VOGONS


First post, by assasincz

User metadata
Rank Member
Rank
Member

Hi all,

First of all, I hope you all are staying safe and healthy,

Second of all - I am currently working on a AMD K6-2 project built , the core of which are:

  • 256MB SD RAM (1x128MB+2x64MB, all PC100)
  • SD-to-IDE adapter and 32GB card

This has to be my most problematic build yet.

I already managed to solve a CPU FSB/multiplier issue - when set top CPU's "proper" config 4x100MHz, the system was greatly unstable (memory errors, display and text artefacts, random reboots and blank screens). I just could make the windows to install and run properly.
After some troubleshooting (swapping all components including the CPUs, setting SD RAM clock to FSB/AGP clock), I eventually figured out that the system had issues with FSB set to 100MHz and 95MHz, although they are properly suppoorted according to manual. Any FSB clock below that (83/75/66/60MHz) worked fine and the system was stable. So there is a 10-switch DIP panel controlling the CPU settings. After staring into the manual for a while, I deduced that from the DIP array 1-6 that the switches 1-3 control the FSB and 4-6 control the multiplier. All possible seetings for FSB switches are already described in the manual. However the manual only describes DIP switch settings for multipliers 2.5x, 3x, 3.5x, 4x, 4.5x. OK, but I wanted to get to the 400MHz level with the CPU, so after some more fiddling and experimenting, I found the positions for multiplier switches to set 5x, 5x5 and 6x. Eventually, I set the CPU to 5x83=415MHz and the system ran just fine, but no higher clock than that. So one problem solved that I though to share with you.

Then came the GPU time. For purposes of this build, I got myself a Manli TNT2 Pro 32MB AGP as I thought this would fit the build nicely. Running Win98SE however, after installing the Detonator drivers (any version) I was unable to get the system to boot at all. Bit of a let down, I tried other loose AGP cards I have on hand that could fit into the slot to see if the problem persists:

  • Geforce4 MX 440 (MSI MS-8866) - SAME ISSUE as with the TNT2 Pro
  • Radeon 9250(Sapphire Atlantis) - WORKED FINE
  • Geforce4 Ti 4800SE (Leadtek Winfast A280) - SAME ISSUE as with the TNT2 Pro

Needless to say, I also tried various PCI graphics cards, these all worked fine, including the TNT2 M64 (Guillemont Gamer Cougar). I sourced MVP3 drivers from Phil that include AGP drivers, but this did not help.

So in conclusion, I could not get only any nVidia AGP card to work properly on the system, no matter the driver version(just 640x480/16color retard mode). Only ATI AGP card and PCI graphics cards worked.

During all of the above process, I tried fiddlign with CPU and SD RAM clock settings to see if these could contribute to the issue, but they do not.

I thought to try with Windows 2000, but for the love of god, I could not get the installation to work and finish properly. I will keep working on this but my suspitions are that the issues will persist.

Do you have any suggestions what could I try next? I can't figure out why the system is so picky with nVidia AGP cards Have any of you ever encountered anything of this sort?

Man, how DID we survive the 2000s?

The attachment mvp3-at.jpg is no longer available

Reply 1 of 30, by Baoran

User metadata
Rank l33t
Rank
l33t

I thought it was common knowledge that there was big problems with AGP cards in super socket systems and that is just how it is. Most recommend voodoo 3 cards because they don't use any AGP features even when using agp version of the card.
a PCI card or voodoo 3 card is the way to go.
We survived 2000s because we were using Penium 2 or Pentium 3 by then and those generally work fine with agp cards.

Reply 2 of 30, by Mamba

User metadata
Rank Oldbie
Rank
Oldbie
Baoran wrote on 2020-03-20, 07:11:

I thought it was common knowledge that there was big problems with AGP cards in super socket systems and that is just how it is. Most recommend voodoo 3 cards because they don't use any AGP features even when using agp version of the card.
a PCI card or voodoo 3 card is the way to go.
We survived 2000s because we were using Penium 2 or Pentium 3 by then and those generally work fine with agp cards.

Woa...

It is not so bad to play with MVP3 and AGP, trust me.
It is possible that your motherboard has issues, for what I can read, it is highly the case.
None of the card you are using should cause any major trouble if you stay under 100MHz of FBS, assuming via 4in1 are installed and everything is ok for other components.
Could you see if this problem occurs at 66MHZ of FSB also? Downclock you CPU.
It is better for the tests to use a normal HD rather than any sort of adapter.
You never know.

Edit: I saw you already played with clocks.
Do you tried to lower agp to 1x andd or change agp aperture size?
My bet is on a faulty motherboard.
Never had such troubles with nvidia cards, pre FX in special, with socket7 systems
It must be something wrong with your MB.

Reply 3 of 30, by Baoran

User metadata
Rank l33t
Rank
l33t

From my experience agp is quite unstable and there can be big differences even between motherboards using the same chipset and it can be increased if some of the capacitors on the motherboard are in not so great condition. Even the psu can affect how well some cards work from my experience of changing psu in one of my super socket 7 systems in the past.

Reply 4 of 30, by Garrett W

User metadata
Rank Oldbie
Rank
Oldbie
Mamba wrote on 2020-03-20, 09:38:

Never had such troubles with nvidia cards, pre FX in special, with socket7 systems
It must be something wrong with your MB.

While that may be the case, it is indeed common knowledge that early non-Intel AGP chipsets don't play nice with AGP cards and are very temperamental. Sure though, 20+ years can certainly cause capacitors to go bad f or example, but there are countless cases of people having issues with ATi, Nvidia, Matrox and all sorts of AGP cards on these boards both from way back then on Usenet and Anandtech/Tom's Hardware/etc forums and even nowadays on Vogons. 3Dfx or PCI cards are suggested for greater stability. Any reason in particular you want an NVidia card in there though?

In all honesty, a K6-2 400 is probably around the ballpark of a PII 233 or 266 in terms of gaming performance and as such, that TNT2 is wasted in there. Even a single Voodoo2 would have no problem saturating that CPU. If the Radeon card produced no issues, then it may be a good alternative, although I am somewhat cautious of ATi drivers from that era 😀.

Reply 5 of 30, by assasincz

User metadata
Rank Member
Rank
Member

Thank you for all your kind feedback.
Just a side note, the I have already recapped the whole Rhino mainboard few weeks ago.
Yes I have tried messing around AGP settings in BIOS, to no avail.
I will be fiddling around some more but from what you describe, it will likely be hit-and-miss until something eventually somewhat works...

Reply 7 of 30, by Mamba

User metadata
Rank Oldbie
Rank
Oldbie

Guys,
I am aware that agp is far from perfect on via chipset.
But what the OP describes frankly is absolutely not common behaviour.

I insist, there is something wrong in the motherboard.

Never ever experienced something like that, apart when the motherboard was the problem.

Could you look into the possibility to cross flash with the DFI bios?
This possibility seems intriguing and you should really check this.
Have you a programmer just in case?

And, are you using a atx psu?

Reply 9 of 30, by assasincz

User metadata
Rank Member
Rank
Member
Mamba wrote on 2020-03-20, 12:37:
Could you look into the possibility to cross flash with the DFI bios? This possibility seems intriguing and you should really c […]
Show full quote

Could you look into the possibility to cross flash with the DFI bios?
This possibility seems intriguing and you should really check this.
Have you a programmer just in case?

And, are you using a atx psu?

I did not find the BIOS updates anywhere for download actually for this board.
And even if I did, I have no programmer, unfortunately,

I am using an ATX PSU through ATX-to-AT adapter...

Reply 10 of 30, by Garrett W

User metadata
Rank Oldbie
Rank
Oldbie

Lol, I have a DFI P5BV3+ and it is an absolute mess of a board. I've had stability issues even with PCI cards, so I can totally relate to the OP here.

Reply 11 of 30, by Mamba

User metadata
Rank Oldbie
Rank
Oldbie
Garrett W wrote on 2020-03-20, 13:01:

Lol, I have a DFI P5BV3+ and it is an absolute mess of a board. I've had stability issues even with PCI cards, so I can totally relate to the OP here.

Oh well,

if it is the case, coupled with the fact that you don't have a programmer, the crossflash could be too risky.
Bios updates I am sure can be found if you want to try.

http://web.inter.nl.net/hcc/J.Steunebrink/k6plus.htm (find the DFI P5BV3+), there are 3 different versions of this beta BIOS. One for each Rev (A, B, or C) of this board. You should check which one is the closest to yours. I wuld exclude A since it has EDO slots. Maybe it is the C.
Link is broken, here the good one (go to the DFI bioses) ftp://ftp.spez.com.ua/driver/mb/via/

https://www.manualslib.com/manual/816826/Dfi- … lus.html#manual

I see there are different revisions for the MB, just like the DFI counterpart. Which one is your? Is it up to date from bios side?

Again, are you using an ATX PSU properly? I see that there should be even an ATX power connector close to the AT one, so you definitely should get rid of the adapter.

Reply 12 of 30, by Deksor

User metadata
Rank l33t
Rank
l33t

Well the thing is, octek's bios is identical to dfi's. On mine I flashed the latest bios modded to support k6-2/3+ CPUs and it works perfectly.

I suspect that octek was one of DFI's subsidiary. The only difference there is, is just a sticker saying octek on the chipset.

Trying to identify old hardware ? Visit The retro web - Project's thread The Retro Web project - a stason.org/TH99 alternative

Reply 13 of 30, by Mamba

User metadata
Rank Oldbie
Rank
Oldbie
Deksor wrote on 2020-03-20, 14:55:

Well the thing is, octek's bios is identical to dfi's. On mine I flashed the latest bios modded to support k6-2/3+ CPUs and it works perfectly.

I suspect that octek was one of DFI's subsidiary. The only difference there is, is just a sticker saying octek on the chipset.

It is your call,
But if I were in you, I would try the crossflash.
I don't see why it should not work, the mb are the same.
And you could put a k6 plus chip, a very nice boost.

Reply 15 of 30, by Mamba

User metadata
Rank Oldbie
Rank
Oldbie
Deksor wrote on 2020-03-20, 16:44:

That's what I did ^^

I flashed DFI's bios (modded by someone) on my "octek", and it worked.

I was referring to the OP.
Could you confirm the version you flashed?
A B or C?
Did you have the same revision of the OP?

Reply 17 of 30, by Mamba

User metadata
Rank Oldbie
Rank
Oldbie
Deksor wrote on 2020-03-20, 18:41:

It's REV C

i don't know which rev the op have. We'll see.

So you had a mvp3-at rev b+ and flashed the dfi bios rev "c", correct?
Could you upload it?

Reply 19 of 30, by Mamba

User metadata
Rank Oldbie
Rank
Oldbie
Deksor wrote on 2020-03-20, 22:34:

No I think my board is equivalent to dfi's p5bv3+ rev c, I just flashed the modded bios which is for I don't know which board

I fail to understand... You flashed the modded bios, ok.
And what motherboard you have (originally)?
Could you upload that bios?