Hello all,

I've been benchmarking away some of the stranger VGA cards that I've got, and there's a few that I figured I'd provide data on, especially since there's so little out there. The two cards that stood out most to me that just "needed" benchmarks were the IIT AGX-014 and NCR77C22E. I hope that you all find this interesting, if not informative.

I used the following cards as a baseline:

Diamond SpeedStar Plus (Tseng ET4000AX 1MB)
Boca Research SuperX VGA (Cirrus Logic GD5426 VLB 2MB)

The test system:

• American Predator Corporation Predator I Plus Motherboard
• Symphony SL82C460 Haydn II Chipset w/ Cache Support
• 66MHz Intel 486DX2-66, C-Step, SL-Enhanced
• 32MB (60ns) Fast Page Mode 72-pin SIMM RAM (2x16MB sticks)
• 4GB Compact Flash Card
• Kingston NE2000-compatible Plug n' Play Ethernet Card
• VLB SMC I/O Controller (integrated on-board)
• Gravis UltraSound MAX Rev. 1.8
• ESS AudioDrive ES1868F
• ESS E-Wave (SerdaCo)

First, there's the Western Digital Paradise Accelerator Pro.

I've got the box and manuals for this beast, but no driver diskettes. This is quite unfortunate, as the IIT AGX-014 is an odd duck. And by odd, I mean odd. It's a 16-bit Windows accelerator, based on the IBM XGA-2. In fact, the name "AGX" is an allusion to the reverse-engineered nature of this thing. It can preform four operands per pixel operation, which was the biggest draw to this architecture. These cards were notorious for cheating in WinBench; specifically, the Hercules Graphite Pro was particularly problematic. Either way, they were regarded as fast cards indeed for certain operations. There's not much out there on these cards, other than that they exist. After scouring the web, the closest thing I could find to hope was on this Vogons thread. I tried and tried anything and everything to find Windows 3.1 drivers for this card, but settled on Windows 95. Windows 95 has built-in drivers that, unfortunately, do not support true color, but do provide more-than-adequate Windows acceleration with this card. I've only got 1MB of 45ns VRAM installed, and wanted to test out claims that Paradise made to the extent that this 16-bit ISA card was "faster than VL-Bus cards." For DOS testing, the card has no VESA BIOS extensions and the TSR is--you guessed it--long since missing. I found the VESA TSR for the Orchid Celsius VLB; it appears to be BIOS-agnostic, and functions properly, allowing me to then load UniVBE and gain full VESA 3.0 BIOS extension compatibility for SVGA usage and testing under DOS.

Then, there's the NCR 77C22E.

This came without a box and manual. I can find one reference to the particular card I have online, and that's in the form of the Boca Research SVGANI. It's a 1MB card, equipped with 70ns DRAM. I had seen this card before, integrated into behemoths like this and this. I was initially skeptical, as this card appears to date to ~1990, and expected poor performance. I could not have been more incorrect. Before spilling the beans entirely, this card is flat-out awful to configure. Let me be clear: awful. Now that that's out of the way, here's the issue: this card displays pretty much everything in Monochrome unless you install a TSR named HIGH_RES.SYS. That TSR is required for Windows 3.1 as well, and this card does have functional Windows 3.1 drivers. It does not appear to have any Windows 95 drivers out there. In DOS, it is as fast as a Diamond SpeedStar Plus equipped with a Tseng Labs ET4000AX. In Windows, it's almost on-par with the AGX-014, and ahead of the Tseng. I do not know if it is a Windows accelerator; such information would be interesting, so if any of you have more on this card, I'd love to hear what you know!

Anyway, I hope you found this interesting. I have been working on a YouTube video on the AGX for a bit now, and wanted to share some of the knowledge here before anything was published. Feel free to check out the video when it drops (hopefully this weekend?); a link to my channel is in my bio. No worries either way, as I'd rather share this and learn some more about these cool cards!