Real Rampage shots?

[<El Capitano>]

Someone at Hardforum has just posted some pictures which are claimed to have been of the Rampage. Take a look at them (http://www.hardforum.com/showthread.php?threadid=134267) and see what you think.

[<El Capitano>]

D'oh, JR's already posted them, sorry :(

[Shaun_Dark_Lord]

Just incase someone other than me can't get to John's site, here are the pics;

http://homepage.ntlworld.com/gerald.marley/Spectre_pic.jpg

http://homepage.ntlworld.com/gerald.marley/Rampage.jpg

[fallen_StorM]

If they ARE real, what does this mean? Does this mean that somewhere out there 3dfx is still working?

-StorM

[Cop]

"That" will never fit in my box!!!!

[fallen_StorM]

Keep in mind, half of what you see there is a "riser card", the green card is the rampage, the thing under it is the riser card. Its really not that big.

-StorM

[Ailuros]

I think this one was covered already in one of John's threads here already. It's a single chip SPECTRE (no sage) and it's not huge. The ancle of the first picture taken in coordinance with the monitor tricks the eye a bit. It's apparent in the second pic that it's not that big. About the size of a v3.

These are "memorabilia" of the company's formerly known as 3dfx labs, and no there doesn't seem a chance in a million to ever see one of those in the future.

[JF_Aidan_Pryde]

Well said Ailuros! :)

[<Dolemite>]

Those are shots of the original bring-up boards.

They were taken around late November.

[dryass]

i dont get it, what the hell is so special about these shots? it looks like EVERY other video card out there.

[homer_simpson]

haha, i agree dryass

[Seldzar]

Heh well the big deal is this:

1) It's a card most of us wanted but will NEVER get.
2) It's the only picture i've personally ever seen of the card that we all wanted so badly.
3) It gives us a glimpse into how they go about testing and how the area that rampage was being worked on looked like.
4) It serves to remind us of what should have been.
5) It's a piece of a video gaming history; what was and what can never be. =(
6) It's a chance for us common folks to see what so many talented 3dfx individuals spent years developing, only to have it all taken away right at the end.

It's completely irrelivent if it looks like other video cards, it's NOT other video cards, it's Rampage. That should be enough to take a gander and remember.

[ 04-26-2001: Message edited by: Seldzar ]

[fallen_StorM]

Originally posted by Seldzar:
Heh well the big deal is this:

1) It's a card most of us wanted but will NEVER get.
2) It's the only picture i've personally ever seen of the card that we all wanted so badly.
3) It gives us a glimpse into how they go about testing and how the area that rampage was being worked on looked like.
4) It serves to remind us of what should have been.
5) It's a piece of a video gaming history; what was and what can never be. =(
6) It's a chance for us common folks to see what so many talented 3dfx individuals spent years developing, only to have it all taken away right at the end.

It's completely irrelivent if it looks like other video cards, it's NOT other video cards, it's Rampage. That should be enough to take a gander and remember.

[ 04-26-2001: Message edited by: Seldzar ]


Damn good job, couldn't have said it better myself!

-StorM

[dryass]

It's completely irrelivent if it looks like other video cards, it's NOT other video cards, it's Rampage. That should be enough to take a gander and remember.

no, i ment physicaly... the card looks like every other card out there.

these pictures are stupid. its a video card going into a slot. big deal?

and no, it was just a "VIDEO CARD", nothing more, nothing less. it DISPLAY VIDEO on your montior. it doesn't give you magical powers like some of you think it was... LOL.

now... if it was something more like a screenshot, or a frame rates, or if the video card was DOING SOMETHING and not just sitting there in a slot, then yes, maybe it would be more intresting.

but cmon, a card going into a slot. "WOW. that is so cool... lol."

[fiezukem]

Even if it is just a card to some off you it is still a 75+ million dollar card. Now that is a lot of money.

[ATMArage]

Despite how much I miss 3dfx and how pumped I was for Rampage, Dryass is right. It's a card that looks like everyother card. It's not what the card looked like that was so great about it, but how it made stuff look. These pics do nothing in that catagory.

[Ailuros]

well on my part both shots where more than interesting, just because I never had a chance to see how those labs look and how those test they run are being performed.

The 3d gaming industry is becoming more and more dull from day to day, so we need something to keep our interest alive.

Nothing bad about it, yet nothing special either. :D

[mellotron]

ailuros, i was just saying that the other day... When 3dfx was around there were flame wars, controversy, and of course 3dfxgamers.com

Also the Rampage wasn't just another card. The card was going to be 3dfx's first venture into a T&L. Also possibly the first SLI T&L config. and also it was going to be the first video card with a programmable T&L chip that would allow game designers to create their own API's. This was going to be 3dfx's big break, however they just ran out of fuel and sold out. Oh well, better days are to come...

[GrimFaceOfReality]

Originally posted by EvilMessiah:
and also it was going to be the first video card with a programmable T&L chip that would allow game designers to create their own API's.

Huh?

[Ailuros]

Nope SPECTRE was not the card to that would have helped 3dfx to get back their name. Rather FEAR, so they say.

Spectre was just a very good competitor to a GeForce 3, another dx8 compliant card. I don't think it would have been able to kill the GF3 or outsell it, just restablish their name in the market in terms of technology not regaining the speed crown.

[ 04-26-2001: Message edited by: ailuros21 ]

[John Reynolds]

Originally posted by ailuros21:
Nope SPECTRE was not the card to that would have helped 3dfx to get back their name. Rather FEAR, so they say.

Spectre was just a very good competitor to a GeForce 3, another dx8 compliant card. I don't think it would have been able to kill the GF3 or outsell it, just restablish their name in the market in terms of technology not regaining the speed crown.

[ 04-26-2001: Message edited by: ailuros21 ]

I don't know. . .the high-end board would've had almost 13GB of bandwidth compared to a GF3's 7.5GB, and bandwidth is the limiting factor in performance for 3d cards today. I don't think it would've annihilated a GF3, but it should've been generally faster. Would've cost a bit more, though, so. . . .

[mellotron]

GFOR, an API is a set of rules (OpenGL, D3D) that tell the video card to do what its told to. However this is very limiting because of the rules. A programmable GPU would have allowed programmers make faster code, as well as introduce new techniques and methods, this would reduce time to produce games as well as give programers more freedom and ability to make the games look and feel the way they want it to. However Rampage, SPECTRE, whatever never hit the shelves and nVidia turned out to be the first with the programmable GPU with the GF3, the R200 (Radeon2) will also support this rather impressive feature.

[Ailuros]

hey I made that conclusion based on your and Dave's quotes. It's not the first time I hear that one you know......hehehehe.

Yes I do think too that dual chip Spectre would have had a hard time to drop as much as the GF3 can.

But single -chip would have accomplished a very good number of sales. At least there wouldn't have been any acussations about missing features this time and it could have made an excellent OEM deal too.

[dryass]

but john, the question for me wouldv'e been. when WOULD 3dfx release the high-end card?

because the way i look at it, if 3dfx was still here right now, they would still be struggling to get things out on time, and nvidia wouldn't be relaxing so much on the geforce3, and it wouldv'e been out a couple weeks ago, due to pressure and competition.

so i mean sure... the high rampage wouldve been great, but i still questioned the delays, cut down, etc they wouldve put on it.

[Ailuros]

Messiah,

just a friendly warning: you'll have a hard time "lecturing" Grim, he knows his stuff hehehehehe. :D

[mellotron]

he seems to know his stuff. It caught me off guard when he wrote "huh?" as a reply to my earlier post. I was a little concerned so i dumbed it down a little. No offence GFOR

[Ailuros]

dryass,

all we can judge is from the specs. If you take them under account, single chip Spectre was close enough to give the GF3 good enough competition.

Dual chip was just one more chip, nothing difficult to get out the door, and the ram wasn't any high speced/rare ram like 3.5 or 3.8ns AFAIK. The price would have been already high for it, there was no reason to exceed the 500$ range with it.

[Ailuros]

he was buffled I guess by your explanation. I'll leave it to him to straighten it out for you hehehehe :D

He's correcting me quite often too, but it only helps me to learn a few basics more :cool:

[dryass]

ailuros21

thats the thing, things that wern't suppose to be difficult, or seemed easy, ARE dificult for 3dfx. voodoo3 was difficult, so was v4/5.

v4 is the BEST example, the card was like what... 9 months ovedue?

lets look at the specs.
32 ram.
1 chip.

thats easy, yet difficult for 3dfx. thats why i questoined 3dfx tactics with rampage, single chip probaly wouldve been there only release (like way delayed and late) then nvidia wouldve bought them out then.

no diss to the rampage, just to 3dfx

[Ailuros]

I agree but there's a little story about napalm.

Rampage was supposed to be the follow up for the v3 line.

However by the time they had a ready desing for it back then it was too slow comparing to the competition so they canned it.

They wanted at first to release a v3 4000 or Gold, with 32mb ram and 32bit colour support, but that project was dumped too and they finally came to the point to take the Avenger chip and combine it with more ram and a "quick" T-buffer implementation to add at least FSSA, motion blur etc etc as features.

They lived under the impression that they can do it in no time, which was obviously not the case.

If management wouldn't have screwed up, there wouldn't have been a Napalm series in the first place. Rampage was supposed to follow up the v3 and the v4/5 line was just a fill in product. Think of it it makes sense. It's bad management and wrong choices not inability in creating, developing technology or even getting to final execution with it.

If you change your opinion on marketing strategies several times and start to panic cause you where confident that the competitors card was to launch at 125mhz but launched at 166mhz, then you're doomed to screw up. Simple as that.

[Ailuros]

by the way if I remember correctly SAGE taped out before they decided to shut down and sell out. :D

edit: well aidan posted the specs already, so I can't see why I shouldn't repost them...

Spectre: Rampage + Sage
Tapeout: Quarter 4 2000
Projected Shipping Date: March/April

Rampage Rasterizer: 200-250MHZ Core
32MB@200MHz DDR/ Rampage Chip
Manufacturing Process: 0.18µ
4 Pixel Pipelines/1 TMU/Pipeline
Fillrate: 800-1000Mpixels/Mtexels
8 Layer Multitexturing via Loopback
Sage: 125Million Vertices/Second Theoretical
50 Million Vertices/Second Sustained
SLI support for Dual Rampage/Single SAGE configuration

Features:
3dfx M-Buffer
128Tap anisotropic filtering
52bit Internal Color rendering/ 0 – 16.0 Color luminosity range
FXT1/DXT1 Texture compression
Higher Ordered Surfaces (HOS)
3D textures support
True PhotoShop filter effects in hardware
Non-Photorealistic rendering
Cube Environment Maps/EMBM/Dot3 BM
YUV Texture formats
DirectX8 1.1Compliant Pixel Shader
DirectX8 1.0Compliant Vertex Shader

[ 04-26-2001: Message edited by: ailuros21 ]

[GrimFaceOfReality]

Originally posted by EvilMessiah:
GFOR, an API is a set of rules (OpenGL, D3D) that tell the video card to do what its told to. However this is very limiting because of the rules. A programmable GPU would have allowed programmers make faster code, as well as introduce new techniques and methods, this would reduce time to produce games as well as give programers more freedom and ability to make the games look and feel the way they want it to. However Rampage, SPECTRE, whatever never hit the shelves and nVidia turned out to be the first with the programmable GPU with the GF3, the R200 (Radeon2) will also support this rather impressive feature.

Programmable GPU does not allow the developers to write their own APIs. Programmable elements (pixel/vertex shaders) are in fact set by API.

[mellotron]

Doh! *embarassed look*

/me crawls back under rock and to never surface for a while...

[John Reynolds]

Originally posted by dryass:
ailuros21

thats the thing, things that wern't suppose to be difficult, or seemed easy, ARE dificult for 3dfx. voodoo3 was difficult, so was v4/5.

v4 is the BEST example, the card was like what... 9 months ovedue?

lets look at the specs.
32 ram.
1 chip.

thats easy, yet difficult for 3dfx. thats why i questoined 3dfx tactics with rampage, single chip probaly wouldve been there only release (like way delayed and late) then nvidia wouldve bought them out then.

no diss to the rampage, just to 3dfx

You citing the V4's release date as a result of "engineering difficulties" makes no sense whatsoever when you think that 3dfx released the V5 5500 first. What, you think the 5500 was easier to engineer than the V4? No, so it was clearly a result of 3dfx using the VSA-100 yields towards the 5500, boards they thought would sell better.

3dfx released the V2, a 3-chip board that could use an analog connector to work in tandem with another board. They released the V5. They didn't have problems with the V3, they had problems taping out the Banshee chip.

And here's some info for you: from what I was told, the Rampage chip required one respin to be market-ready. That means the single-chip boards would've been out by now, if that info is correct. In contrast, the GF3 chip has gone through, I think, 4-5 respins. While there are tons of pros and cons to the whole single chip vs. scability argument, there is something to be said for smaller chips. 3dfx probably would've beaten Nvidia to market with a DX8 compliant board this time around, unless Nvidia pushed a A4/A5 batch of chips out onto the market to win this PR battle.

But, hell, it's all moot now. I just wanted to point out that your argument had some serious holes in its logic. At least IMO.

[Ailuros]

It's one of the rare circumstances where I hate holes....... :D

[dryass]

theres no escaping it. 3dfx died becuase of what i just said. you can talk all the jibber jabbish about "well.. 3dfx WOUDLVE released rampage.. but..." see, key words there is "woudlve" and " but".

imo, i think rampage would of NOT saved 3dfx, i think 3dfx wouldve done the same thing, they were HEADED to do the same thing, release it late, then loose more money, then let nvidia buy all there cores and junk. imo atleast.

besides, i wasnt saying v4 had enginering troubles, but i was saying the v4 was suppose to be out the same time v5, or maybe even before.. oh wait.. wasn't it suppose to be out even when the v3 was released? talk about one hell of a delay...

with 3dfx past of delays, poor exacution, imo.. there wouldve been no way they coudlve got the rampage out in time to save them, and if they did, they were to deep in a hole too dig out of. so either way, they were screwed from "day 1" from when all this "downfall" started to happen. imo atleast...

But, hell, it's all moot now. I just wanted to point out that your argument had some serious holes in its logic. At least IMO.

exactly, but i think your arguement as well is got holes, and opinions. they always have imo.

[ 04-26-2001: Message edited by: dryass ]

[Ailuros]

well arguing over it won't lead anywhere anyway. The company is gone and it's finished. Neither side can't be sure of "would have's" and "if's".

It really depends on the perspective you take on different matters.

Where I still disagree with you dryass, is that conversations like these are useless. From the technology side of it it ain't. They do it at B3D and any other forum too. They discuss edram eventually and Bitboy's tech too, although it's still doubtful that they will present anything substantial.
It is boring lately in the 3D market and most of it has been covered already. So we just seem to do flashbacks to have something to talk about.

If you have more productive issues you would like to discuss, feel free and post a thread and I'll be happy to join in.

[dryass]

it may be boring to you guys, but i never have ANY time to hang out in message boards, so to me, anything in the 3d world is new.

you had all your fun about talking about this stuff.. but i missed out. besides, i have a tendosy to crash threads, and to throw them off-topic. to much to say, but no ones mentions in other topics what i like to talk about, only once in awhile will someone talk about something that will strike my intrest.

[Ailuros]

I didn't get the impression that this one did strike your interest. It sounded more like a reality call on your behalf.

I presented the history as I believe it to be since 98' when the rampage concept started to appear. A company's fall doesn't happen just in one or six months. It always takes time.

I just wanted to show how and where I believe the chain reaction started. It's enough to stumble a bit in this business and it's a matter of time until you finally fall.

The purpose of these kind of discussions went into the wrong direction though. All I care about is the technology behind it, not why and where the company went belly up.

And the tech behind Spectre is by far not unimpressive, a GF3 - and soon to be more known about- Radeon2 competitor. Some differences and some similarities. These kind of discussions are useless only for those who really show some interest in what a dx8 compliant card can hold.

[John Reynolds]

Originally posted by dryass:
theres no escaping it. 3dfx died becuase of what i just said. you can talk all the jibber jabbish about "well.. 3dfx WOUDLVE released rampage.. but..." see, key words there is "woudlve" and " but".

imo, i think rampage would of NOT saved 3dfx, i think 3dfx wouldve done the same thing, they were HEADED to do the same thing, release it late, then loose more money, then let nvidia buy all there cores and junk. imo atleast.

besides, i wasnt saying v4 had enginering troubles, but i was saying the v4 was suppose to be out the same time v5, or maybe even before.. oh wait.. wasn't it suppose to be out even when the v3 was released? talk about one hell of a delay...

with 3dfx past of delays, poor exacution, imo.. there wouldve been no way they coudlve got the rampage out in time to save them, and if they did, they were to deep in a hole too dig out of. so either way, they were screwed from "day 1" from when all this "downfall" started to happen. imo atleast...

exactly, but i think your arguement as well is got holes, and opinions. they always have imo.

[ 04-26-2001: Message edited by: dryass ]

You know, if you can't get your facts straight then there's no point in even taking up an opposing viewpoint to your supposed arguments.

[Ailuros]

John,

(not too far from the topic) what's your take on quincunx?

Is it actually rotated grid or is it ordered grid + 5-tap blur filter, giving same results as 2x RGMS as in IQ?

Also once there is the difference between the T-buffer and M-buffer just multisampling AA, or was there more to it?

With rumours flying around that NVIDIA might start using some 3dfx tech in the NV30, how high is the possibility to see an M-buffer there?

edit: hope I don't get a yes/no/no kind of answer again....
:D

[ 04-26-2001: Message edited by: ailuros21 ]

[John Reynolds]

Originally posted by ailuros21:
John,

(not too far from the topic) what's your take on quincunx?

Is it actually rotated grid or is it ordered grid + 5-tap blur filter, giving same results as 2x RGMS as in IQ?

Also once there is the difference between the T-buffer and M-buffer just multisampling AA, or was there more to it?

With rumours flying around that NVIDIA might start using some 3dfx tech in the NV30, how high is the possibility to see an M-buffer there?

edit: hope I don't get a yes/no/no kind of answer again....
:D

[ 04-26-2001: Message edited by: ailuros21 ]

Those rumors are completely bogus. Ignore them.

Quincunx is a 5-tap filter. The multisampling that's enabled is done so independently, which is causing a bit of confusion. It's just that the drivers limit the user's choices, I think. And with the GF3, 2x is RGMS and 4x is OGMS (getting 4x to use a rotated grid would've required more gates and Nvidia didn't want to bother with it, or so I've been told). So the 2x + Quincunx is 2x RGMS with the filter applied. What's my take? I would suggest reading the opinions of those who've seen it first hand, like Reverend. He's stated that it definitely blurs but that the blur itself isn't really noticeable during actual gaming, which is interesting. Of course the same could be said for dithering in smoke or banding in lights during actual gameplay, so in the end it'll probably just come down to personal preferences. Some people are more sensitive to certain visual anomalies than others, so. . . .

As for T vs. M buffer stuff, I just looked back over a Rampage presentation file I've had since last year and it doesn't mention anything on the latter, so I'm not sure.

[multigl]

Quincux = Rotated Grid Multiple Sampling Full Scene Anti Aliasing. take a pixel, rotate sub samples around it, blend, wallah you got rotated grid FSAA.

[John Reynolds]

Originally posted by MultiGL:
Quincux = Rotated Grid Multiple Sampling Full Scene Anti Aliasing. take a pixel, rotate sub samples around it, blend, wallah you got rotated grid FSAA.

That's not correct. Quincunx is not "rotating" subsamples. . .it's blending color values from neighboring texels in a pattern from which it derives its funky name.

Oh, and no offense but your signature is a bit erroneous. If Christ was the only begotten son of God, then he did not die of a broken heart. If he was just a typical mortal, though one with a message, then he probably wasn't too happy being nailed up and ending that way (ending because, as a typical man, he wouldn't have been raised from the dead by God). So it depends on your perspective on the man. Personally, I find it somewhat offensive, though I certainly don't want to launch this thread deep into off topic territory.

[ 04-26-2001: Message edited by: John Reynolds ]

[Ailuros]

John,
has Reverend finished his article? I'd love to read it. (ok ok I'll go check) :D

I read somewhere that it does not blur the whole scene, just certain parts of it and that in some occasions there is still some aliasing apparent in closer distances, which seems not to be the case in far distances.
Is this true or was it just mambo-jumbo on the withdrawn FO3D article?

edit:

Sorry guys, but I've pulled my article. I just found out that my Quincunx source was not reliable, or put another way, he was talking complete bollocks! I'm really pissed!
It's not a complete loss though - all the buildup theory was sound, but the Quincunx stuff is slightly flawed. I'm currently drawing from multiple sources to get this thing sorted and hiring a hit-man to take down my source. Hold tight.

Tom P Monkish
No I'm not Tom bloody Pabst!

:D

[ 04-26-2001: Message edited by: ailuros21 ]

[multigl]

well john i stand corrected. as for the take a pixel, rotate sub samples around it, boom FSAA MS style, i was trying to explain RGMS. then again my knowledge on FSAA is rather limited; theres only so much a 16 year old can know.

[Ailuros]

hey multi don't get offended buddy. That was the reason I asked to clarify it for me.

I refuse to treat you as a minor. I prefer to treat you as an adult on equal grounds. Meaning your age is irrelevant. :D

[John Reynolds]

Originally posted by ailuros21:
hey multi don't get offended buddy. That was the reason I asked to clarify it for me.

I refuse to treat you as a minor. I prefer to treat you as an adult on equal grounds. Meaning your age is irrelevant. :D

Pffft. There should be age requirements for the internet, same as drinking or a driver's license. ;)

[Ailuros]

I'm sober.
Go on :D

[John Reynolds]

Well, I'm not sure if what I wrote is correct or not. Read this thread: http://www.beyond3d.com/forum/noncgi/Forum1/HTML/002459.html :confused:

[ 04-27-2001: Message edited by: John Reynolds ]

[Ailuros]

errrrrrrrr it's a dead end......

[John Reynolds]

There was a period at the end of the url. I edited the post and removed it, so the link works now.

This copy 'n paste stuff is just too damn complicated. :rolleyes:

[<Vince>]

It's apparent in the second pic that it's not that big. About the size of a v3.

The initial test boards used the left over Voodoo5 5500 PCB's. Thus, it's the size of a Voodoo5 5500, not a Voodoo3. Althought I'm sure if it made it into production, it could (better have been) shrunk to a smaller, cheaper PCB.

The Dual 'Rampage/SAGE' board fit on a V5 PCB aswell just for a sence of scale.

and also it was going to be the first video card with a programmable T&L chip that would allow game designers to create their own API's

Umm... no. See-below.

GFOR, an API is a set of rules (OpenGL, D3D) that tell the video card to do what its told to. However this is very limiting because of the rules. A programmable GPU would have allowed programmers make faster code, as well as introduce new techniques and methods, this would reduce time to produce games as well as give programers more freedom and ability to make the games look and feel the way they want it to.

This is a Vertex Shader, not an API. A Vertex Shader is a sub-set feature of an API such as OpenGL 1.2 or DX8+. Everything that you said applies to a Vertex Shader. SAGE had DX8 1.0 complient Vertex Shaders.

[Vince]

Doh!!!

Sorry guys, didn't see that there was more than 1 page.. new Forums :-)