Yes, that looks like the r300g driver.
I would expect llvmpipe to be slower than 300g, since (AFAIK) 300g is already using llvm for the vertex shaders. If llvmpipe turns out to be faster for some reason that would be a very interesting data point, or I guess it might mean that 300g is not using the llvm vertex shader path yet.
Actually I just looked at the XBMC wiki and it says pretty clearly that Shader Model 3 hardware is the minimum for decent performance. That matches what we have seen from user reports (maybe even yours ?) where 5xx hardware (SM3) runs well and 4xx hardware (SM2) does not. The X1250 has an R4xx 3D engine, ie shader model 2.
Did you ever try running with fglrx on the X1250, and if so what was performance like there ?
The FAQ hints at the ability to select basic (ARB) shaders rather than the advanced (GLSL) shaders but those sections of the wiki are blank - do you know if that option actually exists ?
I would expect llvmpipe to be slower than 300g, since (AFAIK) 300g is already using llvm for the vertex shaders. If llvmpipe turns out to be faster for some reason that would be a very interesting data point, or I guess it might mean that 300g is not using the llvm vertex shader path yet.
Actually I just looked at the XBMC wiki and it says pretty clearly that Shader Model 3 hardware is the minimum for decent performance. That matches what we have seen from user reports (maybe even yours ?) where 5xx hardware (SM3) runs well and 4xx hardware (SM2) does not. The X1250 has an R4xx 3D engine, ie shader model 2.
Did you ever try running with fglrx on the X1250, and if so what was performance like there ?
The FAQ hints at the ability to select basic (ARB) shaders rather than the advanced (GLSL) shaders but those sections of the wiki are blank - do you know if that option actually exists ?
Comment