PIII & V6k

[osckhar]

Hi to everybody,  

Well, I would like  know what Board for PIII with ChipSet VIA is totally compatible with the V6k?  

Regards,
Oscar

[omega_supreme]

I asked Hank the same thing (only one tiny difference) and he says basically the SIS and VIA chipsets should work. But just to be sure ask Rolo01. If im not mistaking he ran his A2.2600 next to a P3 1100.
I think is was an ASUS mainboard.

[osckhar]

I asked Hank the same thing (only one tiny difference) and he says basically the SIS and VIA chipsets should work

Well, I have tried:
MSI MAX REV.2 645DX: Lock Ups!
DFI SIS 645DX: Lock Ups!
GX Board Dual PIII 450MHz: It doesn't want boot up!
My BABY is very, CAPRICIOUS. Rather, it is very sensitive to the AGP BUS.

Tomorrow will come in a ABIT SA7, I will try it with my v6k. I also will try it with a ASUS P3BF.

If im not mistaking he ran his A2.2600 next to a P3 1100.
I think is was an ASUS mainboard.

Yes, Roland has one ASUS P3BF.

Regards,
Oscar

[omega_supreme]

Yes that was the board I meant. Why dont you try that one?
If I accepts P3 1100 Mhz youcan put in a pretty decent CPU

[Rolo01]

I told Osckhar he can even use a Celeron-1400 ( Tualatin core ) with an adapter on the P3B-F. I tried many boards before I found some that work well with my 6000. The results were posted on the 6000 resource page.

[hanksemenec]

Hi,

Funny thing you, have been asking about PIII. I have been trying to chase down a problem with KT266 PRO. Nothing works under 2K or XP. But when I went to Win98 everything runs stable under 3DHQ B10 driver.
Same hardware, just different OS and driver.

Q3ADemo locks in about 3 seconds, in the same place under timedemo demo001 with all XP and 2k drivers. This happens even under single chip settings.

With Win98 the demo goes all the way through, and Q3A runs over night.

I can see the command fifo is executing an AGP fetch instruction which should not exists and write pointers are right behimd the read pointers or on top. I have seen this ondition under Power PC.

Please can you verify if Win98 is stable with 3DHQ B10 driver on your rig. This is a good data point for me, since I have only KT266 MB.

Hank

[Obi-Wan Kenobi]

Well What I do know is that the VIA Apollo Pro 133A is a very stable chipset in many conditions. The ABiT VP6 is my P3 board of choice, it supports Dual Pentium!!!/EB also, I have it with 2x 1Ghz P!!!/EB 1Ghz and 4x 256MB PC-133 Mircon Crucial, I suppose you could try that one, I dunno what you think about it Hank, the chipset is about the same age as the Voodoo5 6000, so I was think'n out loud hehe.

Friendly Regards

Obi-Wan Kenobi.

[Rolo01]

Hank,
does it have to be a Pentium 3 ?
If yes, I can run some tests in a week from now, at the moment I only have 2 Pentium-4 PCs at home ( with Win98 ), the Pentium 3 comes back in approx. 1 week.

[osckhar]

Hi,

Well, yesterday came in a ABIT SA7. I tried it with the v6k and I got the same mistakes like the others SIS 645DX BOARDS. Anyway, this board is more estable than the others 645DX Boards.

Please can you verify if Win98 is stable with 3DHQ B10 driver on your rig. This is a good data point for me, since I have only KT266 MB.

I apologize for the dealy, Hank. Well, today came in a MSI KT3 ULTRA REV2. This board is IMPRESSIVE, it is totally stable with the v6k. I have tried the 3DHQ B10 driver under WINXP, and it is very unstable, the 3DMark2001 reboot the PC!  
The same thing utilizing 645DX Board and 3DHQ B10 and WinXP!

Regards,
Oscar

[hanksemenec]

Looks like I will have to wade through a bunch of stuff.

KT266 PRO 2x 1GHZ PIII

win98SE - 3DHQ Beta 10 - card runs for ever
wnXP - any driver - Q3ADEMO (timedemo 1, demo demo001) lock ~ 3Sec
win2k - any driver - Q3ADEMO lock ~ 3 seconds

Srverworks chipset 2x 1GHz PIII

Same result as above
V5-5500 runs ok with same drivers set

(Hint bridge might be doing something stupid, since all chips lock in the same way on 6k, checked the read pointers and command fifo registers on each VSA100 chip)

K7S5A rev 5.0 AMD XP2400+

no problems under 2k. I have to check out latest drivers.

Well it looks like a lot of work, I need YOU to check 3DHQ B10 under win98SE, the Glide source is different from win2k/xp.

Hank

[osckhar]

(Hint bridge might be doing something stupid, since all chips lock in the same way on 6k, checked the read pointers and command fifo registers on each VSA100 chip)

Hank, any solution?

Well it looks like a lot of work, I need YOU to check 3DHQ B10 under win98SE, the Glide source is different from win2k/xp.

Well, with the MSI KT3 ULTRA REV2 and QuakeIII Arena timedemo 001 (4 pass):
Win98SE and 3DHQ B10: V6k Works very well!

WinXP and 3DHQ B10:
No FSAA: Seems to be the v6k works fine. Not LOCKS UP!
2XFSAA: Problems, the v6k has LOCKS UP to the second pass.

Hank, any test more? This driver seems to be very unstable!

Regards,
Oscar

[hanksemenec]

Well a little more on the bug.

everytime I run the Q3ADEMO, FIFO hole counter reports that 4 DWORDS are missing to close the hole.

Magically Hint bridge has 4DWORD Delayed Transaction FIFO. It is possible that the system discards the writes, and locks the system.

I'll try to get rev CC Hint Bridge and see if an undocumented feature slipped into rev B Hint chips . The only way to get arround this is to implement Hole Counter interrupt from 6k and recopy the fifo contents to avoid a lock.

That means that the driver would have to keep a local copy of fifo contents, not a long one, and do a move to VSA100 fifo. With PIII it mght be a small performance penalty.

Hank

[ggab]

Originally posted by hanksemenec

Well a little more on the bug.

everytime I run the Q3ADEMO, FIFO hole counter reports that 4 DWORDS are missing to close the hole.

Magically Hint bridge has 4DWORD Delayed Transaction FIFO. It is possible that the system discards the writes, and locks the system.

I'll try to get rev CC Hint Bridge and see if an undocumented feature slipped into rev B Hint chips . The only way to get arround this is to implement Hole Counter interrupt from 6k and recopy the fifo contents to avoid a lock.

That means that the driver would have to keep a local copy of fifo contents, not a long one, and do a move to VSA100 fifo. With PIII it mght be a small performance penalty.

Hank

[?]these (great) comments/suggestions, can be implemented in currents drivers?

[osckhar]

More Tests:

ASUS P3BF v.1.03: V6k doesn't want boot!

Regards,
Oscar

[Rolo01]

Hola Osckhar,
which BIOS did you use on the P3B-F ?
The latest version ( 1.08 ) does not work at all with the V5-6000.
I recommend you look for Bios 1.06, this one enables the high multi that is needed for the Tualatin-Celerons, and works flawless with the 6k. I think I found it on www.asus.com.tw or something.