Great findings. Is Dungeon Keeper II a DX6,7 or 8 game?

Did you run into issues with other games?

I noticed that with Incoming, on the latest driver the graphics break, just like if you play it on a modern machine. But use an older 7x driver and it works fine.

Maybe I should start using older graphics drivers for my future projects, because I notice a lot of little odd issues, such as not being able to select some resolutions in Forsaken and other glitches.

philscomputerlab wrote:

Great findings. Is Dungeon Keeper II a DX6,7 or 8 game?

Dungeon Keeper II is a DirectX 6.1 game - this version is on the DKII CD as well. It runs fine with 8.1 provided you use compatible hardware / drivers.

The black screen encountered using HW acceleration on newer graphics cards is a directX + driver issue. On some newer rigs, if you manually copy the missing DX6 files, it will run correctly but will often crash.

philscomputerlab wrote:

Did you run into issues with other games?

I noticed that with Incoming, on the latest driver the graphics break, just like if you play it on a modern machine. But use an older 7x driver and it works fine.

Maybe I should start using older graphics drivers for my future projects, because I notice a lot of little odd issues, such as not being able to select some resolutions in Forsaken and other glitches.

Actually, Half Life will throw the same error as DK II with 61.76 - weird right? I haven't tested incoming yet.

Ah that rings some bells.

I actually got an original copy, comes on 2 CDs.

AFAIK it's another game that has some Environment Mapped Bump Mapping support for Matrox G400 cards. I either couldn't get it working, or didn't know what to look for. AFAIK it was the Lava, but I couldn't see a difference. Maybe a certain patch is needed I don't know.

Not sure about Half-Life. I had it running on a GeForce FX with the latest driver. But the game was patched to the latest version, not sure.

When doing a benchmark marathon on a s370 via 133 board i experienced some weird stuff. It went something like this: tnt1 fine, tnt2m64 no post, tnt2ultra fine, geforce3 fine, geforce4 lockups, geforce4 mx fine, fx5950 lockups etc. etc.

No post even with a voodoo4! (all other voodoo's ran fine).

I don't remember which cards worked exactly but the board drove me nuts (using the same driver for each card up to geforce4). It was an Asus tuv4x. I have some slot1 via boards with the same behaviour.

Swapped it with an Asus tusl2-c and started all over. Benched over 20 cards and had zero troubles.

It is a combination of the agp driver and graphics driver i believe, but i hate these boards for it. they are also a little slower compared to intel based boards so i have no reason to ever use one in a system.

It seems to work better with Ati cards but i never tested myself.

meljor wrote:

When doing a benchmark marathon on a s370 via 133 board i experienced some weird stuff. It went something like this: tnt1 fine, tnt2m64 no post, tnt2ultra fine, geforce3 fine, geforce4 lockups, geforce4 mx fine, fx5950 lockups etc. etc.

No post even with a voodoo4! (all other voodoo's ran fine).

I don't remember which cards worked exactly but the board drove me nuts (using the same driver for each card up to geforce4). It was an Asus tuv4x. I have some slot1 via boards with the same behavior.

I benched everything from a riva128 to a FX5900XT and had no post issues whatsoever. I did experience lockups with some cards, but after installing more appropriate nvidia drivers, they ran fine. Each card works best with a certain driver version - one day I'll make a list. It's not the motherboard, it's the video drivers - this is the whole point of this thread.

My Voodoo 4 4500 won't work in my ECS board either, or in any of my AGP 4x motherboards - but works fine in my Abit AN7-S.

I had similar problems with my Asus CUV4X-M - turn out two 1000uf caps near the AGP port were bloated. Swapped them and it ran fine.

meljor wrote:

It is a combination of the agp driver and graphics driver i believe, but i hate these boards for it. they are also a little slower compared to intel based boards so i have no reason to ever use one in a system.

It seems to work better with Ati cards but i never tested myself.

No, they are not. VIA boards are a bit faster (10-15% even depending on the board) both me and philscomputerlab made threads (phil even made youtube videos) showcasing this. Socket 370 intel boards are in fact slower than their VIA counterparts - this is the main reason I use a VIA board over my two intel boards that are gathering dust in their respective boxes. FPU speed differs as well - I got marginally better results on VIA boards in some Aida64 FPU tests, while intel boards scored a little better in arithmetic tests like CPU Queen.

meljor wrote:

i have no reason to ever use one in a system.

Over the years, I've had a lot of VIA boards and a lot of Intel boards - while I can certainly say that late VIA boards are pretty bad compared to their intel counterparts (overclocking and performance-wise), early VIA chipsets like the MVP3, the Apollo Pro, the KT133, 266, 333 and KT880 were rock solid and really fast. I've had no issues with any of these chipsets, ever. The KT880 for example is my favorite socket A chipset - it's as fast if not a little faster than the nforce2, and is A LOT more stable. Also, unlike the nForce, dual-channel works flawlessly, even with mismatched memory modules, and it will take almost any ram you put in it and run fine (unlike all my nforce boards witch are extremely picky about ram). The KT333 is one of the last chipsets to support 3.3V AGP cards - the fastest chipset you can run a voodoo 3 on (using a barthon CPU).

philscomputerlab wrote:

Ah that rings some bells.

AFAIK it's another game that has some Environment Mapped Bump Mapping support for Matrox G400 cards. I either couldn't get it working, or didn't know what to look for. AFAIK it was the Lava, but I couldn't see a difference. Maybe a certain patch is needed I don't know.

I tried my Matrox G450 with the patched DK II - environment bump mapping worked fine, but the game was rather slow, and kept crashing to desktop every other level.

philscomputerlab wrote:

Not sure about Half-Life. I had it running on a GeForce FX with the latest driver. But the game was patched to the latest version, not sure.

It ran with my FX 5900XT, but kept crashing to desktop. I find the GF4 Ti series are the best card for playing this game w/o crashing and it runs great at 1280x1024. It ran a lot better on the FX, but well...

kanecvr wrote:

No, they are not. VIA boards are a bit faster (10-15% even depending on the board) both me and philscomputerlab made threads (ph […]
Show full quote

meljor wrote:

It is a combination of the agp driver and graphics driver i believe, but i hate these boards for it. they are also a little slower compared to intel based boards so i have no reason to ever use one in a system.

It seems to work better with Ati cards but i never tested myself.

No, they are not. VIA boards are a bit faster (10-15% even depending on the board) both me and philscomputerlab made threads (phil even made youtube videos) showcasing this. Socket 370 intel boards are in fact slower than their VIA counterparts - this is the main reason I use a VIA board over my two intel boards that are gathering dust in their respective boxes. FPU speed differs as well - I got marginally better results on VIA boards in some Aida64 FPU tests, while intel boards scored a little better in arithmetic tests like CPU Queen.

meljor wrote:

i have no reason to ever use one in a system.

Over the years, I've had a lot of VIA boards and a lot of Intel boards - while I can certainly say that late VIA boards are pretty bad compared to their intel counterparts (overclocking and performance-wise), early VIA chipsets like the MVP3, the Apollo Pro, the KT133, 266, 333 and KT880 were rock solid and really fast. I've had no issues with any of these chipsets, ever. The KT880 for example is my favorite socket A chipset - it's as fast if not a little faster than the nforce2, and is A LOT more stable. Also, unlike the nForce, dual-channel works flawlessly, even with mismatched memory modules, and it will take almost any ram you put in it and run fine (unlike all my nforce boards witch are extremely picky about ram). The KT333 is one of the last chipsets to support 3.3V AGP cards - the fastest chipset you can run a voodoo 3 on (using a barthon CPU).

Lets not forget ISA. With Intel chipsets you have to choose between ISA slots or 133MHz FSB without OCing the AGP bus. With VIA you can have it all even on a Tualatin.

I said the board drove me nuts, but also said i believe it is a combination of agp and graphics driver. But that's the awful thing with this chipset: some things work great, other things do not and you have to find the problem again (after switching agp cards for example).

Aopen had a good example of a good mvp3 board but even that one can't top my p5a's. The ali boards are the ones i prefer. A lot of people prefer also the mvp3 so it just depends which board you have i guess.

The Tuv4x may have some bad caps, don't know, nothing looks wrong with them but it could very well be. But i seem to get troubles with most via slot1 and s370 boards i ever used and it all comes down to stability in 3d games.
Never had these problems with intel boards.

When i compare results from a slot1 via and a 440bx, the bx always scores better. When i compared the tuv4x with the tusl2-c the intel was faster in every test in 3d, as well as in everest (i use the older versions, it is now aida).
Maybe other brands do it better with the via chipset but that is the problem, as with intel the performance differences are far smaller, and it just works right.

The plus on the via boards is indeed the isa support and the higher amount of ram that can be used. But i don't need more than 512mb of ram for my 98se machines. for windows xp i use faster machines.

I'm with you about the via socket a boards: they work very well and i didn't have the problems i had with the older versions. Loved my kt133a, kt266a and kt333 boards in the past. Then i went with the nforce2 and never had a problem with that either (a7n8x deluxe) and it was a bit faster. Never used a kt600 or newer so can't comment on that.

DDR ram could cause problems, yes, but there were a lot of inferior brands. Using kingston modules always fixed problems (also with via boards). Dual channel ran perfect with them on the nforce2.

Still have an Asus a7n8x deluxe and a a7v333. Both great boards.

oh, no, my nForce boards won't take quality ram... the Abit won't take Corsair modules - none whatsoever. My A7N8X doesn't like Geil and Kingmax modules. My Epox posts with all types of ram except Kingston.

On early FSB 100 slot 1 boards Intel chipsets generally have an advantage. Testing proves otherwise for late Slot 1 boards (Apollo Pro 133) and Socket 370 boards. The 810 and 815 are nice and stable, but are a touch slower then the Apollo Pro. My Abit VH6 and ECS P6VXA blow my Epox EP-S31A and Abit ST6 out of the water. Overall the ECS P6VXA has the best performance / stability ratio out of all my boards. It scored higher then all other boards (regardless of chipset) even with a slower CPU! (950mhz vs 1GHz). Performance wise, it depends a lot on the CPU FSB:

At 100MHZ: i440bx>Apollo PRO 133>i815
At 133MHz Apollo PRO 133>i815 (since the i440bx doesn't officially support FSB 133, but I have 440 boards that claim support for and work fine at 133)

Also, all my VIA boards support Tualatin CPUs after BIOS updates - I've ran a 1200MHz Celeron at 1530MHz (FSB133*11.5x) in my ECS with no issues whatsoever. On the other hand, none of my Intel boards supports the Tualeron w/o mods (although the ECS should)

P.S. - both my Asus CUV4X-M and CUV4X-E are a little weird. Not the best VIA boards I tested. Nice OC options tough, but not as stable as some of my SOYO boards.

How did you manage to run a 1200 Celeron at 11.5x? I thought these things had a locked multiplier?

I had a VIA Apollo 133, it only worked great with the ATi Radeon 7200, the FX5500 and GeForce 6200 only worked at AGP2x, and if I forced AGP4x, I would see glitches or BSOD.
The motherboard had a single ISA slot, but it wasn't soldered.
Just a note: the FX5500 and GeForce 6200 CD drivers didn't recognize their GPUs!!!! Maybe it was something else breaking it?

meljor wrote:

How did you manage to run a 1200 Celeron at 11.5x? I thought these things had a locked multiplier?

No idea. In BIOS it was set to 12x, but the resulting frequency was 1529 instead of 1600... in any case, as far as I remember, my Tualeron could be down-clocked by multiplier, but it could not be overclocked. On my Asus board for example it would post at 800MHz (8x100) after a CMOS reset. Default is 12x100 for it.

same issue here with an Asus CUV4X-CM and also... Apollo 133A... Pentium 3 1GHZ
I can't get the full potential of my geforce TI 4800SE, AGP 2X is max I can go otherwise the card would glitch out or crash.
6665 3D marks in 3D mark 2001, what is even less than what the Thread starter got with his Ti4200....

thread is a bit old I know.. but maybe a solution has been found to use the card with AGP 4X here?

you can try disabling fast writes or playing around with different 4in1 driver versions, this old thread claims that 4.29 will fix the issue: https://forums.tomshardware.com/threads/asus- … -4x-agp.496240/

wouldn't get my hopes up though, there's about a million threads out there describing issues with this chipset.

I'm currently on 4.43 which should already have this fix. also tried disabling fast write with no result

I also noticed it really depends on the card..
with the TI 4800SE I can't go higher than 2X, and with a Geforce 4 MX460 I can't go higher than 1X what makes the card slower than a Geforce 2 MX 400 at 4X.
its really weird..
I just shelved the machine again and been working with my P3B-F P3 800 and Voodoo 3 until a solution is found that works...