• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.

Rare GPUs / Unreleased GPUs

unknown.png


https://www.ebay.com/itm/Nvidia-GeF...h=item1a6e7b4125:g:LqgAAOSwh7tcL7x4:rk:2:pf:0
 
I've recently unpacked the Ageia Maplewood samples after nearly 3 years in storage and decided to redo all my testing. To my surprise they actually do all function, and after 10 hours of poking DIP switches, reinstalling drivers, and rebooting the PC I have each card set to POST without issue. I have done some minimal testing using FluidMark 1.0.0 but have yet to get around the .dll hell that is stopping any of the PPU accelerated games from running properly in Windows 7.

Some updated photos and preliminary test results pictured below.
P1310472.JPGDSC03605.JPGDSC03599.JPGDSC03602.JPG

1546827688595.png


The constrained PCI-E is pretty clearly bottlenecking performance for these cards. It would have been interesting to see how much further performance scaled on an x8 or x16 interface. The new goal is to record performance in PhysX PPU accelerated games, with a very distant stretch goal being to get in touch with anyone who worked with the Ageia engineering team to discuss what is actually inside this revised silicon, what changes were made, and what the final revision product would have offered if it had ever made it that far.
 
I've recently unpacked the Ageia Maplewood samples and decided to redo all my testing. To my surprise they actually do all function, and after 10 hours of poking DIP switches, reinstalling drivers, and rebooting the PC I have each card set to POST without issue. I have done some minimal testing using FluidMark 1.0.0 but have yet to get around the .dll hell that is stopping any of the PPU accelerated games from running properly in Windows 7.

Some updated photos and preliminary test results pictured below.
View attachment 114132View attachment 114129View attachment 114130View attachment 114131

View attachment 114133

The constrained PCI-E is pretty clearly bottlenecking performance for these cards. It would have been interesting to see how much further performance scaled on an x8 or x16 interface. The new goal is to record performance in PhysX PPU accelerated games, with a very distant stretch goal being to get in touch with anyone who worked with the Ageia engineering team to discuss what is actually inside this revised silicon, what changes were made, and what the final revision product would have offered if it had ever made it that far.
Awesome post!

What a coincidence, i was literally just moments ago researching PhysX Engineering Samples... only came up with this one lowly picture:

1546828148116.png
 
Last edited:
Holy crap is that a rare card! Oh, if I were a collector, I'd get and mount it on my wall!

Appears to be only an 8132, only 32MB of ram and not 64MB ...

it's got some Aftermarket heatsinks ...

Missing the proprietary Power Board

Seems to have some damage with missing SMD capacitors at the very least...






 
The riva v3400 TNT was sold as an upgrade to the TNT Riva AGP sold in Micron computer system. Mine did not have any branding other than Nvida Riva and could have been Asus or it could have been micron that made it. It got replaced as soon as I was back in the United States with a Diamond Multimedia card.
 
Hello :)

i'm from Germany and follow this thread since a few years as silent reader and i just created an account to contribute some pictures from my ES/rare cards collection.

This was my very first rare card in my collection: ATI R300 card (If i remember correcty, a Radeon 9700 Pro)

R300_1.JPG R300_2.jpg R300_3.JPG
R300_5.JPG R300_6.JPG
 
Here is another one of my favorite cards from my collection: NV28 test board. It has four dip switches to change the device id.
With all switches on, the driver detects the card as a GeForce 4 Ti 4800, all switches off, the card is detected as Quadro 4 980 XGL.

There are possible more combinations, but i diddnt't want to break the card.

NV28_1.JPG NV28_2.JPG NV28_3.JPG
NV28_4.JPG
 
Here is another one of my favorite cards from my collection: NV28 test board. It has four dip switches to change the device id.
With all switches on, the driver detects the card as a GeForce 4 Ti 4800, all switches off, the card is detected as Quadro 4 980 XGL.

There are possible more combinations, but i diddnt't want to break the card.

View attachment 115632 View attachment 115633 View attachment 115634
View attachment 115635
That looks like a development board.
 
Back
Top