Provides partial precision exponential 2x.
expp dst, src.{x|y|z|w} |
---|
Where:
Vertex shader versions | 1_1 | 2_0 | 2_x | 2_sw | 3_0 | 3_sw |
---|---|---|---|---|---|---|
expp | x | x | x | x | x | x |
The exp - vs instruction operates differently depending on vertex shader versions.
In vs_1_1, the expp instruction gives the following results:
v = the scalar value from the source register with a replicate swizzle dest.x = pow(2, floor(v)) dest.y = v - floor(v) dest.z = pow(2, v) (partial-precision) dest.w = 1
In vs_2_0 and up, the expp instruction gives the following results:
v = the scalar value from the source register with a replicate swizzle dest.x = dest.y = dest.z = dest.y = pow(2, v) (partial-precision)
In vs_2_0 and up, the instruction works like this:
float V = the scalar value from the source register with a replicate swizzle dest.x = dest.y = dest.z = dest.y = pow( 2, V ) (partial-precision)
The instruction provides at least 10 bits of precision.