Nice work- first tweak program that enables aniso under 29.60- but... I played a little with it, because I wanted to see, if GF4 Cards still break down with aniso in D3D- so I used 3DM2001se patch- well the newest. I chose only the fillrate test- becasue thats the one that measures the fillrate With no FSAA and Bilinear filtereing, I got 986 MTexel/s in single texturing, 2243 in Multitexturing- fine. So I switched on Aniso- and with 16tap I got 610.7/610.7- OK nV still did not fix it- but now the but: id made no difference if I chose 16tap, 32tap or 64tap- the fillrates were allways the same. So you seem to have found a way to force aniso , but the degree of aniso filterring seems to be allways the same- regardless of 16/32/64 setting. Some points to the FSAA settings: The value 8 settings produce the exact same fillrates like the 2X setting. 6X and value 7 ( what you call 4X 9tap) produce the same fillrates and they are omly slightly below NoFSAA: No FSAA: 986( single Text) and 2244( Multitext) with bilinear filtereing Value 7 and 6X: 980 and 2211 MTexels/s I repeated these measures several times, because I thought it was only an artifact, but Value 7 and 6X is slightly slower then NoFSaa. So my questions: what level of aniso is activated, can you fix it to chose different levels? And Value 8 seems to be 2x FSAA, Value 7 and 6X are only slightly slower then no FSAA. Keep on the good job, hope I could hepl fixing some bugs. Oh I use WinXP-home , nv29.60, the newest nVhardpage on a Athlon XP 1700+, Epox 8k3a+ and a gainward TI4200-128MB XP 650 @ 300 core/540mem.