The year we see benchmarks! (maybe!)
To recap- I have a Rampage board (see this page for more info), but no dongle... Lets go from there...
After pictures of my Rampage board turned up on the German 3dfx Voodoo Alert forum I was approached by one of the board members Loeschzwerg if I'd like a dongle constructed. Do cops like doughnuts? oh ya. What is a dongle you ask? Read on...
This is how a typical video adapter outputs, aside from the slight blur caused by my less than stellar camera skills.
This is how Rampage outputs- Due to a bug in the chip the DAC output is reversed (high is low and low is high)
The top picture is actually a Photoshop corrected picture of the image above.
The dongle- Pictured above is the last known Rampage dongle of the perhaps five that were fabricated.
The current owner has been kind enough to provide me with high resolution pictures and other assistance so that this project could get off the ground finally.
Version one of the new dongle from Loeschzwerg
Version two- Some tweaking of the resistance values are required to get the brightness up but at least were in the ballpark!
The next version of the dongle will further improve image quality.
So far the only compiled win9x drivers available were found on Rashly's site which were uploaded some time ago. I'm trying to track down who compiled and uploaded them and if newer versions can be generated. These are alpha drivers that look to be from 3dfx.
A few people outside 3dfx look like they might be onboard for driver development and I'm working with the contacts to see if anyone from 3dfx would like to assist.
Would you like to help? Contact me at this address, all communication will be with the strictest confidence.
Documents & Files-
Power Point presentation from Immersion 2000 event dated 3/9/2000
Power Point presentation 4/12/1999
US Patent 6778181 detailing how Rampage functions.
Rampage Win9x drivers
Rampage .30C BIOS dump.
Looks like Microsoft Visual Basic 6.0 is used to compile the drivers.
Applications tried with Win9x drivers-
1) X-29 demo (fail) OpenGL test.
2) Direct3D Tunnel test from Righteous3D Voodoo Graphics disc. (fail)
3) Half-Life under Direct3D (menus ok, hangs on loading screen.)
4) NFL Blitz under Direct3D (fail)
5) GLQuake, Quake 2 or Quake 3 (fails) with 3dfxogl.dll (tried renaming opengl32.dll) file in game directory or system directory. Quake 2 works in software mode.
6) 3DMark 2001SE, crashes when D3D tests start. 3DMark 2000 refuses to run at all as it thinks there isn't enough memory.
7) OpenGVS image quality test- Runs in Direct3D, but only seems to work well in wire frame mode.
8) 3D Mark 99 MAXX- Locks up like Half-Life
9) Max Payne- This actually runs but is badly banded, almost like a bad V5-6000 board with the screen being like a staircase.
10) Wizmark D3D benchmark- Hangs on beginning of test.
11) Using the built in DirectX diagnostics, DirectDraw tests work perfect but hangs up as soon as the Direct3D test are run.
12) GLRally- Fails to run (OpenGL)
Development notes from 3dfx intranet late 2000-
11/15/00 5:27 PM Stuffed Rampage bringup board being reworked. Rework is fixing power supply connection issues. Estimated time to completion "a while". Probably about 8 hours to rework and verify.
11/16/00 3:20 PM With a lot of work, the board engineers have been able to rework one bringup board. We currently have applied power to the first Rampage chip. Bringup of the chip is underway. We are continuing to work on a clean up of the database for the first bringup board.
11/16/00 8:45 PM The engineers have been working thorough intermittent contact problems in the socket area this evening. Progress has been slow because the chip responds for awhile then stops responding.
11/17/00 5:35 PM Rampage bringup has made little progress today. This morning the engineers were able to read a few registers before the board stopped working. During the day the debug effort continued to point to socket issues. Therefore, we are currently mounting a chip down on the board in an attempt to make more progress. In addition, we are completing three more of the initial bringup boards. We are planning on reworking at least one more board. The respin of the initial bringup board is underway. The current target for completion of the respin is Saturday about noon. We will not attempt to fab more boards until we have high confidence in the database. We will quick turn 3 to 5 boards to make sure the board is correct before building any larger quantity. The second bringup board is continuing to make progress in layout including incorporating the new chip model. This board may be ready for fab by the middle of next week.
11/22/00 Here is what has happened in the last 24 hours. The new version of the initial bringup board bare fab arrive this morning and was sent to the assembly house. We should get new stuffed boards back Wednesday afternoon. The socket fits well on the new board. Our next big push needs to be to get the memory working and stable. We got the DAC's aPwr and GND hooked up, and now we get relatively clean and stable garbage on the video monitor. We've been able to flash and read the ROM reliably, although there may be work left in the 32K-64K address range. We're bringing up our second board with Joe Minacapelli's patented socket holding device, and other rework to ensure stability. PCI clock insertion workaround and PCI clock PLL fixes are underway. We've been able to do a 2d blit, but because the memory system's not stable, it didn't look perfect, but it did look like it did a blit. We tried bringing up the windows driver with Rampage as the secondary device, and it recognized the card, and allowed us to debug some of the modeset code.
11/29/00 Rampage is running but there are some stupid bugs. The DAC output is reversed ... high is low and low is high which turns out to be rather sucky since displaying is difficult. Looking at a dongle to invert the inverted output. Flash ROM bug where part can't be written... Fixes for both are possible in metal layers. 5 boards are operational.
11/30/00 DAC output ‘dongle’ fitted. Rampage running D3D applications. And soon after that the D3D DX8 demos were running and Quake3 debugging in a window (which we’ve also seen pictures of)
3dfx tools provides little additional information I'm afraid.
Overclock tab lists 0mhz clockspeed!
Mad Onion 3D Mark 2001 information.
3DMark 99 information
Everest graphic properties screenshot.
May Payne "staircase" error. Everything looks like it is shifted over.
Videos so far-
Booting into Windows 2000 Pro.
Running Quake II in software mode.
Loeschzwerg doing testing in Germany.
Running OpenGVS image quality test, only wire frame mode is reliable.