Note
Access to this page requires authorization. You can try signing in or changing directories.
Access to this page requires authorization. You can try changing directories.
Component-wise reciprocal.
| rcp[_sat] dest[.mask], [-]src0[_abs][.swizzle] |
|---|
| Item | Description |
|---|---|
dest |
[in] The address of the results dest = 1.0f / src0. |
src0 |
[in] The number to take the reciprocal of. |
Remarks
Use this instruction for reduced precision reciprocal, independent of the strict requirements for divide.
Maximum relative error is 2-21. (The error tolerance just matches rsq)
The following table shows the results obtained when executing the instruction with various classes of numbers.
| src | -inf | -F | -denorm | -0 | +0 | +denorm | +F | +inf | NaN |
| dest | -0 | -F | -inf | -inf | +inf | +inf | +F | +0 | NaN |
This instruction applies to the following shader stages:
| Vertex | Hull | Domain | Geometry | Pixel | Compute |
|---|---|---|---|---|---|
| X | X | X | X | X | X |
Minimum Shader Model
This instruction is supported in the following shader models:
| Shader Model | Supported |
|---|---|
| Shader Model 5 | yes |
| Shader Model 4.1 | no |
| Shader Model 4 | no |
| Shader Model 3 (DirectX HLSL) | no |
| Shader Model 2 (DirectX HLSL) | no |
| Shader Model 1 (DirectX HLSL) | no |