Having now sampled some DX 10 hardware and software (patched CoH and DX10 codesamples) I am beginning to come to the conclusion that the current strategy for MS and GFX makers has done rather well in generating signficant revenue for a technology that is basically a paper launch.
To elobrate it would appear none of the current generation of so call DX 10 hardware can run DX10 with anywhere near required performance (8800 and 2900 excl'd, and even their performnance is far from stellar), this could change with driver/software optimization but to get a 30% aggregated performance improvement across the generation seems unlikely. Rough educated guess on performance improvement requred. And also the next generation will probably have to fit in with the corporations speed bump strategy ie. geforce 9 can not cull to much off the remaining 8 series sales
And with this state of affairs developers are unlikely to invest much resources in to developing DX10 code paths when in reality most people will have to resort to the DX9 for acceptable eye candy/performance ratio. There will probably be noticeably exceptions with games that have cutting edge tech as one of their unique selling points, but thats like 1 or 2 games a year.
People will cite immature code and the likes but even improvements here would seem unlikely to produce signifcant enough results. But just to help shoot down my own argument I will be trying the nvidia beta drivers when i get home from work to see if there is any change, nvidia did recommend these for CoH which i only just found out. Also coders can do some wonderfull things with optimisation, a'la late PS2 games, but its the poor performance DX10 samples thats is worrying me because these are simple enough techniques being demo'd
To elobrate it would appear none of the current generation of so call DX 10 hardware can run DX10 with anywhere near required performance (8800 and 2900 excl'd, and even their performnance is far from stellar), this could change with driver/software optimization but to get a 30% aggregated performance improvement across the generation seems unlikely. Rough educated guess on performance improvement requred. And also the next generation will probably have to fit in with the corporations speed bump strategy ie. geforce 9 can not cull to much off the remaining 8 series sales
And with this state of affairs developers are unlikely to invest much resources in to developing DX10 code paths when in reality most people will have to resort to the DX9 for acceptable eye candy/performance ratio. There will probably be noticeably exceptions with games that have cutting edge tech as one of their unique selling points, but thats like 1 or 2 games a year.
People will cite immature code and the likes but even improvements here would seem unlikely to produce signifcant enough results. But just to help shoot down my own argument I will be trying the nvidia beta drivers when i get home from work to see if there is any change, nvidia did recommend these for CoH which i only just found out. Also coders can do some wonderfull things with optimisation, a'la late PS2 games, but its the poor performance DX10 samples thats is worrying me because these are simple enough techniques being demo'd