rampage is so I can compare the screen images to make sure that it is actually running in hardware mode, its clearly running in hardware mode unless the 4500 is running software too. Both the DX7 and DX8 tests look identical. Several of the DX tests that Gino provided a link to have RGB and Ramp emulation modes as well and you can clearly tell the difference in quality and framerate.

Thandor wrote on 24.06.13 at 12:15:14:


I agree that it looks like an intel Seattle board in the clamp test. You can also see another mb in the background. But the BIOS could be the standard intel or one used by OEMs like Compaq, Gateway that had it in their systems at the time.

A better picture in this listing:
http://www.ebay.co.uk/itm/Intel-Desktop-Board-AA-720940-213-SE440BX-2-Slot-1-Mot...

Cheers,
G1nX

Quake III Arena-

http://www.youtube.com/watch?v=-L5hXXaX28I&feature=youtu.be


Bless you Hank for being so chatty tonight, you gave me a vital clue.

Quote:



1) Hank confirmed this on my phone, some memory manegment features of Rampage were not enabled in drivers so they could focus on debugging the core. The ram is still being utilized, but not sure how much.

2) Speaking of the core, the first batch of Rampage chips needed surgical "rewiring" with an electron beam on the underside!

3) AGP functions had been disabled, this version of Rampage is pretty much a PCI device in an AGP slot. (this was my big clue)

4) Rampage was at least a year out from retail- Other chip revisions to fix issues, drivers and debugging needed to happen first.

So, based on what Hank had told me about the aspects of AGP, I disabled AGP Texture Acceleration in the DX control panel and all my tests were functional! From that point, I went into the BIOS and tried rolling the AGP apature size from where I had it at 32 mb to 64 and then down to 4mb at which point things really started rocking!

Those are very good news So the texture acceleration caused the problems, mainly due to the deactivated memory management features I assume.
With activated texture acceleration Direct3D optimizes the texture swaping from system memory to the video memory. This is one of the big advantages of AGP over PCI, beside the bus speed.

What configurations did you use to run Q3A?

I'm still a bit confused about the DirectX/Diag stuff. Of course the drivers and the card can't do D3D9 features, but they make use of the libraries. Basic direct3d commands are still present.
The layer model of Direct3D:
Win32 application -> calls Direct3D API (not the hardware layer) -> Direct3D accesses the HAL (vga driver) -> driver accesses hardware

Do you/Hank think it might be possible to activate advanced memory management?