VOGONS


Reply 20 of 24, by waterbeesje

User metadata
Rank Member
Rank
Member

Drivers. Plug&pray. Went up and down the hill a few times more than is like, but I'm getting somewhere!

Windows and the mouse didn't seem to like eachother. Just by trial and error I discovered Windows would not respond to the mouse when ctmouse was loaded. Reboot without loading ctmouse and Windows did whatever I wanted!..

Next up, the graphics card.
The Diamond Stealth 64 Vram drivers would crash Windows on startup, locking the entire system. The universal driver library from Vogonsdrivers contains some s3 drivers, resulting in the same lockup. Trying to find any other driver did not result in anything useful... Seems these Vision 968 cards are only did for Windows 95 and 98?
Any way I swapped the card with the original 864. Also no drivers that worked. I did find some 864 drivers library on the net and I could load Windows with them. Yay!
But the picture looked kinda distorted... Not good, could it be bad drivers or bad video Ram? Swapped back to the 968 and Windows did boot! But with the same glitchy distorted picture. I know the 968 is a good one since I tested it on Win98 before without these glitches (on my IBM Aptiva 2137). So the drivers must be broken....

Tired of driver trouble I got to my Matrox MGA2064, downloaded the first MGA driver from Vogonsdrivers and BAM! Worked like a charm! No distortion, full 24 of 32 bit colour on windows 3.11!

Next up: sound. The Crystal CS4237b chip is quite some capable card. It emulates everything from opl3 to Roland modes and I already have configured it for DOS, playing some music from the rest utility.
In fact, the drivers contain a DOS and Windows 3.1 setup program.
Running setup in Windows: not a single hassle. Shut down Windows, editing autoexec.bat and config.sys to get the drivers onto the right boot menu items and reboot the system.
Windows gets me a black screen and after a few seconds returned to the command prompt.
Testing from DOS: working fine.
Check startup parameters all fine.
Reboot to minimal drivers: himem+sound.
Testing from DOS: working fine.
Windows refused to start.
Disable sound driver from DOS, reboot, Windows started with errors (sound related).
Remove Windows sound drivers, set drivers for DOS, reboot, start Windows, reinstall drivers, reboot, Windows refuses.
(No filthy language allowed at our home but I came close).

In the end I found that the config for the soundcard is plug and play but can be overridden by editing a little ini file and give the config.sys sound driver the /O switch (override).
Fiddling with some addresses and stuff kinda made this a soft-non-plug&pray card 😁

Reboot, test in DOS went ok, restart Windows, reinstall drivers set autoexec and config, and rebooted.

Windows boots right onto the desktop.

Me be happy again 😁

Stuck at 10MHz...

Reply 21 of 24, by waterbeesje

User metadata
Rank Member
Rank
Member

Can't wait till I get my new part to complete Bakbeest: there's an Intel DX4 on its way.

It's a WT version, SK051. Not WB unfortunately, but should be a nice addition any way.

Stuck at 10MHz...

Reply 22 of 24, by waterbeesje

User metadata
Rank Member
Rank
Member

And the DX4 is running!

No struggled at all, put it in, changed voltage from 5v too 3,3v and it's all set!

Performance went up by about 25-30% so I'm happy 😀
Cachechk also reported that memory speeds is up.
Int cache 10,2ns, was 15ns
Ext cache 26,2ns, was 30ns (???)
Main mem 34,4ns, was 38,1

So I guess it's pretty much maxed out on hardware now...
- Intel DX4
- 256kB cache, 15ns
- 64MB ram (2x 32MB, EDO)
- Matrox Millennium, 4MB
- Crystal CS4237b soundcard (with onboard IDE)
- 1,2GB Quantum Fireball
- some 2x atapi CDROM drive that seems to fail (drivers won't load)
- 1,44MB FDD
- 1,2MB FDD
- Zip 250 drive

Maybe some 5x86 upgrade would be possible, but that's the only thing I can think of. But that CPU would end up in another system, probably my Aquarius motherboard.

Replacing the sound card could be an option, but no need yet. This one seems to support everything I need so far.

Adding drives (because I can) would require an additional card with IDE or SCSI. It is possible, but no plans yet.

Attachments

Stuck at 10MHz...

Reply 23 of 24, by SodaSuccubus

User metadata
Rank Member
Rank
Member

Are 5x86 133s even ment to be faster then a Intel DX4 WB without being OCd?

I remember when I was playing around with one. With optimized ram timings and a good PCI board it scored like, 34-35 FPS in doom. My DX4 Overdrive was doing about 33. Iv seen WB DX4s report 40.

I for sure thought a 133 would atleast outdo a WB DX4 by a few more frames.

Reply 24 of 24, by waterbeesje

User metadata
Rank Member
Rank
Member

I think a 5x86 should be a little faster at 33x4 versus 33x3 for a dx4. Not a full 33% but at least L1 cache and internal cache run faster. L2 and main memory I don't expect to improve much.

Also Intel is known to be the best "real" dx4, with AMD being just a little slower, and cyrix/IBM/st/to a bit slower again. 5x86 is just a slightly higher multiplier over the slightly inferior AMD. Should be called DX5 or something. (Or dx4 ,and the official dx4 should be named dx3)

The cyrix 5x86 is again another story, sharing much with its big brother 6x86 and more like half a 486, half Pentium, remaining the old socket.

If you are comparing to a Pentium overdrive, well that's actually a Pentium, not a "real" 486. Should be quite a bit faster than a dx4 already when clocked equal 😀

Stuck at 10MHz...