Gx00 driver bug fixes...

[rob_b]

Sounds like a nice registry hack, but did
anyone respond to the question about what
bugs this fixes ? .... rob_b

[Kosh Naranek]

Haig
Why didn't you just say so in the first place instead of just suggesting ppl 2 run their ram @ CAS 2 ?

If you had said the thing about dual CPU and Abit boards a lot more ppl would have answered in that direction.

Sorry I cant't help because I only have a single CPU and an ASUS P3V4X Via Apollo Pro

[Haig]

What part of

what happens if you put "SDRAM Latency equals 2" in your system bios?

don't people understand? This is not a suggestion, just a curious question waiting for an answer.

I thought this was clear enough that I wanted to know what would happen if they used this setting.

In any ase, enough of symantics, I received feedback from 17 people confirming this also works.

Haig

[Gurm]

Exactly! I said:

Turn off P3 optimizations, it'll make your drivers less buggy overall.

You said:

INSTEAD, why not set your memory to CAS2?

(Emphasis mine...)

And we all failed to see the correlation. In fact, I still fail to see the correlation. I also fail to see how fiddling with the memory CAS settings could in any way be an alternative to disabling buggy code in the drivers...?

- Gurm

------------------
Listen up, you primitive screwheads! See this? This is my BOOMSTICK! Etc. etc.

[Heed]

Haig:
I think that your inquiry came accross as an alternative to disabling the 3d optimizations in the drivers. That's why I asked about the drivers relationship with sdram speed and the new instruction sets. Granted, the thread turned into "my cas 2 can beat up your cas 3" for awhile, but your inquiry was in the context of the general (i.e. for all matrox users) nature of the thread's beginning.

Now that it's clear that you had something specific in mind, the inquiry makes more sense (i.e. it's in a specific context). Just some confusion on everyone's part to begin with. Glad you received some positive feedback for what you were looking for.

Just trying to unmuddy the waters a bit

Cheers,
Heed.

[Haig]

Well,

if you insist on bickering about this, then why is my 2nd post so hard to understand?

I'm just curious to know if this will help the dual CPu crashes in win2k that's all.

Now for the he said she said my uncle said part

You said that I said

INSTEAD, why not set your memory to CAS2?

That is NOT what I said. My 1st post clearly stated:

Instead of that, what happens if you put "SDRAM Latency equals 2" in your system bios?

There is a clear difference between what happens and why not

Stating what happens if... clearly tells people "hey, what happens if you do this...Does it work or not?"

Stating "why not do this this" tells people to use an alternative method and live with it which is the way you are reading it.

Now that all this is clear, may I have some feedback please?

Haig

[Gurm]

Ah, ok. You were referring to the dual CPU bit.

Ok. Point one - I don't think MY suggestion (disabling P3 optimizations) has a heck of a lot of bearing on SMP stability.

Point two - on MY machine, it doesn't make a lick of difference in terms of stability. Quake3 crashes a bit faster with CAS2. Of course it's also generating more frames. *shrug*

- Gurm

------------------
Listen up, you primitive screwheads! See this? This is my BOOMSTICK! Etc. etc.

[Haig]

I thank you. In case anyone is interested, here's what we are seeing with the dual cpu problem.

- On our test benches, we only see it on a compaq system with a reliant chipset. The latest certified drivers have fixed it.

- Our game server has been running on a P2B-D and never had this problem.

- Most of the complaints are from Abit users, (which we do not see in house),but some have fixed it with the following workarounds for now:

- disable PIII optimizations

or

- put your SDRAM Latency to 2

or

- do both.

Haig

[PaulS]

Hi Haig,

I have a chaintech 6BDU, 2 X 6000 PIII, PC133 RAM, nothing OC'd. Have always run at CAS 2, and have always had the SMP corruptions in W2k. Recently swapped my G400 into a single processor board, (where it performs fine) so I can't say if this has been fixed with the latest drivers. My new video card (not a Matrox unfortunately) doesn't have the SMP corruptions.

Kind Regards,

Paul

[Haig]

Thanks for the feedback Paul. With this motherboard, we do see the same problems but it isn't as constant as what we saw with the compaq and Abit boards. However, it is there.

Thanks,

Haig

[DanielJ]

Haig,

What about the Tyan Tiger 133 boards (S1834)? I noticed a number of people having problems with them. Including myself. I have tried the PIII hack and the CAS2 trick.

Please post any new ideas on how to fix the problem, or anything you want me to try and give you feedback on.

Thanks,

Daniel

[DGhost]

Haig,

I have had this (the dual cpu hang on boot) problem too, and i have seen several other posts. Speaking from both personal experience and other peoples findings, it would appear the problem is as follows:

The G400 5.1x series drivers for Windows 2000 have an issue with VIA 694X northbridge based dual motherboards. When both CPU's are active and busmastering is enabled in the drivers, it will bring up a BSOD when it trys to initialize the graphics driver. This is independant of the AGP GART driver used (so far), the Windows 2000 version (non-SP1 or SP1), and the only way to get a working display driver is to install the 5.0x drivers (i run 5.06 right now, 5.04 works fine, i have not tried 5.03 but i have heard it works).

I have heard reports that this is a known and reproduced issue, i have heard reports saying otherwise.

Just wanting to clear up some issues ;P

-Luke

[Haig]

With the S1834, we can't reproduce this in house but I have seen a few reports on NG's stating that our card (G400) is wiping out the motherboard bios.

Luke - We tried a few VIA 694X northbridge based boards with our current internal build and couldn't reproduce it.

Haig

[This message has been edited by Haig (edited 28 September 2000).]

[DGhost]

Cool. The board i've been using is the MSI 6321 (694D-A).

I look forward to a new release =)

Oh, and out of curiosity, when will you know if Whistler supports true DualHead features?

-Luke

[This message has been edited by DGhost (edited 28 September 2000).]

[DanielJ]

Haig,

I can run the Tyan Tiger 133 (S1384) with the new drivers (5.14) in W2K with 2 cpu's. Great. Or is it?

The trick seems to be to disable busmastering in PowerDesk. This would indicate a great performance loss. I am not currently bothered with this because I am not doing anything graphic intensive. I just wanted the "Monitor" tab to get 100Hz vertical refresh rate, for now.

Browsing through the registry keys ("User" and "User3D") I noticed the BusMastering option was still set to "1" (true?). Do I have BusMastering or not?

Just wondering,

Daniel