Also two questions before second bug report..
1.is multi_draw_indirect fully supported in current Nvidia HW (and Fermi) without CPU support in driver or is implemented some part in software I say that because this originated from AMD extension so seems AMD really supports in HW..
2.seems fbo_no_atachments extensions is equivalent to Dx11.1 TIR target independant rasterization if so how it's possible implementable in Nvidia first gen d3d11 HW (fermi) if I read on some sites that TIR was just one thing current D3D 11 HW doesn't supported.. what are the differences then?
Second sample is in 430-compute shader sample seems I get fixed by removing readonly and writeonly qualifiers from storage buffer and then fixing a bug by g-truc
ie changing
byCode :Out.Ouput[gl_GlobalInvocationID].Texcoord = In.Input[gl_GlobalInvocationID].Texcoord;
Code :Out.Ouput[gl_GlobalInvocationID.x].Texcoord = In.Input[gl_GlobalInvocationID.x].Texcoord;
so seems
this should work but doesn't work
Code :layout(binding = BUFFER_INPUT) readonly buffer iBuffer { vertex Input[]; } In; layout(binding = BUFFER_OUTPUT) writeonly buffer oBuffer { vertex Ouput[]; } Out;
i get:
Compute info
------------
Code :0(22) : error C0000: syntax error, unexpected identifier, expecting "::" at token "iBuffer" 0(27) : error C0000: syntax error, unexpected identifier, expecting "::" at token "oBuffer" and fixing it by ex: //layout(binding = BUFFER_INPUT) readonly buffer iBuffer layout(binding = BUFFER_INPUT) buffer iBuffer //works!
At least for now with this fix it works!



