VOGONS


Reply 20 of 45, by 386SX

User metadata
Rank l33t
Rank
l33t

Every time I think about the VSA-100 specs I ask myself why it couldn't be what it would have sense to be a theorical Voodoo3 chip when the original chip seems like a modernized boosted Banshee, the VSA-100 chip should have been the "perfect" Voodoo3 but in the 1998 and the latest Voodoo architecture. I never found a meaning about the position of the VSA-100 in the market, how far such product could have last in the 2000 market cosidering the others chips around already having similar features if not more even years before..

Reply 21 of 45, by The Serpent Rider

User metadata
Rank l33t++
Rank
l33t++

3dfx wasted resources on stuff like acquiring STB, instead of big investment in RnD. At that time, STB was one of the big OEM manufacturers of Nvidia cards. It was downward spiral from there.
Original plan was to release VSA-100 cards as high-end and Banshee refresh (Voodoo 3) as low-end offering. That plan failed, VSA-100 was nowhere to be ready in 1999. After that they've wasted even more time on cobbling together SLI card as another stop-gap solution.

I must be some kind of standard: the anonymous gangbanger of the 21st century.

Reply 22 of 45, by TelamonLivesOn

User metadata
Rank Member
Rank
Member

I think that the most logical path is rather than reverse-engineering the VSA-100, I will likely build my own version of the Quantum3D Mercury Brick using 3Dfx Voodoo 2 parts, along with writing special drivers to allow it do more advanced SLI scaling (each card has 2 voodoo 2 chipsets on-board, which will draw even and odd lines respectively, along with the other 3 cards alternating frames). Thank you for everyone's inputs on my crazy idea 😁 I will likely create a different thread about high res scans of people's quantum3d bricks.

Reply 23 of 45, by 386SX

User metadata
Rank l33t
Rank
l33t
The Serpent Rider wrote on 2021-02-25, 18:31:

3dfx wasted resources on stuff like acquiring STB, instead of big investment in RnD. At that time, STB was one of the big OEM manufacturers of Nvidia cards. It was downward spiral from there.
Original plan was to release VSA-100 cards as high-end and Banshee refresh (Voodoo 3) as low-end offering. That plan failed, VSA-100 was nowhere to be ready in 1999. After that they've wasted even more time on cobbling together SLI card as another stop-gap solution.

I understand this view. I always imagined that they probably didn't feel the time was changing faster than ever and there were some hints in the tech market suggesting that earlier.. the Pentium cpu.. the PlayStation game console.. some powerful competitors already having 2D/3D solutions with slow but sometimes better solution. It was not all about the frame rates for seconds and the only single chip solution before the Voodoo3 was the slowed down Banshee? The Voodoo3 might have saved the company I agree if it was the VSA-100, but it might have been even not enough either if another completely new chip wasn't already in design and the "Voodoo3" brand was totally wrong if that road was taken.
Considering how well the Voodoo Graphic and the Voodoo II went, it might have been wise to skip the Voodoo3 brand as a Banshee II card @ 143Mhz and focusing totally on a next gen GPU, considering proprietary APIs wasn't going to have a sense lasting much longer. Instead the expectations that even the "Voodoo3 boxes" design/graphic had, increased the problems even more imho.

Reply 24 of 45, by Putas

User metadata
Rank Oldbie
Rank
Oldbie

I don't see how the VSA-100 could succeed as a highend next to the Voodoo3. It was more expensive while being rarely faster.
Voodoo3 was fast and simple, it was the VSA-100 that would deserve canceling more.

Reply 25 of 45, by The Serpent Rider

User metadata
Rank l33t++
Rank
l33t++

I don't see how the VSA-100 could succeed as a highend next to the Voodoo3

Same way TNT and TNT2 did - large texture support, more memory, 32-bit color, AGP 4x and all new fancy texture compression feature (similar to S3 Savage 3D). With Voodoo 3 3000/3500 probably being released as part of VSA-100 family and Avenger chips marketed as Banshee II or just as Voodoo 3 1000/2000 offerings.

Last edited by The Serpent Rider on 2021-02-27, 12:38. Edited 2 times in total.

I must be some kind of standard: the anonymous gangbanger of the 21st century.

Reply 26 of 45, by 386SX

User metadata
Rank l33t
Rank
l33t
Putas wrote on 2021-02-27, 04:33:

I don't see how the VSA-100 could succeed as a highend next to the Voodoo3. It was more expensive while being rarely faster.
Voodoo3 was fast and simple, it was the VSA-100 that would deserve canceling more.

Just idea but I meant that if the original Voodoo3 was meant to be and renamed "Banshee II" @ a lower freq (like 143Mhz) which would have meant having more chips to sell, and at the same time the "VSA-100" design as the "real" Voodoo3 "maybe" in the 1998 up to the half-1999 at best, it might have survived the new millennium.
Probably I agree imho if the VSA-100 proect was skipped at all and never ever even named it might have gone the same way but at that point it was clear releasing the original Voodoo3 that the company was (at best) taking time.. to do what the VSA-100? Considering the amount of success the V1 and V2 had, I suppose it could have skipped even the Voodoo3 chip as said and launching it without owning a PCB manufacturer as a Banshee II replacement, cheap enough to compete with TNT1 and TNT2 and others solutions but most important meanwhile the design of a complete new processor.
Even "names" were important back then.. the brand "Geforce" was a powerful one.. the brand and tech "T&L" too. The brand Voodoo was already out of its time and to benefit from the prevous successes, the Voodoo "3" in the best scenario had to be a new generation GPU.

Reply 27 of 45, by The Serpent Rider

User metadata
Rank l33t++
Rank
l33t++

Avenger chip just felt right as low-end offering - perfect for OEM and gaming on a budget, where additional 3D features weren't so important. Both ATI and Nvidia did the same trick later on with Radeon 7000/7500 and GeForce 4 MX cards.

I must be some kind of standard: the anonymous gangbanger of the 21st century.

Reply 28 of 45, by 386SX

User metadata
Rank l33t
Rank
l33t
The Serpent Rider wrote on 2021-02-27, 12:23:

I don't see how the VSA-100 could succeed as a highend next to the Voodoo3

Same way TNT and TNT2 did - large texture support, more memory, 32-bit color, AGP 4x and all new fancy texture compression feature (similar to S3 Savage 3D). With Voodoo 3 3000/3500 probably being released as part of VSA-100 family and Avenger chips marketed as Banshee II or just as Voodoo 3 1000/2000 offerings.

Exactly that seems like a good possible idea. The VSA-100 wasn't a next generation completely redesigned GPU that needed years to be designed and produced considering even S3 was doing a good Savage4 chip, slower but with all the features those times required.. Matrox even with the great and never enough used Enviromental Bump Mapping. 32bit colors, full AGP compatiblity and usage, large textures support were already features out there on the market.
Also there's another factor that seems like another big problem, the time factor. I might remember wrong but I think to remember both the Avenger and the VSA-100 weren't released at the time supposed to. And in that specific moment months were like years technologically. Seems like everything went wrong.

Reply 29 of 45, by 386SX

User metadata
Rank l33t
Rank
l33t
The Serpent Rider wrote on 2021-02-27, 12:50:

Avenger chip just felt right as low-end offering - perfect for OEM and gaming on a budget, where additional 3D features weren't so important. Both ATI and Nvidia did the same trick later on with Radeon 7000/7500 and GeForce 4 MX cards.

Right but together with others much higher end architectures. Let's just think at the Geforce2 MX for example.. it was basically the perfect cheap video card to own in its time considering at which price it could have been bought.
I had one, low end PCB build nothing high end and it could go from 175Mhz of core to 220Mhz or similar for the GPU.. technologically it was years ahead everything. At that point I wasn't surprised to see later cards like the Geforce4 or the Radeon 7500 whith the success they had with the higher end chips from both Nvidia and ATi.
The Banshee was basically there for the same reason but at the point going back then I would have bought the Banshee n times more than the Voodoo3 2000 I bought for much higher price. Or even a S3 Savage3D or Savage4 I could have found everywhere at low prices.

Last edited by 386SX on 2021-02-27, 13:05. Edited 1 time in total.

Reply 31 of 45, by Putas

User metadata
Rank Oldbie
Rank
Oldbie
The Serpent Rider wrote on 2021-02-27, 12:23:

I don't see how the VSA-100 could succeed as a highend next to the Voodoo3

Same way TNT and TNT2 did - large texture support, more memory, 32-bit color, AGP 4x and all new fancy texture compression feature (similar to S3 Savage 3D). With Voodoo 3 3000/3500 probably being released as part of VSA-100 family and Avenger chips marketed as Banshee II or just as Voodoo 3 1000/2000 offerings.

That is an awkward analogy. TNT2 was a straight replacement of TNT, just like Voodoo3 to Banshee.

The benefits of marketing something below its tier or delaying it for the next family escape my imagination.
Wasn't Velocity the Banshee II you seek?

386SX wrote on 2021-02-27, 12:54:

The VSA-100 wasn't a next generation completely redesigned GPU that needed years to be designed and produced considering even S3 was doing a good Savage4 chip, slower but with all the features those times required.

But for 3dfx it was quite difficult to make. Devil is in the details.

Reply 32 of 45, by The Serpent Rider

User metadata
Rank l33t++
Rank
l33t++

That is an awkward analogy. TNT2 was a straight replacement of TNT

TNT is mentioned only for feature set, which was better, even if performance wasn't all that groundbreaking, compared to Voodoo 2.

Wasn't Velocity

Velocity is still Avenger chip and the cheapest offering available. Some cards were even branded as Voodoo 3 1000, but with 8 Mb.

I must be some kind of standard: the anonymous gangbanger of the 21st century.

Reply 33 of 45, by 386SX

User metadata
Rank l33t
Rank
l33t
Putas wrote on 2021-02-27, 13:36:
That is an awkward analogy. TNT2 was a straight replacement of TNT, just like Voodoo3 to Banshee. […]
Show full quote
The Serpent Rider wrote on 2021-02-27, 12:23:

I don't see how the VSA-100 could succeed as a highend next to the Voodoo3

Same way TNT and TNT2 did - large texture support, more memory, 32-bit color, AGP 4x and all new fancy texture compression feature (similar to S3 Savage 3D). With Voodoo 3 3000/3500 probably being released as part of VSA-100 family and Avenger chips marketed as Banshee II or just as Voodoo 3 1000/2000 offerings.

That is an awkward analogy. TNT2 was a straight replacement of TNT, just like Voodoo3 to Banshee.

The benefits of marketing something below its tier or delaying it for the next family escape my imagination.
Wasn't Velocity the Banshee II you seek?

386SX wrote on 2021-02-27, 12:54:

The VSA-100 wasn't a next generation completely redesigned GPU that needed years to be designed and produced considering even S3 was doing a good Savage4 chip, slower but with all the features those times required.

But for 3dfx it was quite difficult to make. Devil is in the details.

I suppose it was but I can't imagine if it was cause a lack of next generation ideas or simply a time factor introduced with the PCB company aquisition. Cause without the T-Buffer logics and the renewed SLI tech the VSA-100 was quite passable as a "fixed" -expected- Voodoo "3 ".

Great interviews the link above, even if I'd have liked to hear more video cards related questions more than interesting graphic market ones. It's a great start from the beginning of the company history but then the question went directly to end of the company, not talking a lot about both the Avenger and the VSA-100 chips that would have been really interesting and also about the Voodoo5 6000 life and what happened during its development.
From what I've understood after the 1 hour and 50 minute point, it looks like really the PCB building decision was the bad one most seems to agree. But still I'd have asked about the Voodoo3 and its position on the market at that point.
Anyway that interesting video I didn't see before seems to convince me that after the Voodoo2 all the graphic PC market did run while the aquisition of the PCB manufacturer seems to have introduced a big pause basically loosing the last train at the station.

Reply 34 of 45, by Putas

User metadata
Rank Oldbie
Rank
Oldbie
The Serpent Rider wrote on 2021-02-27, 13:40:

That is an awkward analogy. TNT2 was a straight replacement of TNT

TNT is mentioned only for feature set, which was better, even if performance wasn't all that groundbreaking, compared to Voodoo 2.

Oh I get it now. I have many doubts whether TNT's success was due to the feature set, my bets are on it being such a well-rounded card.

The Serpent Rider wrote on 2021-02-27, 13:40:

Wasn't Velocity

Velocity is still Avenger chip and the cheapest offering available. Some cards were even branded as Voodoo 3 1000, but with 8 Mb.

My point is the Avenger chip covered the 100-250 dollars range with ease, fulfilling the role of Banshee II as well. When more expensive Napalm was ready it would not justify a higher price anyway.

Reply 35 of 45, by The Serpent Rider

User metadata
Rank l33t++
Rank
l33t++

My point is the Avenger chip covered the 100-250 dollars range with ease

The truth is, they didn't. 3500 was an overpriced gimmick card and 3000 was competing with TNT2 Ultra, some of which even had VIVO. Marketing wasn't on their side, because it's simple numbers - 16Mb with 16-bit color card vs 32Mb with 32-bit color card for the same price (or even cheaper). "Do the math" - Atari Jaguar.
Things on the low-end side were even worse, because Nvidia, Matrox, ATI and S3 had plethora of offerings for customer in practically any price range. And 3dfx had Velocity and V3 1000, which came to market too little, too late. In other worlds 3dfx failed the initial goal of acquiring STB - setting strong presence in OEM market.

Last edited by The Serpent Rider on 2022-06-01, 13:48. Edited 2 times in total.

I must be some kind of standard: the anonymous gangbanger of the 21st century.

Reply 37 of 45, by 386SX

User metadata
Rank l33t
Rank
l33t

If I did understand the discussion in the interesting video linked above (not my first language anyway), it looks like the answers quite direct that was given to the PCB manufacturer problem (after 1 hour and 50 minutes) seems like that transition took too much time that basically has been the biggest problem.. I can believe that if the aquisition took that long to be completed the time introduced was imho probably a lot and the competitors were probably ahead in developments.
I never understood which was the position of the 3500 card that should have been the highest performer of that line and instead was a multimedia probably also quite expensive to complete in the package, accessories etc.. "finally" had the wanted 183Mhz line but also I think came out quite late (?) or that was the feelings I remember having when going to the computer eletronic fairs of those years. And while I might be wrong I think to remember the Voodoo5 line took also some time to get released. It was imho a time where one single year passed really fast.. from 1998 to 2000 it was like everything was changing faster than people could even read reviews in the PC battle.