New Parhelia Drivers!

[K6-III]

Just got an error:

PowerDesk-HF-Error

"An unexpected error has occured"

Reinstalls of the driver won't get rid of it. Any ideas???

[K6-III]

Nevermind (went away by itself after trying to access it a whole bunch of times and doesn't seem to be coming back)

[K6-III]

What's up with FastWrites being off with these drivers???

[XHotKolaX]

Stability with side-banding on mayhap?

Other than 3DMark and benchmarks outputing misleading numbers, I haven't seen but a 1-2% difference in performance in either direction with Fast Writes on.

[K6-III]

Sure doesn't hurt to have the option, especially since I can turn Fast Writes on and off in my bios...

[JaG]

Originally posted by K6-III
Just got an error:

PowerDesk-HF-Error

"An unexpected error has occured"

Reinstalls of the driver won't get rid of it. Any ideas???

May it be, that you've got .NET v1.1 installed? Then this may happen. As Haig described, you should deinstall .NET v1.1 and the driver then install .NET v1.0 (that one on the Parhelia disc) and install the drivers again.

[JaG]

Originally posted by K6-III
What's up with FastWrites being off with these drivers???

I noticed that phenomenom, too, but after reactivating the FastWrite feature, manually by registry edit nothing too important changed.

If your mainboard handles FastWrites automatically (no certain bios switch) and you want activate the feature simly search for 'FastWrite' in the regedit. It should be six reg_dword entrys. Correct them all to value '1' and the feature will be active. If your mainboard needs to be configured in bios, enable the FastWrites feature there, before changing the registry.

[JaG]

Originally posted by K6-III
Sure doesn't hurt to have the option, especially since I can turn Fast Writes on and off in my bios...

Ok.

Now it's clear. Activate it first there and than modify the registry as described.

[JaG]

Thank you TDB for your detailed quote.

You're completely right. I think we matroxusers all expected Matrox to do so - admit the bug and deal with the consequences, because of their reputation.

But the situation on market in fact became harder and harder in the last years. Now it's not enough any more to have only the best 2D. As well in the professional or med areas do 3D applications become more and more popular. Matrox had to release a product which could realize these demands here and they had to do it quickly before the competitors got those markets, too. Surely I won't imply that the bbz and the management really new s.th. of the bug when they released the Parhelia, but if so, I on the other hand can in parts understand that they had to release it even though.

They or at least too many peaple talk since mid of 2000 from new developments like G800 or G1000. Most of all rumours are based on a small true core. But the development became harder and harder from the beginning of the 'Geforce' releases on. Now are much more transistors to integrate into a chip than before the realization of pixel and vertex shaders. These 'processing parts' take the most die size, much more than the 2D, video or at least RAMDAC or TMDS. They had to function! Nvidia and ATi had already got practical knowledge. They worked on their third gen. for Matrox Parhelia is their first gen. with ps and vs. From release to release of competitors the transistor number increased. Chip developement costs much more money than four years ago. And the product cycle releases became shorter and shorter. It's much more difficult for a firm like Matrox to hold their standards here.

In my eyes it's no mismatch in 2D quality. Parhelia has excellent 2D outputs. Video is brilliant, too. The 3D point is annoying, no matter, but who had bought a new card with the 3D facilities of Parhelia if they released it this year instead of the last one. I think much less people than those who did it anyway, last year.

Last year Parhelia still included some technologies which were completely new for the market: DX9 support at least in parts, two DVI-D and three RGB outputs at super high resolutions, a 256-bit wide mem. bus, a 512-bit wide Controller Array, 10-bits per color channel... A short time later (at R300 release) many of these selling arguments wouldn't have been significant enough any more. Matrox' only selling argument would have been their quality. So they perhaps decided. Such a decision is difficult, because basicly both ways are problematic and wrong.

Now, nine months later, we know, that a complex architecture like Parhelia is hard work for Matrox. Matrox is no firm which wants to grow at any price (like i.e. nvidia). The firm size is easily comprehensible and shouldn't grow so much. They already liked in the past to survive 'depressions' with that strategie. Normally they do things 'completely' not 'fast', that's right. But here we don't have to forget the Software point as well. Still there are some core features, like vs 2.0, which are not 'really' activated. So a later release would have been rather problematic at this point, too.

In fact I think they underestimated the banding problem. A complete rma of all affected cards would be a total catastrophy, I expect. So there is no other way to go here as to dement it, as hard this would be.

[K6-III]

Originally posted by JaG
May it be, that you've got .NET v1.1 installed? Then this may happen. As Haig described, you should deinstall .NET v1.1 and the driver then install .NET v1.0 (that one on the Parhelia disc) and install the drivers again.

Nope, .Net 1.0 SP2

[JaG]

Hmmm.

Than I might see the effect some time, too...

On my Parhelia disc is only the .NET and sp1. Haig didn't mention problems with sp2, but who knows?

[K6-III]

Originally posted by JaG
I noticed that phenomenom, too, but after reactivating the FastWrite feature, manually by registry edit nothing too important changed.

If your mainboard handles FastWrites automatically (no certain bios switch) and you want activate the feature simly search for 'FastWrite' in the regedit. It should be six reg_dword entrys. Correct them all to value '1' and the feature will be active. If your mainboard needs to be configured in bios, enable the FastWrites feature there, before changing the registry.

Thanks.

[JaG]

[K6-III]

I like these drivers. 9275 3dmarks at 238/619 in 3dmark2k1: http://service.futuremark.com/compare?2k1=6364627

[K6-III]

I guess 1002 3DMark03 is OK: http://service.futuremark.com/compare?2k3=635300