Interview with ATI Catalyst SPM

[nooneyouknow]

Just a few comments.

1 - The statement about the most stable drivers actually refers to Microsoft's own documentation. Since MS keeps track of all those errors (via that nice message box about reporting the crash), apparantly, MS has fewer reports from ATI users than anyone else. I don't think that represents all the issues but I would think that the ratio is pretty reflective. As for quantity, I would like to see the quantity for sure since everyone seems to thinkthat nvidia has more units in the market than ATI.

2 - AF uses either Bilinear or Trilinear sampling depending on whether performance or quality setting is chosen AFAIK.

 

[KimB]

Just a few comments.

1 - The statement about the most stable drivers actually refers to Microsoft's own documentation. Since MS keeps track of all those errors (via that nice message box about reporting the crash), apparantly, MS has fewer reports from ATI users than anyone else. I don't think that represents all the issues but I would think that the ratio is pretty reflective. As for quantity, I would like to see the quantity for sure since everyone seems to thinkthat nvidia has more units in the market than ATI.

That is interesting, and I would most certainly like to know the exact numbers. nVidia certainly has far more products in people's PCs today. Combine that with the fact that many crashes are not really related to the video drivers, but other system configuration issues, and it becomes obvious that nVidia could easily develop a quick disadvantage here. In my PC, every occasional program crash that I've had on my GeForce4 occurs in the exact same form on the Radeon 9700 (Of course, I can't give accurate data on frequency of such crashes, as they're still very, very rare and I haven't recorded them, but they're at least similar in the symptoms).

The only real stability problem that I've tracked down to nVidia's drivers recently is with regards to nView. Some programs still don't seem to like it enabled.

2 - AF uses either Bilinear or Trilinear sampling depending on whether performance or quality setting is chosen AFAIK.

Which sort of makes sense for surfaces that are already using bilinear or trilinear filtering, but text in some games is made to use point sampling, and when bilinear/trilinear is forced, it causes the text to become blurry. NWN shows this pretty well, and so does Morrowind if the texture LOD is lowered.

 

[martrox]

I have a bone to pick with the Radeon 9700's anisotropic filtering implementation. It forces bilinear/trilinear filtering on all surfaces. This means that you will get blurry text in some games, depending on your graphics settings.

I haven't run into this problem.... yet. However, I'd gladly accept a little blurry text for excellent 4X/6X FSAA & 16X AF at 1024X768 or 1280X960 with really good speed, which is something I could never do with my GF4TI4600.....

 

[nooneyouknow]

Just a few comments.

2 - AF uses either Bilinear or Trilinear sampling depending on whether performance or quality setting is chosen AFAIK.

Which sort of makes sense for surfaces that are already using bilinear or trilinear filtering, but text in some games is made to use point sampling, and when bilinear/trilinear is forced, it causes the text to become blurry. NWN shows this pretty well, and so does Morrowind if the texture LOD is lowered.

I will do some investigating into this since I am very interested in this now that you brought it up. How does it look on Nvidia hardware since they use Trilinear samples for their AF correct?

 

[Dave Baumann]

I will do some investigating into this since I am very interested in this now that you brought it up. How does it look on Nvidia hardware since they use Trilinear samples for their AF correct?

Although I've not actually seen what Chalnoth states, I think he deos bring up a good point as well.

As well as 'Performance' and 'Quality' modes there should be an option to force Anisotropic, but with the application preference for the filtering type. So, in this mode the slider will still activate the level of Anisotropy, but the application will select Bilinear ('Performance') or Trilinear ('Quality') (or Point, as Chalnoth states) sampling as the developer or user (via the graphics options) requested.

 

[Doomtrooper]

Oh and lets not forget a overclocking tab...I may have mentioned this before

 

[Tahir2]

Oh and lets not forget a overclocking tab...I may have mentioned this before

I have noticed something strange with my Radeon 9700 Pro. It has been in my system for a while now and has two case coolers blowing on top of it normally.

When I first purchased the card these coolers were not present but I noted that the graphics card would get pretty hot on the back where the GPU was as well as on the top corner where there is a small, flat heatsink. That is when I added the case coolers on top... however the extra heat generated wasnt too much to write home about, I was just doing it as I overclock regularly. Now I remved the two coolers from my system a few days ago and touched the back of the Radeon 9700 Pro as well as where the heatsink is on the top corner (on the back) and I was surprised they were a helluva lot hotter. Weird.

Anyways... personally speaking I dont think adding an overclocking tab is wise. Too many new users would potentially destroy their cards and it is a feature that cant be supported for that reason (I bet the return rate would shoot up).

Just my tuppence worth.

PS anyone know why my Radeon 9700 Pro is generating a lot more heat than it did before? Perhaps its a BIOS change but I haven't added any additional hardware since buying the Radeon 9700 Pro. Hrmmm..

edit: typo

 

[Doomtrooper]

I agree overclocking is Risky for a IHV but how many OEMS that make ATI cards ship with overclocking tabs ??

Unitech uses the Rage3D tweak
Gigabyte has one
Tyan..