Home My Page Projects Code Snippets Project Openings diderot

# SCM Repository

[diderot] View of /branches/pure-cfg/src/compiler/IL/gen/low-il.spec
 [diderot] / branches / pure-cfg / src / compiler / IL / gen / low-il.spec

# View of /branches/pure-cfg/src/compiler/IL/gen/low-il.spec

Tue Feb 8 16:12:38 2011 UTC (9 years, 6 months ago) by jhr
File size: 2345 byte(s)
```  Working no mid to low conversion
```
```# specification of operators for LowIL version of the IR.  Each line (other than comments)
# specifies an operator using four fields, which are separated by ":".  The fields are
#	name
#	argument type		(optional)
#	arity
#	comment			(optional)
#
# type-indexed arithmetic operations
Add : ty : 2 :
Sub : ty : 2 :
Mul : ty : 2 :
Div : ty : 2 :
Neg : ty : 1 :
LT : ty : 2 :
LTE : ty : 2 :
EQ : ty : 2 :
NEQ : ty : 2 :
GT : ty : 2 :
GTE : ty : 2 :
Not : : 1 : boolean negation
Max : : 2 :
Min : : 2 :
Sin : : 1 :
Cos : : 1 :
Pow : : 2 :
#
### vector operations
# Dot<n>(u, v)	-- computes dot product of u and v; n specifies u and v's arity
Dot : int : 2 :
# Cross(u, v)	-- computes cross product of u and v
Cross : : 2 :
# Select<ty,i>(u)  -- select ith element of vector u; ty specifies u's type (IVec or Vec)
Select : ty * int : 1 :
# Norm<n>(u)   -- computes length of vector u; n specifies u's arity
Norm : int : 1 :
# Scale<n>(s,u) -- multiply scalar s time vector u; n specifies u's arity
Scale : int : 2 : scalar*vector multiplication
# InvScale<n>(s,u) -- divide vector u by scalar s; n specifies u's arity
InvScale : int : 2 : vector/scalar division
CL : : 1 : linear anisotropy measures
PrincipleEvec : ty : 2 : principle eigenvector; ty is result vector type
Subscript : ty : 2 :
Floor : int : 1 : compute real floor of a vector
#
### conversions; the real to int forms are vector ops
IntToReal : : 1 :
TruncToInt : int : 1 :
RoundToInt : int : 1 :
CeilToInt : int : 1 :
FloorToInt : int : 1 :
#
### image/kernel operations
LoadVoxels : RawTypes.ty * int : 1 : load a vector of voxel values from an address
# PosToImgSpace<V>(u) -- transforms the world-space position u into the image-space specified by V.
PosToImgSpace : ImageInfo.info : 1 : transform a world-space position to image-space
# GradToWorldSpace<V>(u) -- transforms the image-space gradient vector u to world space
GradToWorldSpace : ImageInfo.info : 1 : transform an image-space gradient to world-space