Okay, I'm starting this new thread to avoid getting off-topic in the games area... ;)

We're discussing the possibility of SLI on Voodoo 1 cards that aren't normally designed to have SLI connectors... :D

Oh and Janskjaer, you were correct about my Orchid Righteous 3D... ::)

The card is in fine condition, and so is the manual and guaruntee card inside. But there was no box or passthrough cable (lucky that I built a passthrough cable that works as a spare!) ;D

If there was a possibility of rigging up SLI for Voodoo 1's - someone would need to work out how the SLI on Quantum Voodoo 1's worked and to WHERE on the card each of the SLI connector pins was routed... :o

...complex I know, but I could probably pull off the soldering as the surface mount components aren't too big... :P

Ah whatever! ::)

This is true  ;) although we may need a schematic of the PCB in order to find the circuit lines that use the SLI.  Not an easy task. :(

I thought it may have been possible to get some SLI functionality of using two Hercules boards, that had the port for the daughterboard for an extra memory boost, but doing this would not work as the port does not no how to seperate the SLI functions for one card to do the vertical, and one card to do the horizontal set of executions on the screen. So this can be ruled out!  :(

Hmm...a little off track there my friend! ::)

The mem upgrade ports wouldn't help as they are for...well...memory! :P

But you are correct about the fact that we'd need schematics of the 'normal' Voodoo 1's in order to find out where the SLI 'lines' for each card would be connected. :(

Trial and error simply isn't possible - unless you have a LOT of Voodoo 1 cards! ;)

Like a SERIOUSLY LARGE number of Voodoo 1 cards! :o

Heheheh! ;D

Hey dude,  :)
Yeah, the thing about the memory ports was the point I was trying to make, in that they would be no good for any kind of SLI function.  They would need a totally new adapter port on the card for an SLI belt connection.

Schematics are now going to be hard to come by. Although I have reason to believe that any SLI lines would feature near the memory port, as that would be the best possible location to have a 'link up'.

Our best chance would be to get a really good high res photo of a Quantum 3D board to see if they marked out and highlited the SLI circuit lines on the board.

What you are discussing is, in theory quite correct.  It is definately possible to link two Voodoo1 chipsets for SLI operation, it is also possible to add a second texelfx processor to each pixelfx FBI.  In practice however this is not simply a case of 'linking wires', and requires driver support in order to function.

How true!  :)

I must admit, I never thought about that!  :-[

Although, it would be a case of physically setting up the hardware aspects of the card, first, wouldn't it?

It would be then best to design the drivers as the secondary procedure, wouldn't it?

Is this how it the task would be approached?

Well, going to the trouble of connecting two voodoo1's together in SLI, which would involve quite some modification, would be largely pointless unless the drivers actually worked.  SLI is software based, the actual 3dfx device driver for a particular software API has to support it in order for it to function.  It only functions when the Glide/Direct3D environment is active.  So without this support, the second card would be ineffective.

So the first stage would probably be the drivers.  Also you've got to ask yourself is it worth it?  The Voodoo2 fulfills this role, and itself is no more than a Voodoo1 with a second TMU.  It is technically feasible to SLI Banshee and Avenger chips also.  Though, again, not as simple as 'joining wires' at certain points.  But worthwhile? Not really IMHO.  If an individual wants the best balance of performance and cost from passthrough solution, then Voodoo2 SLI is the best bet, this is essentially the same as, and better than, what you're describing here anyway, so it's not worth doing as 3dfx have done it already. ;)

Okay, the topic WAS only theoretical ya' know... ::)

...so let's keep the rumour mill churning! ;)

DRIVERS: - This issue could be addressed by modifying the Quantum Voodoo 1 drivers to allow SLI with ANY Voodoo 1 card. :D
(this procedure would be similar to what has been done with the Creative MiniGL - it was modified to run on Banshee first of all, then for specific Banshee models) ;)

Also I agree that the Voodoo 2 SLI setup is the best to use as it's already working. ::)

The point I was getting at is that I'm a CRAZY MODDIN' bloke who'll modify just about anything! :o

(my XBOX is only one example of this!) ;D

So despite the fact that it's rather SILLY to even think about doing this - it's just a bit of MODding fun ya' know! ;)

CHILL! 8)

Oh ya' and I REALLY shouldn't have started this topic so early in the morning! ::)

Check out the typo: - Vodoo, heheheh! ;D :P ::)

Quote:

Okay, the topic WAS only theoretical ya' know... ::)

I had guessed. ;)


Quote:

DRIVERS: - This issue could be addressed by modifying the Quantum Voodoo 1 drivers to allow SLI with ANY Voodoo 1 card. :D

Well how's your C? :P


Quote:

(this procedure would be similar to what has been done with the Creative MiniGL - it was modified to run on Banshee first of all, then for specific Banshee models) ;)

MiniGL is a standalone stripped down ICD designed for the ID Software quake engine.  It's not really comparable with this subject matter. :)


Quote:

So despite the fact that it's rather SILLY to even think about doing this - it's just a bit of MODding fun ya' know! ;) CHILL! 8)

Too right...  8)

Quote:

Well how's your C? :P

I'm not too great at coding, but what the heck...which brings us to the next point... :P


Quote:

MiniGL is a standalone stripped down ICD designed for the ID Software quake engine.  It's not really comparable with this subject matter. :)

Exactly, it's not NORMALLY comparable...but first of all I said this was THEORETICAL. ::)

Second of all, if you were to modify drivers you'd probably only need to hack the GLiDE dll's so they would work with a non-quatum card. :D

After all, there are already modified drivers for Windows XP and Voodoo 1. So just hack the Quantum GLiDE dll's so they work with normal Voodoo 1's and then experiment with the drivers to get them to work... :D

Like I said it's all theory...and should be taken with a pinch of salt. ::)

wrote on 25.04.04 at 15:38:43:

I'm not too great at coding, but what the heck...which brings us to the next point... :P Exactly, it's not NORMALLY comparable...but first of all I said this was THEORETICAL. ::) Second of all, if you were to modify drivers you'd probably only need to hack the GLiDE dll's so they would work with a non-quatum card. :D After all, there are already modified drivers for Windows XP and Voodoo 1. So just hack the Quantum GLiDE dll's so they work with normal Voodoo 1's and then experiment with the drivers to get them to work... :D Like I said it's all theory...and should be taken with a pinch of salt. ::)

This is not just a question of "hacking" the glide 2.x libraries, though you are near the mark when you say that.  I understand what you're trying to say and I have already stated that it's "theoretically" possible.  Trust me I am taking it with a pinch of salt.  Just "opinion", which is the whole beauty of the medium that is the messageboard. :)

Now I'm thinking hacked Voodoo2 drivers on the Voodoo1, any thoughts??? ::)

wrote on 25.04.04 at 20:02:31:

Now I'm thinking hacked Voodoo2 drivers on the Voodoo1, any thoughts??? ::)

Maybe....I think the biggest difference between the V1 and V2 was the RAM and the second TMU, so possibly disabling one TMU from V2 drivers could make them work on the V1. Then the issue of a hardware connection comes to mind, how would that be achived?

Look at the Voodoo2 if you want to figure out SLI.
By following where the lines go we could figure out where V1 lines needs to go.
But There is only one person which that I know of could give us some pics of a v1 SLI. And he already have given one ( but maybe to smal for following lines ).
that man is called: ( maybe he wil be pissed for putting down his name right know I hope not) Procerus. Because that is the only guy with a Dual bord SLI setup.
After figuring out the lines we could see if it is possible.
I can code C (although I am quite a rookie at this level ).
But if we would have a source of the Obsidians  >:( it would go a lot easyier. DAMN why wont Quantum3D give us the SOURCES!!!!!!!!!

Just a quick question here. The Voodoo 2 had dual TMUs, giving it twice the fill rate and essentially twice the power of a voodoo 1. SO...wouldn't hooking up two voodoo 1s in the holy grail "perfect" SLI still give essentially just the power of a Voodoo 2?

Hmm...good point amp! ;)

It would be sorta' like a Voodoo 1.5 actually! :P

Since there are TWO TMU's and TWO FBI's, whereas a Voodoo 2 would only have one FBI... ::)

(and for all the morons out there on the net, this has NOTHING to do with the FBI! FBI = Frame Buffer Interface) ::) :P

wrote on 05.05.04 at 22:03:34:

Hmm...good point amp! ;) It would be sorta' like a Voodoo 1.5 actually! :P Since there are TWO TMU's and TWO FBI's, whereas a Voodoo 2 would only have one FBI... ::) (and for all the morons out there on the net, this has NOTHING to do with the FBI! FBI = Frame Buffer Interface) ::) :P

You must mean 2.5  ;D But also, when you consider the real-world conditions, the actual benefits of SLI are somewhat bogged-down, on both the V2 and the V5. Also, the FBI, if I'm correct, normally doesn't do as much work as the TMUs, so only having a single FBI vs. two wouldn't be a huge performance increase. I think this project is simply not worth the time, unless you have it, but I know I could be doing much better things (especially considering the cost of a couple V2s nowadays, and the better driver support).

He I scanned my voodoo2 board yesterday.
But to really track every line in the board you have to take the chips of it. But I can say that they are heading for the Ramdac, and 2 or 3 ar going to the FBI.

Yah, well it's a NO GO then! :P

Unless someone has a DEAD quantum V1 SLI board and LOTS of spare time with some de-soldering braid and a multimeter... ::)

...PHEW, no thanks! :o

Wel, I can find a dead V2. But mainly we need to now what exactly every pin of a FBI chip and the ramdac chip does. If we know that it won't be that hard.