Strange, I pulled the ROM to dump it in programmer and it read OK, checksum is OK. I pushed the chip back and it booted. I run VMT and VESATEST and there are no problems with VRAM. Dumped VBIOS attached if someone would need...

1Generated by Martin's VGA13/VESA driver test 1.46
2Host CPU id: 581h, host OS: DOS
3VESA VBE 2.0 Rendition Verite 2200 V1.61 [8192 kB]
4ies Ltd.
5served
6N/A
7OEM software revision 48.49
8VBE/AF version 0.0
961 videomodes available on this card
10
11 #0 = 11Ch - 320x200/8
12 #1 = 10Dh - 320x200/15
13 #2 = 10Eh - 320x200/16
14 #3 = 10Fh - 320x200/32
15 #4 = 11Dh - 320x240/8
16 #5 = 11Eh - 320x240/15
17 #6 = 11Fh - 320x240/16
18 #7 = 120h - 320x240/32
19 #8 = 121h - 384x288/8
20 #9 = 122h - 384x288/15
21#10 = 123h - 384x288/16
22#11 = 124h - 384x288/32
23#12 = 125h - 400x300/8
24#13 = 126h - 400x300/15
25#14 = 127h - 400x300/16
26#15 = 128h - 400x300/32
27#16 = 129h - 512x384/8
28#17 = 12Ah - 512x384/15
29#18 = 12Bh - 512x384/16
30#19 = 12Ch - 512x384/32
31#20 = 100h - 640x400/8
32#21 = 12Dh - 640x400/15
33#22 = 12Eh - 640x400/16
34#23 = 12Fh - 640x400/32
35#24 = 101h - 640x480/8
36#25 = 110h - 640x480/15
37#26 = 111h - 640x480/16
38#27 = 112h - 640x480/32
39#28 = 130h - 720x400/8
40#29 = 131h - 720x400/15
41#30 = 132h - 720x400/16
42#31 = 133h - 720x400/32
43#32 = 134h - 720x480/8
44#33 = 135h - 720x480/15
45#34 = 136h - 720x480/16
46#35 = 137h - 720x480/32
47#36 = 138h - 768x576/8
48#37 = 139h - 768x576/15
49#38 = 13Ah - 768x576/16
50#39 = 13Bh - 768x576/32
51#40 = 103h - 800x600/8
52#41 = 113h - 800x600/15
53#42 = 114h - 800x600/16
54#43 = 115h - 800x600/32
55#44 = 13Ch - 856x480/8
56#45 = 13Dh - 856x480/15
57#46 = 13Eh - 856x480/16
58#47 = 13Fh - 856x480/32
59#48 = 105h - 1024x768/8
60#49 = 116h - 1024x768/15

61#50 = 117h - 1024x768/16
62#51 = 118h - 1024x768/32
63#52 = 140h - 1152x864/8
64#53 = 141h - 1152x864/15
65#54 = 142h - 1152x864/16
66#55 = 144h - 1280x960/8
67#56 = 145h - 1280x960/15
68#57 = 146h - 1280x960/16
69#58 = 107h - 1280x1024/8
70#59 = 119h - 1280x1024/15
71#60 = 11Ah - 1280x1024/16
72
73VideoModeNumber = 4118h
743145728 B allocated for FrameBuffer
75LFB Selector = C7h
76LFB PhysBasePtr = CE000000h
77Measured refresh rate = 60.0 Hz
78Measured 7 FPS (FrameBuffer>VRAM)
79(transfer speed 22 MB/s)

RayeR wrote on 2024-03-08, 04:14:

Strange, I pulled the ROM to dump it in programmer and it read OK, checksum is OK. I pushed the chip back and it booted. I run VMT and VESATEST and there are no problems with VRAM. Dumped VBIOS attached if someone would need...

Nice. Sometimes contacts in these sockets form thin oxide layer on their contact surfaces over the years. It's always good idea to remove all socketed chips and clean their pins and the sockets contacts. If contact cleaner spray doesn't help, at least it doesn't hurt 😀 A good practice is to start repairing thing from the simplest possible problems first.

P.S. As for my dead TNT2 card - now I have suitable spare EEPROM chips, but I can't find any suitable BIOS-dump file for it.

Formula 1 now playable on v2100 and v2200 previously worked only on v1000
Test: Running Formula 1 on a Rendition Verité V2200: Works (fully functional)

RaVeN-05 wrote on 2024-03-13, 11:06:

Formula 1 now playable on v2100 and v2200 previously worked only on v1000
Test: Running Formula 1 on a Rendition Verité V2200: Works (fully functional)

I think that may be not truth.
Formula 1 might have always worked on v2100 and v2200 😉
Just nobody configured it properly on his v2100 and v2200:

sharangad wrote on 2024-03-16, 17:07:

If you could try @agentbsod's fix for f1. He says with the F1 drivers (redline and verite) in the F1 folder and v10003d.uc in windows\system F1 runs without the F1 wrapper. He also says you need verite.vxd in the system folder. I believe the V2000 ref drivers already have verite.vxd as an active driver.

Just give it a go. If it works there's yet another way to play F1 on V2000.

RayeR wrote on 2024-03-08, 04:14:

Strange, I pulled the ROM to dump it in programmer and it read OK, checksum is OK. I pushed the chip back and it booted. I run VMT and VESATEST and there are no problems with VRAM.

Don't know how you tested the card but the procedure should be this:
1.you remove previous card
2. turn on pc without card
3.turn off and plug new card

If you skipped point 2. it may not work.
The reason is bios needs to detect new card and the only way to do it properly is by removing the card first so that bios detects empty slot.

robertmo3 wrote on 2024-03-17, 19:33:

2. turn on pc without card

Hm, interesting, I never do this and new VGA was detected succesfully all the time. Maybe some specific MB has VGA re-detection issues...

robertmo3 wrote on 2024-03-17, 14:25:

I think that may be not truth. Formula 1 might have always worked on v2100 and v2200 ;) Just nobody configured it properly on hi […]
Show full quote

RaVeN-05 wrote on 2024-03-13, 11:06:

Formula 1 now playable on v2100 and v2200 previously worked only on v1000
Test: Running Formula 1 on a Rendition Verité V2200: Works (fully functional)

I think that may be not truth.
Formula 1 might have always worked on v2100 and v2200 😉
Just nobody configured it properly on his v2100 and v2200:

sharangad wrote on 2024-03-16, 17:07:

If you could try @agentbsod's fix for f1. He says with the F1 drivers (redline and verite) in the F1 folder and v10003d.uc in windows\system F1 runs without the F1 wrapper. He also says you need verite.vxd in the system folder. I believe the V2000 ref drivers already have verite.vxd as an active driver.

Just give it a go. If it works there's yet another way to play F1 on V2000.

We don't know whether this works on all V1000s and V2000s. Most of them come with verite.vxd (@AgentoftheBSOD has a proprietary driver where the driver vxd is called something else).

It's entirely the possible the default verite.vxd that ships with V2000 boards (and other V1000s) is incompatible with it. Somebody needs to test both with reference drivers . If the card comes with a verite.vxd it and is incompatible it probably can't be replaced as it's an active driver.

[EDIT] Then again it may work with the reference supplied vxd. Someone needs to test it.[/EDIT]

I did not found my diamond s220 , searching for hours... so I bought a 4mb agp version on ebay (the last one before the crazy price).

I know agp not supposed to add something but I bought an agp because I like to have the videocard in top slot, more air flow... it's the cheap late year 98-99 rendition version... it have so less components, 1024x768 is so blurry...

Test
My system is a 440bx tualatin 1300 at usually 1733, so usually 133 fsb and 89 agp, it seems not liked it so i run it at 100 fsb (66 agp) for now... (cpu speed seems not improve the frame rate at all) . The driver installed is late v5... something and works perfectly except in dos mode that is very slow like 1 fps and when I hit the alt+enter, it go to normal speed... vquake and vérité tomb raider run at a good 25-35 fps.

I did try some old games from 96 to some 99 year that i'm amazinly those games works, not well known to be good to rendition. (Half run very good at 20-30fps, Nfs porshe low detail give 20-30fps, unreal tournament low 20 fps, homeworld smooth but menu have white blocks... )

My card is clocked original 56mhz core and 110mhz memory. Without heatsink and very good airflow, it go very stable at 60mhz core and 120 memory. It crash at 63 mhz vio normal 3.5v. At 60, the gpu is warm like recorded 60 degres but as anthony said, not very hot but my airflow is very very good. In my year 97-99 p166 it was no airflow, so heat sink helped to make it cooler. The heat is similar to a voodoo of the era...

Those cheap card do not have voltage regulator I think it draw power direct from the agp so I can put the vio at 3.7v and the card can overclock more but... 65 mhz, a 2mhz gain. (Yes it's weak haha)

The sgram memory is mosys -8ns I think it's the limiting factor, it crash firm at memory speed limit.
I think the v2100 diamond s220's nec or samsung memory was able to overclock more. In powerstrip, the other memory divider not work (it would be easy to find the bottleneck)

Also 4mb is not great as some games not want do 800x600 and 3dmark99 complaint about the ram on triple buffering

Well, i'm quite happy of the card and the nostalgia that make me feel. We was strong at this time in 97-00 to play games at 15-20fps... hahaha 😅 (those days, I can only play 10 minutes at low fps... it give me headaches fast... especially in Shogo with the mouse lag ) 😅🤣😂

The downside of this particular card is the very blurry output and 4mb is not great for 98-99 games.

Some thermal imagery of my card

We see inside the chip

we can estimate the size of the die 😁

Nice photos, yeah, that cool times when the 1st gen. of 3D accelerators doesn't require even a heatsink, neither a fan, nor took 3 slots nor sucked a kilowatt from aux PSU cables... 😀
I remember playing Unreal 1st time on Voodoo1 and P166@200 also was not much high FPS thriller but the graphics was fantastic compared to all previous games...

Here's a photo of my card to add in a collection here... BTW has VR also some DOS support like 3Dfx (glide.ovl) or just a windows?
I didn't have a time to try it under win yet... 2 more memory chips are on the back side (8MB total).

RayeR wrote on 2024-03-22, 04:18:

Nice photos, yeah, that cool times when the 1st gen. of 3D accelerators doesn't require even a heatsink, neither a fan, nor took […]
Show full quote

Nice photos, yeah, that cool times when the 1st gen. of 3D accelerators doesn't require even a heatsink, neither a fan, nor took 3 slots nor sucked a kilowatt from aux PSU cables... 😀
I remember playing Unreal 1st time on Voodoo1 and P166@200 also was not much high FPS thriller but the graphics was fantastic compared to all previous games...

Here's a photo of my card to add in a collection here... BTW has VR also some DOS support like 3Dfx (glide.ovl) or just a windows?
I didn't have a time to try it under win yet... 2 more memory chips are on the back side (8MB total).

Yes, nice card you have , better than my 4 mb

my card have 4 chip too but it's 128k bank memory, yours is 256k (I think the rendition use 2 chip for texture and 2 other for another things, the sale brochure say that it use 2x64 bit so it may need 4 chip, correct me if I mistaken)

(Also I plans to try to change the chip on my card for faster one and... the 512k chip is very more common and easy to find... tnt2... if someone thinknitnwould work... it will make a 16mb, it's supposed to be supported...)

You should try it, it's easy to work with in windows for me the drivers works great, just dos that need patches for make games work with the card...

It will work good without an heatsink but use it in a well vented case... 😉

The rendition vérité is very interesting era of my teen gaming pc.

Me and all my friends bought a 50$ s220. We put heatsink, update the vbios, good overclock and played 3d games. The 3dfx voodoo 1, we never think of it because it was so expensive.

So for me, the rendition was our first 3d experience in pc gaming for the first 2-3 years 😉

LeCarl wrote on 2024-03-22, 04:54:

Me and all my friends bought a 50$ s220. We put heatsink, update the vbios, good overclock and played 3d games. The 3dfx voodoo 1, we never think of it because it was so expensive.

Excuse my ignorance, but isn't this S200 a Diamond's Rendition Verite V2200? What BIOS did you upgrade as it's already an "upgraded V2100"?

I still have my old QDI Legend Rendition Verite V2200 AGP and I've never bothered to mess with its BIOS. It looks like this one: https://www.vgamuseum.info/images/palcal/rend … p_3d_top_hq.jpg

LeCarl wrote on 2024-03-22, 04:54:

my card have 4 chip too but it's 128k bank memory, yours is 256k (I think the rendition use 2 chip for texture and 2 other for another things, the sale brochure say that it use 2x64 bit so it may need 4 chip, correct me if I mistaken)
(Also I plans to try to change the chip on my card for faster one and... the 512k chip is very more common and easy to find... tnt2... if someone thinknitnwould work... it will make a 16mb, it's supposed to be supported...)

Yes, it would be nice to try. If there are properly wired all necessary address lines it could work. BIOS may detect memory size automatically. Let us know if it will work.

I'm banging my head on the wall trying to get a Diamond S220 (v2100) PCI working well with a 486 DOS build and VQuake

The motherboard is a Zida Tomato 4DPS, version 3.1 (unsure, but the jumpers layout matches the 3.x version and not the earlier 1.x - 2.x) - BIOS is version 1.71 - CPU is a Am5X86 currently running at 133Mhz

The Graphic card is a Diamond Stealth S220 - BIOS version is 1.36, so it should be already the last one.

For the sake of testing, there's no soundcard, I've disabled in the BIOS all on-board devices (COM, LPT, Game Port) which are not essential

I'm currently using MSDOS 6.22 on a Compact Flash directly attached via adapter to IDE1. I'm using the demo version of Quake (1.06) found on Phil's Benchmark Disk and I've added VQuake.exe 1.06b12 and patched it to run on a V2xxx card via V2dosfix. I've tried different version of speedy3d UC files, but I don't see any noticeable difference in quality or speed.

These are the oddities I'm experiencing:

-Bus mastering not working. Both VQUAKE and the Ren Dma Test report that. The mainboard should support it, although I'm well aware that some 486 boards had troubles with PCI and bus mastering. Since this is a fairly well known board, did anyone else experiences problems with PCI and bus mastering?

-I can't use any later version of VQuake (version 1.07 or 1.08). It results in an error on start-up: "SPD3D.UC file not found or not compatible". I've tried several combinations of VQuake and SPD3D.UC to no avail. On the other side, version 1.06 seems to run nicely with different UC packages, although differences seem minimal (0.1 -0.2 fps more or less) .

-If I try to set any resolution higher than 400x300, the game crashes with a "pointer out of range" error.

I know that VQuake DOS was made with V1000 in mind, but I haven't seen any mention that V2xxx boards can cause such problems.

Performance is also quite disappointing. At 320x200 demo1 runs at an average of 17.8 FPS. That's only 3 more frames per seconds compared to what I achieve with software rendering on a ET6000 (14.6 FPS).

Well this kinda sounds like a few issues you're aware of (bus mastering being mostly broken on PCI equipped 486 motherboards, but it being essential to good performance with a verite) and vQuake being very CPU hungry. I think you're just really bottlenecked and while there's certainly some room for improvement, I don't think you'll get about 20 fps with this setup.

DrAnthony wrote on 2024-05-25, 14:44:

Well this kinda sounds like a few issues you're aware of (bus mastering being mostly broken on PCI equipped 486 motherboards, but it being essential to good performance with a verite) and vQuake being very CPU hungry. I think you're just really bottlenecked and while there's certainly some room for improvement, I don't think you'll get about 20 fps with this setup.

I think you're right. I never had a Rendition back in the day; I went straight from a Virge to a Voodoo2 and never looked back. Speaking of Voodoo, I also experimented with this card using the DOS port of Hexen 2, since the author also made a GL-DOS port. With this 3D card the game becomes somewhat playable even on a 5x86. I wanted to check if VQuake would benefit from HW acceleration (of a Verite chip, in this case) in the same way, but it seems VQuake uses the card to do just the bare minimum and all the heavy lifting is still done by the CPU.

Today, I had some spare time and plugged the board into a barebone Pentium 4 PC I assembled on the fly... In this system everything worked fine and I'm able to push the Verite up to 60 fps at 320x200 and 33fps at 640x480. Slower than in software mode, but the image of course looks nicer.

I'm going back to my 5x86 and do more testing at 160Mhz and with some other microcode I found in old archives.

Indeed, also keep in mind that vQuake really didn't run significantly faster on the 2nd gen Verites compared to the original parts. It was a fairly straightforward port of the software renderer that just took advantage of what translated well to the hardware and left everything else to the CPU rather than the more ground up rewrite GLQuake took. Also Quake was hyperoptimized for the pipelined fpu of the Pentium for free operations that were quite expensive on other designs of the era.

Is there any chance of playing GlQuake with that CPU?

Putas wrote on 2024-05-26, 16:04:

Is there any chance of playing GlQuake with that CPU?

Well it depends on what you consider playing. I've seen plenty of people running Voodoos on 486s for the novelty of it all, but even the fastest ones are going to chug a bit. If I can recall correctly, even with a Voodoo II (with that quite useful triangle setup engine) pulled low 20's at basically any resolution in GLQUAKE. A 2nd gen Verite on a 486 is basically an unknown to me and I'd love to see where it lands. My guess would be right around where vQuake is hitting.

DrAnthony wrote on 2024-05-26, 19:22:

A 2nd gen Verite on a 486 is basically an unknown to me and I'd love to see where it lands. My guess would be right around where vQuake is hitting.

Here you go:

The system specs:

Motherboard: Zida 4DPS v3.1
Chipset: SiS 85C496/85C497
BIOS Date: 10/17/96
CPU: AMD Am5x86 (ADZ)
External Cache: 4 x 64KB, 15ns
Main Memory: 2 x 16MB, 72 Pin, 60ns
I/O Card: -
Video card:
1 - Hercules Dynamite 128 - Tseng Labs ET6000 2MB PCI (90MHz MCLK)
2 - Diamond Stealth S220 - Rendition Verite v2100 4 MB PCI (BIOS v1.36 - 55 MHZ Clock)

Quake Demo v1.06 (from Phil's benchmark disk)
Software rendering: QUAKE.EXE 1.06, default settings, Mode 0 (320x200 VGA)
Hardware rendering: VQUAKE.EXE 1.06b12 patched with V2DOSFIX, default settings (NODMA, can't do otherwise on this board), Mode 0 (320x200).

TIMEDEMO DEMO1 @ 133MHz
Software rendering: 14.6 FPS
Hardware rendering: 17.8 FPS

TIMEDEMO DEMO1 @ 160MHz
Software rendering 17.0 FPS
Hardware rendering: 20.4 FPS

Note: DMA should be worth an extra 10% in performance, but it doesn't work on this motherboard. On a 160MHz 5x86, it's borderline playable with a Veritee. Graphics are also nicer, which is an added bonus. Enabling antialiasing results in a -1 FPS hit. Since distant objects textures tend to be somewhat blurry at this low resolution, using D_MIPSCALE 0.5 makes them sharp again, but it has a -2FPS penalty.

NOTE: For the love of God, I can't get Vquake version 1.07 or 1.08 work neither on this PC nor in my makeshift Pentium 4 test bed. The EXE crashes to DOS with the following error:
"Problem Starting Verite. Probable cause: missing or outdated microcode (spd3d.uc). Error code 0x406"

I've tried with different .uc files, patching the exe and other stuff out of desperation. Note that, on the P4 system, I get this error both with this specific card and with another Rendition V2200 board (this time with AGP connector) I have at home. If someone has a verified working copy of VQuake 1.08, would he be so kind to send me the .exe and the matching .uc file? I can't explain what's going on. One theory I have it's that the EXE was compiled with specific Pentium code and won't run on anything that isn't a P5 or P6.