...about this two mod?
* Voodoo 5 5500 Agp 2x @ physical Agp 4x
* NForce 2 Ultra Board Agp 8x @ Agp Universal with vmod
http://www.forumzone.it/topic.asp?TOPIC_ID=22932&FORUM_ID=40&CAT_ID=25&Topic_Title=3DFX+V5+5500+Agp+Mod+%26+AN7+AGP2x+Mod+%26+V%2DAGP+Mod%2E&Forum_Title=3dfx+Hardware+e+Software+%2D+Discussioni

In some other thread I was asking for the same thing more or less but Komponent said vddq for agp are also signaling rails. So this dude didn't cut the traces and run a VI/O to VDDQ directly. Personally I wouldn't feel comfortable having 3.3v running through my nforce2 agp bridge. What I didn't get is the possibility to run it without the vI/O to vddq but the only Italian I understand (very few) are due to French.

P.S. I had the AN7 and its a bad ocer (stops ~236mhz bus). At socket A kt333 is available. Would make more sense if it was done an a 754 or that tasty NForce3 939 DFI mobo :)

Of course we have some dubts: we know that there is the possibility that somethings could be burned. Afterall it's a good news and also a signal that tells us that something still moves.

Yeap, its a good thing something is still moving. Always had an naive thought that the voodoo 5 gets juice (vcore/vmem) from the psu socket and not from the agp but I guess other people tried that before.

wrote on 28.06.07 at 11:25:02:

In some other thread I was asking for the same thing more or less but Komponent said vddq for agp are also signaling rails. So this dude didn't cut the traces and run a VI/O to VDDQ directly. Personally I wouldn't feel comfortable having 3.3v running through my nforce2 agp bridge. What I didn't get is the possibility to run it without the vI/O to vddq but the only Italian I understand (very few) are due to French. P.S. I had the AN7 and its a bad ocer (stops ~236mhz bus). At socket A kt333 is available. Would make more sense if it was done an a 754 or that tasty NForce3 939 DFI mobo :)

heh or even nicer a Dual Socket 940 board  ;D
Take Tyan's Thunder K8W for example  8)
http://2cpu.co.kr/review/bkc/s2885.gif

here the board in a more detailed shot:
http://img409.imageshack.us/img409/285/opteronboardbigmx3.th.jpg

fit this with two beefy Opteron 165CPU's and OC them to a nice 3.2Ghz hehe and if you don't like OCing CPU's take two beefy Opteron 285 CPU's they run dual core 2.6Ghz each and have 2x 1MB L2 hereby if the mod will succeed put in your 5500 AGP or 6000 and have all the fun you can with it hehe, this board has NUMA technology thereby each CPU 2 in total with 4 cores have DUal DDR per CPU [not per core ;) ] with that this system has a nice ram bandwidth of 12.8GB per sec with a max amount of 8GB ram per CPU a total of 16GB 8)

I know may cost a lot but for a 3dfx system heh could get quite fast in a lot of games and maybe the FSAA x4 as 8 modes might get vewry playable as well @ 1024 x 768 x16 as 32.

but then again it's the mod that has to work and that's not my thing but in theory it is possible like shown on the nForce2 Ultra board, same type of AGP x8 slot problem.

That is a very good news  ;D
Can somebody translate the Mod stepts ?

In the pictures this dude sews off the agp plastic restricts so that the card can fit the slot and then shoots 3.3v from the back of the board to the proper agp pin also in the back. The most interesting thing is in the beginning where he seams to manage to make the v5 post with 1.65vagp. But is it stable or its just a simple post proving a no-burn state and then to manage to get stability (as it concerns the voodoo....I have some doubts about the sodomization of the nforce) he gets the vi/o on the vddq....strange and unfortunately systran doesn't translate much of the page :(

Hope Komponent kicks in to the rescue :)

Edit: got it.....with 1.65vagp the v5 is instable....the vi/o->vddq is necessary.

Edit 2: happy reading :)


http://www.megaupload.com/?d=4F7RP74S

wrote on 28.06.07 at 11:25:02:

In some other thread I was asking for the same thing more or less but Komponent said vddq for agp are also signaling rails. So this dude didn't cut the traces and run a VI/O to VDDQ directly. Personally I wouldn't feel comfortable having 3.3v running through my nforce2 agp bridge. What I didn't get is the possibility to run it without the vI/O to vddq but the only Italian I understand (very few) are due to French.

wrote on 28.06.07 at 17:09:09:

In the pictures this dude sews off the agp plastic restricts so that the card can fit the slot and then shoots 3.3v from the back of the board to the proper agp pin also in the back. The most interesting thing is in the beginning where he seams to manage to make the v5 post with 1.65vagp. But is it stable or its just a simple post proving a no-burn state and then to manage to get stability (as it concerns the voodoo....I have some doubts about the sodomization of the nforce) he gets the vi/o on the vddq....strange and unfortunately systran doesn't translate much of the page :( Hope Komponent kicks in to the rescue :)

Hi, dudes,
Not that any "rescue" could came from myself over this subject, but if being asked anyway I can share my oppinion as long as it is understood that I am no owner of the absolute knoledge and that what I say is only based on my personal experience and should ever be regarded with caution. But first, sorry dudes, you could have allready noticed that I no longer visit the 3DFX Community's forums as often as I used to; but that is because of my dayly job; I am at work from morning to evening, 6 days of 7 every week, so I do not have free time to continue the work on 3DFX projects anymore, at least for this period, but I am still very happy to browse every evening here at Falconfly's Forums to see the news.
Sorry for the off-topic and back on the subject: The VDDQ lines are not signaling lines, but they supply power to the I/O lines, that are ofcourse bi-directional. A 1.5V-only AGP motherboard signals through 1.5V powered lines to the video card and a 3.3V-only video card signals through 3.3V powered lines to the mainboard. It is simple enough for everyone to understand, realy: the main problem is not the VDDQ, but the bi-directional I/O. The 2 devices must "talk" the same language or problems will arrive, sooner or later...
I see a bigger problem with this mod, though: The dude applies external 3.3V power to the AGP VDDQ pin under the MB's PCB, directly to the slot. But the electrons do not have a mind of their own, so the power applied to the pin will flow through the whole circuit: from the pin to the video card through the AGP slot, but from the pin to the MB's chipset too, through the PCB's trace, so now there is an external 3.3V power applied to it. I am sorry, but I cannot agree with that... Add that to the overvolting caused by the 3.3V VIO to a 1.5V rated item and now you have the complete picture. What I see is a torture to the setup and I do not think that it could support it for ever, but ofcourse I could be dead wrong so do not take my words for granted.
But... there could be a safe way to do a simmilar mod. As you all know, the VSA100 GPU is natively 1.5V signaling compatible along 3.3V too, but the only video card that uses this feature is the Voodoo4 AGP. A safe way to modify a Voodoo5 to work with 1.5V signaling would be to do the modding the Voodoo4 style. I will post bellow a picture with informations on how the standard selection is made for the VSA100 on a Voodoo4 AGP PCB. Ofcourse, I got the info after un-soldering the GPU out of the video card and after research and study and as always you should treat the subject with caution...

http://img404.imageshack.us/img404/4507/voodoo4agp4x15vdx4.jpg

I simply had no spare time anymore to continue the project with a Voodoo5, sorry... so for now, the italian dude that did the present mod got closer than anyone to the final destination and I wish him alot of good luck to continue his work.

some Mobo's allow to rise the AGP Voltage over the 1.65 Volts. ( maximal 1.95Volts ).
Just wonder if this could be enough for the V5 to be stable. Then the mod will consist only to remove the AGP plastic restricts and go!

We defenetlly need an translation for this mod. Can somebody help ?

I also find amazing that this dude did it..if his mod is not the best, this is something else....but it is a step ahead..

I'm curios to hear Hank opinnion about this..

Amigamerlin is translating the text for Forumzone / Engboard you need to wait a little ;)

The 3DMark score is very low.
An 2400MHz CPU, 1.5 Gb RAM and 200MHz FSB should score over 2500 points in this benchmark..he gets only 1800 points.
I think that just only one GPU is working, taking into account that the same score i get with a regular voodoo4.
Did somebody already tried this mod ?

Noone in Forumzone i think... I don't try because my Msi Kt3 Ultra 2 with barton 3200+ and 1 gb dimm pc2700 for me it's enought ;)

http://img166.imageshack.us/img166/9751/v555003dmark2001se640x4ps1.gif

http://img515.imageshack.us/img515/2038/v555003dmark2001se192mhhx1.gif

Max_r, is that score with default 3dmark2001 settings???

They seem really low, I scored 2800 with an XP1800, 512MB PC2700 and the voodoo 5 @ default.

Or did you run it under XP?? that could make a difference

Of course the screens are under Xp ;) but I didn't do it!
The first one is done with V5 @ Default and the second one is done with V5 @ Oc 1:1 192 MHz.

Tons of 3.3v agp cards laying around, spare palomino, spare ddr memory, spare psu....but the damn nforce2 board (chaintech 7njl4) is biosless. Would do the trick and leave the system on for months in order to check it, but the board needs a bios. Any ideas (besides hotflashing....my other boards don't support the chip's voltage :( ) ?

@outlaw: Why specially a board with a nforce2 chipset ?
It could be done i think also for a VIA K8T800 or Nforce3 chipset, because the AGP specifications are the same..... or im wrong ?
Any ideas about this possibility  ?

Btw., without bios, there are not many chances..better you get another board.

And again, the scores is very very low for such a machine.
I have tryed yesterday my XP-Mobile @2500 with a Voodoo4 and the score is very near to the score from this dude ( about 1900 points )
Br.
Radu.

wrote on 20.07.07 at 06:30:41:

@outlaw: Why specially a board with a nforce2 chipset ? It could be done i think also for a VIA K8T800 or Nforce3 chipset, because the AGP specifications are the same..... or im wrong ? Any ideas about this possibility  ? Btw., without bios, there are not many chances..better you get another board. And again, the scores is very very low for such a machine. I have tryed yesterday my XP-Mobile @2500 with a Voodoo4 and the score is very near to the score from this dude ( about 1900 points ) Br. Radu.

Yes, Nforce3, kt800pro or kt880 would be candidates for the project but I mention NForce2 since that is my only spare board I can use that wont bother me if it gets burned to ashes.

s for the bios, I m looking for a machine with an interface (like usb or serial ir something) that would support programming bios chips. Found a few on ebay but they all support the old-school bios chips not the tiny ones used nowdays.

Take a look here, whats happend with an intel board:

http://www.3dfxzone.it/enboard/topic.asp?TOPIC_ID=5095&whichpage=2

Icefire tried this mod...

So, take care !

I have here a K8T800 board, an 3500+ CPU, a Voodoo5 ( my only one ) and also the will to try it. ;D
If something goes wrong, what will be burned ? Motherboard alone, or also the voodoo card.

Regards.
Radu.

It is well known 875/865 had weak agp buses when it comes to modding. I remember people raising the vddq in order to link it with the vtt for memory since mems v.r.s back then didn't follow the rule vtt=1/2vdimm after a certain value so you ended up with 3.3v or more for vdimm but vtt was stuck at values of ~1.45v. Most of whom I knew that linked the vddq with vtt at values above 1.65v ended up with a dead board. On the other hand socket A boards and NForce 2 ones especially where a bit more tollerant in voltage modding (8RDA, NF7-s, 7njs, A7n8x)....especially the weak asus could last endless mods...had a rev. without the known vdimm shut down protect system above 3.15v and back then (with vtt following as it should) I was able to get stable (no drops under load) 3.4v at my winbond bh-5 whilst vddq was at 2.1v and vcore at 2.05v. That is why the amd platform might be a better platform to experiment on.

p.s. btw don't forget that via has the vlink circuit no the agp bus. Don't know how it will react if it receives 3.3v out of nowhere :). Had an asus 939 kt800pro board that stuck oc at ~280Mhz (no 3d above) and when I tried to raise the vddq with a newly found mod in extremesystems....the board died....not on modding....on vddq raising.

Well, the sistem should boot fine with the from bios increased voltage ( 1.75 Volt ).
I will be happy if it will start at the default voltage... i don't want to burnd something..especialy my voodoo card  :o
Afterthat if it's instable at this voltage, i think about rising the voltage tot 3.3V.

If the card must just be inserted in the AGP Slot, then the moding is only to remove agp plastic restricts. So my question is why the Voodoo5 5500 AGP4x prototipes ( which are fitting in the Slot without to be cut ) didn't work in a AGP4x/8X Mobo ?
@ Voodoo5 5500 AGP4x owners: did somebody try to start such a card in a Nforce2 Mobo ?

Well, so far i made another experiment !

In order to force the Voodoo5 card to operate with 1.5 Volts ( AGP 4x ) i connect the AGP A2 pin to ground.
In this way, the Mobo should detect the card as a AGP4x card and supply it with 1.5Volts.
Unfortunately the system did not start ! ( Epox Post error message is 2d ) ... My Board Epox 8K5A2 did not have the possiblity to change the AGP Voltage in order to go over the 1.65 Volts...
What i expect with this ?
I was sure that with this both modification ( A2 grounded and AGP Voltage increasing ) the card will start in an Universal AGP Slot ( AGP1x,2x,4x ) and ...very important ... i can select the AGP4x in BIOS !
If this could be possible, then this will be for me a good evidence that the card can be pluged in a AGP4x/8x Slot without problems !
Fell free to comment my experiment !

P.S. Nor the voodoo card and Mobo were damaged ...

wrote on 22.07.07 at 08:37:08:

Well, so far i made another experiment ! In order to force the Voodoo5 card to operate with 1.5 Volts ( AGP 4x ) i connect the AGP A2 pin to ground. In this way, the Mobo should detect the card as a AGP4x card and supply it with 1.5Volts. Unfortunately the system did not start ! ( Epox Post error message is 2d ) ... My Board Epox 8K5A2 did not have the possiblity to change the AGP Voltage in order to go over the 1.65 Volts... What i expect with this ? I was sure that with this both modification ( A2 grounded and AGP Voltage increasing ) the card will start in an Universal AGP Slot ( AGP1x,2x,4x ) and ...very important ... i can select the AGP4x in BIOS ! If this could be possible, then this will be for me a good evidence that the card can be pluged in a AGP4x/8x Slot without problems ! Fell free to comment my experiment ! P.S. Nor the voodoo card and Mobo were damaged ...

Nice try, Radu, but, do not get upset at me, nice but futile...
First, a statement that there is no doubt about its truth: it does not matter if you are able to select 4x or even 8x mode for AGP in BIOS. A Voodoo5 will allways default to PCI66 mode. It is just the way those cards were built.
Now... the state of the A2 pin is questioned almost only into AGP universal 2.0 slot powered mainboards. With very few exceptions, the 1.5V-only boards assume that only 1.5V/0.8V video cards could be inserted into the keyed slot and after all, have no possibility to provide 3.3V signaling if such a card would ever be fitted into so there is no point to check for a grounded A2 pin for AGP 8x motherboards.
With AGP universal 2.0 mainboards the TYPEDET value is examined at power-up and the video card do negociate with the motherboard's chipset for a common compatible signaling mode. In your special case, in this pre-boot sequence the 8k5a2 MB finds that a 1.5V video card in inserted (due to the state of the modded TYPEDET) and selects 1.5V signaling to use to comunicate to the GPU(s). Allthough VSA100 is capable of this mode, on the Voodoo5 PCB it is hardwired for 3.3V only and the 1.5V signals sent by the MB are not able to cross the treshhold required by the GPU. Therefor the mainboard concludes that it is not able to "talk" to the video card and returns a 2Dh error code, as expected.
Again, friends, do not get confused about the AGP mode and the signaling value apparent link. A video card if built compatible, should be able to opperate correctly in AGP 2x mode with 1.5V signaling with no problem. the trouble is the other way: it should not be possible for a video card to run at AGP 4x or 8X mode with 3.3V signaling (ofcourse, we do not care about that since our Voodoo5 cards will default to PCI66 mode anyway, but I was just stating the facts).

Hi Komponent,

I'm glad that you are back in "3dfx" modding bussines !
Thanks for your feedback, and as usual some remarks :

In my experience i conected only the AGP Pin A2 to ground  - that tells the Mobo to supply the card with 1.5Volts. But the GPU continues to work at 3.3Volts !
I must connect the TYPEDET Pin from both VSA100 also to ground in order to tell the GPU's to use 1.5 Volts signals.
This step was not made because i was not able to find this GPU's Pin's on the card and i tried to avoid any risk The Foto that you posted is for the Voodoo4 ! Can u post pls a picture with the exact location ( better the soldering point ) of this pins at voodoo5 ?

When the AGP Voltage will be increased over 1.5volts ( i.e. 1.8 Volts ), then it will cross over the treshhold required by the GPU - so that the card will start working with 1.5V  8)
Do u agreed with this ?

Thanks for your 3dfx "technical " support !  ;)

Radu.

wrote on 22.07.07 at 12:19:41:

Hi Komponent, I'm glad that you are back in "3dfx" modding bussines ! Thanks for your feedback, and as usual some remarks : In my experience i conected only the AGP Pin A2 to ground  - that tells the Mobo to supply the card with 1.5Volts. But the GPU continues to work at 3.3Volts ! I must connect the TYPEDET Pin from both VSA100 also to ground in order to tell the GPU's to use 1.5 Volts signals. This step was not made because i was not able to find this GPU's Pin's on the card and i tried to avoid any risk The Foto that you posted is for the Voodoo4 ! Can u post pls a picture with the exact location ( better the soldering point ) of this pins at voodoo5 ?

wrote on 28.06.07 at 20:08:41:

I simply had no spare time anymore to continue the project with a Voodoo5, sorry...

... but I will try... maybe soon...

Radu, you know from the e-mails that I sent you in the past what is my oppinion. If you strap the GPU's TYPEDET pin you will have to provide a treshold for the 1.5V signaling.

Te italian dude "ciacara" modified an Asrock 4Core Dual-SAta2 board, in order to take the voodoo5 !
;D

His courage is impressive !

the link remain the same:

http://www.forumzone.it/topic.asp?TOPIC_ID=22932&FORUM_ID=40&CAT_ID=25&Topic_Title=3DFX+V5+5500+Agp+Mod+%26+AN7+AGP2x+Mod+%26+V%2DAGP+Mod%2E&Forum_Title=3dfx+Hardware+e+Software+%2D+Discussioni