2007-05-03, 01:24
MartinC:
yea i take input from anybody, just that finas had very nice comparison values . the values standing by themself on a new xbox tells me very little (and i didn't have to to dowload the link you provided). could you see if you can find the same files finas used? most should be on apples's homepage.
finas:
I checked myself and lq pixelshader have no real performance increase from hq with either one buffer or two, so no need to test.
i'm thinking that the fsb difference might have a big impact. we shove data around between different memory locations quite alot, and the lower fsb might slow that down. could explain why the mpeg2 video's showed so minor performance increase as it isn't as complex to decode.
for the mpeg2 part (doesn't work on h264), there is one way that would limit the amount of data moving, wich i know speeds decoding up by a bit, just never got it stable enough to use. it might be even better on the dx1480 due to the slower ram access.
yea i take input from anybody, just that finas had very nice comparison values . the values standing by themself on a new xbox tells me very little (and i didn't have to to dowload the link you provided). could you see if you can find the same files finas used? most should be on apples's homepage.
finas:
I checked myself and lq pixelshader have no real performance increase from hq with either one buffer or two, so no need to test.
i'm thinking that the fsb difference might have a big impact. we shove data around between different memory locations quite alot, and the lower fsb might slow that down. could explain why the mpeg2 video's showed so minor performance increase as it isn't as complex to decode.
for the mpeg2 part (doesn't work on h264), there is one way that would limit the amount of data moving, wich i know speeds decoding up by a bit, just never got it stable enough to use. it might be even better on the dx1480 due to the slower ram access.