On Fri, Apr 26, 2002 at 11:09:20AM -0700, Ian Romanick wrote: > Double Buffering --------------- Yes ...but not page flipping and no single buffer visuals. > Gamma Control ------------------ No
gamma works in q3/rtcw here - is this something else? > Page Flipping ------------------ Yes, but I'm not 100% sure. No, I posted a kludgey patch to enable it, should be in the list archives. > Table Fog ---------------------- Unsure No. > TCL Hardware ------------------- Not currently, but soon. Yes (in the tcl-0-0-branch) > Texture Units per pipeline (3) - 2 currently, soon to be 3. Daryll's tree has this. Some work to do to get this patch working in tcl. > Texture > ======= > Cache -------------------------- What exactly do you mean? Textures are loaded into card memory only, at the mo, and swapped LRU. Daryll's tree has "try card , then agp", I've a nearly finished patch for "try card / then agp / swap out of agp when we detect enough card mem" because that seems far better than swapping mru. (though to be honest running with Bilinear filtering and Daryll's slightly less optimal agp texturing is a smoother experience than trying to do these agp->card swaps) At the moment we copy texture from mesa into kernel, into agp buffer then blit in 64kb chunks, that doesn't make much sense for agp, we may as well just copy (with 3dnow prefetch / a.n.other optimisations etc) straight into agp offset? For the card space we might be better avoiding the copy somehow (although the kernel does copy_from_user efficiently if you compile for k7/p3/4 etc) > Vertical Sync ------------------ Yes, but I'm not 100% sure. There's no interrupt handling in the drm / user call back or anything for vertical sync. tcl driver doesn't wait for v sync. (Page flipping patch I sent commented out the vertical sync wait - given the frame rates > monitor refresh with it enabled I'm not even sure it worked anyway) > Volume Textures ---------------- Yes, but I'm not 100% sure. No (might work as a sw fallback?) > W Buffer ----------------------- Unsure. No. > W Fog -------------------------- Yes, but I'm not 100% sure. No. > > Z Plane > ======= > Z-buffer: 16, 24, 32 bits ------ 16 and 24 int (some artifacts with 24bit between sw fallback / hw multipass and PolygonOffset in things like quake) > Driver Optimisations > ==================== > 3DNow! - Yes Some in mesa for sw. > SSE ---- Yes Some in mesa for sw. Some for the codegen stuff (i.e quicker glBegin...glEnd path) -- Michael. _______________________________________________ Dri-devel mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/dri-devel
