Skip to content

[libclc] Improving vector code generated from scalar code #140008

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 1 commit into from
May 16, 2025
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
135 changes: 90 additions & 45 deletions libclc/clc/include/clc/clcmacro.h
Original file line number Diff line number Diff line change
Expand Up @@ -14,149 +14,194 @@

#define _CLC_UNARY_VECTORIZE(DECLSPEC, RET_TYPE, FUNCTION, ARG1_TYPE) \
DECLSPEC RET_TYPE##2 FUNCTION(ARG1_TYPE##2 x) { \
return (RET_TYPE##2)(FUNCTION(x.x), FUNCTION(x.y)); \
return (RET_TYPE##2)(FUNCTION(x.s0), FUNCTION(x.s1)); \
} \
\
DECLSPEC RET_TYPE##3 FUNCTION(ARG1_TYPE##3 x) { \
return (RET_TYPE##3)(FUNCTION(x.x), FUNCTION(x.y), FUNCTION(x.z)); \
return (RET_TYPE##3)(FUNCTION(x.s0), FUNCTION(x.s1), FUNCTION(x.s2)); \
} \
\
DECLSPEC RET_TYPE##4 FUNCTION(ARG1_TYPE##4 x) { \
return (RET_TYPE##4)(FUNCTION(x.lo), FUNCTION(x.hi)); \
return (RET_TYPE##4)(FUNCTION(x.s0), FUNCTION(x.s1), FUNCTION(x.s2), \
FUNCTION(x.s3)); \
} \
\
DECLSPEC RET_TYPE##8 FUNCTION(ARG1_TYPE##8 x) { \
return (RET_TYPE##8)(FUNCTION(x.lo), FUNCTION(x.hi)); \
return (RET_TYPE##8)(FUNCTION(x.s0), FUNCTION(x.s1), FUNCTION(x.s2), \
FUNCTION(x.s3), FUNCTION(x.s4), FUNCTION(x.s5), \
FUNCTION(x.s6), FUNCTION(x.s7)); \
} \
\
DECLSPEC RET_TYPE##16 FUNCTION(ARG1_TYPE##16 x) { \
return (RET_TYPE##16)(FUNCTION(x.lo), FUNCTION(x.hi)); \
return (RET_TYPE##16)( \
FUNCTION(x.s0), FUNCTION(x.s1), FUNCTION(x.s2), FUNCTION(x.s3), \
FUNCTION(x.s4), FUNCTION(x.s5), FUNCTION(x.s6), FUNCTION(x.s7), \
FUNCTION(x.s8), FUNCTION(x.s9), FUNCTION(x.sa), FUNCTION(x.sb), \
FUNCTION(x.sc), FUNCTION(x.sd), FUNCTION(x.se), FUNCTION(x.sf)); \
}

#define _CLC_BINARY_VECTORIZE(DECLSPEC, RET_TYPE, FUNCTION, ARG1_TYPE, \
ARG2_TYPE) \
DECLSPEC RET_TYPE##2 FUNCTION(ARG1_TYPE##2 x, ARG2_TYPE##2 y) { \
return (RET_TYPE##2)(FUNCTION(x.x, y.x), FUNCTION(x.y, y.y)); \
return (RET_TYPE##2)(FUNCTION(x.s0, y.s0), FUNCTION(x.s1, y.s1)); \
} \
\
DECLSPEC RET_TYPE##3 FUNCTION(ARG1_TYPE##3 x, ARG2_TYPE##3 y) { \
return (RET_TYPE##3)(FUNCTION(x.x, y.x), FUNCTION(x.y, y.y), \
FUNCTION(x.z, y.z)); \
return (RET_TYPE##3)(FUNCTION(x.s0, y.s0), FUNCTION(x.s1, y.s1), \
FUNCTION(x.s2, y.s2)); \
} \
\
DECLSPEC RET_TYPE##4 FUNCTION(ARG1_TYPE##4 x, ARG2_TYPE##4 y) { \
return (RET_TYPE##4)(FUNCTION(x.lo, y.lo), FUNCTION(x.hi, y.hi)); \
return (RET_TYPE##4)(FUNCTION(x.s0, y.s0), FUNCTION(x.s1, y.s1), \
FUNCTION(x.s2, y.s2), FUNCTION(x.s3, y.s3)); \
} \
\
DECLSPEC RET_TYPE##8 FUNCTION(ARG1_TYPE##8 x, ARG2_TYPE##8 y) { \
return (RET_TYPE##8)(FUNCTION(x.lo, y.lo), FUNCTION(x.hi, y.hi)); \
return (RET_TYPE##8)(FUNCTION(x.s0, y.s0), FUNCTION(x.s1, y.s1), \
FUNCTION(x.s2, y.s2), FUNCTION(x.s3, y.s3), \
FUNCTION(x.s4, y.s4), FUNCTION(x.s5, y.s5), \
FUNCTION(x.s6, y.s6), FUNCTION(x.s7, y.s7)); \
} \
\
DECLSPEC RET_TYPE##16 FUNCTION(ARG1_TYPE##16 x, ARG2_TYPE##16 y) { \
return (RET_TYPE##16)(FUNCTION(x.lo, y.lo), FUNCTION(x.hi, y.hi)); \
return (RET_TYPE##16)( \
FUNCTION(x.s0, y.s0), FUNCTION(x.s1, y.s1), FUNCTION(x.s2, y.s2), \
FUNCTION(x.s3, y.s3), FUNCTION(x.s4, y.s4), FUNCTION(x.s5, y.s5), \
FUNCTION(x.s6, y.s6), FUNCTION(x.s7, y.s7), FUNCTION(x.s8, y.s8), \
FUNCTION(x.s9, y.s9), FUNCTION(x.sa, y.sa), FUNCTION(x.sb, y.sb), \
FUNCTION(x.sc, y.sc), FUNCTION(x.sd, y.sd), FUNCTION(x.se, y.se), \
FUNCTION(x.sf, y.sf)); \
}

#define _CLC_V_S_V_VECTORIZE(DECLSPEC, RET_TYPE, FUNCTION, ARG1_TYPE, \
ARG2_TYPE) \
DECLSPEC RET_TYPE##2 FUNCTION(ARG1_TYPE x, ARG2_TYPE##2 y) { \
return (RET_TYPE##2)(FUNCTION(x, y.lo), FUNCTION(x, y.hi)); \
return (RET_TYPE##2)(FUNCTION(x, y.s0), FUNCTION(x, y.s1)); \
} \
\
DECLSPEC RET_TYPE##3 FUNCTION(ARG1_TYPE x, ARG2_TYPE##3 y) { \
return (RET_TYPE##3)(FUNCTION(x, y.x), FUNCTION(x, y.y), \
FUNCTION(x, y.z)); \
return (RET_TYPE##3)(FUNCTION(x, y.s0), FUNCTION(x, y.s1), \
FUNCTION(x, y.s2)); \
} \
\
DECLSPEC RET_TYPE##4 FUNCTION(ARG1_TYPE x, ARG2_TYPE##4 y) { \
return (RET_TYPE##4)(FUNCTION(x, y.lo), FUNCTION(x, y.hi)); \
return (RET_TYPE##4)(FUNCTION(x, y.s0), FUNCTION(x, y.s1), \
FUNCTION(x, y.s2), FUNCTION(x, y.s3)); \
} \
\
DECLSPEC RET_TYPE##8 FUNCTION(ARG1_TYPE x, ARG2_TYPE##8 y) { \
return (RET_TYPE##8)(FUNCTION(x, y.lo), FUNCTION(x, y.hi)); \
return (RET_TYPE##8)(FUNCTION(x, y.s0), FUNCTION(x, y.s1), \
FUNCTION(x, y.s2), FUNCTION(x, y.s3), \
FUNCTION(x, y.s4), FUNCTION(x, y.s5), \
FUNCTION(x, y.s6), FUNCTION(x, y.s7)); \
} \
\
DECLSPEC RET_TYPE##16 FUNCTION(ARG1_TYPE x, ARG2_TYPE##16 y) { \
return (RET_TYPE##16)(FUNCTION(x, y.lo), FUNCTION(x, y.hi)); \
return (RET_TYPE##16)( \
FUNCTION(x, y.s0), FUNCTION(x, y.s1), FUNCTION(x, y.s2), \
FUNCTION(x, y.s3), FUNCTION(x, y.s4), FUNCTION(x, y.s5), \
FUNCTION(x, y.s6), FUNCTION(x, y.s7), FUNCTION(x, y.s8), \
FUNCTION(x, y.s9), FUNCTION(x, y.sa), FUNCTION(x, y.sb), \
FUNCTION(x, y.sc), FUNCTION(x, y.sd), FUNCTION(x, y.se), \
FUNCTION(x, y.sf)); \
}

#define _CLC_TERNARY_VECTORIZE(DECLSPEC, RET_TYPE, FUNCTION, ARG1_TYPE, \
ARG2_TYPE, ARG3_TYPE) \
DECLSPEC RET_TYPE##2 FUNCTION(ARG1_TYPE##2 x, ARG2_TYPE##2 y, \
ARG3_TYPE##2 z) { \
return (RET_TYPE##2)(FUNCTION(x.x, y.x, z.x), FUNCTION(x.y, y.y, z.y)); \
return (RET_TYPE##2)(FUNCTION(x.s0, y.s0, z.s0), \
FUNCTION(x.s1, y.s1, z.s1)); \
} \
\
DECLSPEC RET_TYPE##3 FUNCTION(ARG1_TYPE##3 x, ARG2_TYPE##3 y, \
ARG3_TYPE##3 z) { \
return (RET_TYPE##3)(FUNCTION(x.x, y.x, z.x), FUNCTION(x.y, y.y, z.y), \
FUNCTION(x.z, y.z, z.z)); \
return (RET_TYPE##3)(FUNCTION(x.s0, y.s0, z.s0), \
FUNCTION(x.s1, y.s1, z.s1), \
FUNCTION(x.s2, y.s2, z.s2)); \
} \
\
DECLSPEC RET_TYPE##4 FUNCTION(ARG1_TYPE##4 x, ARG2_TYPE##4 y, \
ARG3_TYPE##4 z) { \
return (RET_TYPE##4)(FUNCTION(x.lo, y.lo, z.lo), \
FUNCTION(x.hi, y.hi, z.hi)); \
return (RET_TYPE##4)( \
FUNCTION(x.s0, y.s0, z.s0), FUNCTION(x.s1, y.s1, z.s1), \
FUNCTION(x.s2, y.s2, z.s2), FUNCTION(x.s3, y.s3, z.s3)); \
} \
\
DECLSPEC RET_TYPE##8 FUNCTION(ARG1_TYPE##8 x, ARG2_TYPE##8 y, \
ARG3_TYPE##8 z) { \
return (RET_TYPE##8)(FUNCTION(x.lo, y.lo, z.lo), \
FUNCTION(x.hi, y.hi, z.hi)); \
return (RET_TYPE##8)( \
FUNCTION(x.s0, y.s0, z.s0), FUNCTION(x.s1, y.s1, z.s1), \
FUNCTION(x.s2, y.s2, z.s2), FUNCTION(x.s3, y.s3, z.s3), \
FUNCTION(x.s4, y.s4, z.s4), FUNCTION(x.s5, y.s5, z.s5), \
FUNCTION(x.s6, y.s6, z.s6), FUNCTION(x.s7, y.s7, z.s7)); \
} \
\
DECLSPEC RET_TYPE##16 FUNCTION(ARG1_TYPE##16 x, ARG2_TYPE##16 y, \
ARG3_TYPE##16 z) { \
return (RET_TYPE##16)(FUNCTION(x.lo, y.lo, z.lo), \
FUNCTION(x.hi, y.hi, z.hi)); \
return (RET_TYPE##16)( \
FUNCTION(x.s0, y.s0, z.s0), FUNCTION(x.s1, y.s1, z.s1), \
FUNCTION(x.s2, y.s2, z.s2), FUNCTION(x.s3, y.s3, z.s3), \
FUNCTION(x.s4, y.s4, z.s4), FUNCTION(x.s5, y.s5, z.s5), \
FUNCTION(x.s6, y.s6, z.s6), FUNCTION(x.s7, y.s7, z.s7), \
FUNCTION(x.s8, y.s8, z.s8), FUNCTION(x.s9, y.s9, z.s9), \
FUNCTION(x.sa, y.sa, z.sa), FUNCTION(x.sb, y.sb, z.sb), \
FUNCTION(x.sc, y.sc, z.sc), FUNCTION(x.sd, y.sd, z.sd), \
FUNCTION(x.se, y.se, z.se), FUNCTION(x.sf, y.sf, z.sf)); \
}

#define _CLC_V_V_VP_VECTORIZE(DECLSPEC, RET_TYPE, FUNCTION, ARG1_TYPE, \
ADDR_SPACE, ARG2_TYPE) \
DECLSPEC __CLC_XCONCAT(RET_TYPE, 2) \
FUNCTION(__CLC_XCONCAT(ARG1_TYPE, 2) x, \
ADDR_SPACE __CLC_XCONCAT(ARG2_TYPE, 2) * y) { \
return (__CLC_XCONCAT(RET_TYPE, 2))( \
FUNCTION(x.x, (ADDR_SPACE ARG2_TYPE *)y), \
FUNCTION(x.y, \
(ADDR_SPACE ARG2_TYPE *)((ADDR_SPACE ARG2_TYPE *)y + 1))); \
ADDR_SPACE ARG2_TYPE *ptr = (ADDR_SPACE ARG2_TYPE *)y; \
return (__CLC_XCONCAT(RET_TYPE, 2))(FUNCTION(x.s0, ptr), \
FUNCTION(x.s1, ptr + 1)); \
} \
\
DECLSPEC __CLC_XCONCAT(RET_TYPE, 3) \
FUNCTION(__CLC_XCONCAT(ARG1_TYPE, 3) x, \
ADDR_SPACE __CLC_XCONCAT(ARG2_TYPE, 3) * y) { \
return (__CLC_XCONCAT(RET_TYPE, 3))( \
FUNCTION(x.x, (ADDR_SPACE ARG2_TYPE *)y), \
FUNCTION(x.y, \
(ADDR_SPACE ARG2_TYPE *)((ADDR_SPACE ARG2_TYPE *)y + 1)), \
FUNCTION(x.z, \
(ADDR_SPACE ARG2_TYPE *)((ADDR_SPACE ARG2_TYPE *)y + 2))); \
ADDR_SPACE ARG2_TYPE *ptr = (ADDR_SPACE ARG2_TYPE *)y; \
return (__CLC_XCONCAT(RET_TYPE, 3))(FUNCTION(x.s0, ptr), \
FUNCTION(x.s1, ptr + 1), \
FUNCTION(x.s2, ptr + 2)); \
} \
\
DECLSPEC __CLC_XCONCAT(RET_TYPE, 4) \
FUNCTION(__CLC_XCONCAT(ARG1_TYPE, 4) x, \
ADDR_SPACE __CLC_XCONCAT(ARG2_TYPE, 4) * y) { \
ADDR_SPACE ARG2_TYPE *ptr = (ADDR_SPACE ARG2_TYPE *)y; \
return (__CLC_XCONCAT(RET_TYPE, 4))( \
FUNCTION(x.lo, (ADDR_SPACE __CLC_XCONCAT(ARG2_TYPE, 2) *)y), \
FUNCTION(x.hi, (ADDR_SPACE __CLC_XCONCAT( \
ARG2_TYPE, 2) *)((ADDR_SPACE ARG2_TYPE *)y + 2))); \
FUNCTION(x.s0, ptr), FUNCTION(x.s1, ptr + 1), FUNCTION(x.s2, ptr + 2), \
FUNCTION(x.s3, ptr + 3)); \
} \
\
DECLSPEC __CLC_XCONCAT(RET_TYPE, 8) \
FUNCTION(__CLC_XCONCAT(ARG1_TYPE, 8) x, \
ADDR_SPACE __CLC_XCONCAT(ARG2_TYPE, 8) * y) { \
ADDR_SPACE ARG2_TYPE *ptr = (ADDR_SPACE ARG2_TYPE *)y; \
return (__CLC_XCONCAT(RET_TYPE, 8))( \
FUNCTION(x.lo, (ADDR_SPACE __CLC_XCONCAT(ARG2_TYPE, 4) *)y), \
FUNCTION(x.hi, (ADDR_SPACE __CLC_XCONCAT( \
ARG2_TYPE, 4) *)((ADDR_SPACE ARG2_TYPE *)y + 4))); \
FUNCTION(x.s0, ptr), FUNCTION(x.s1, ptr + 1), FUNCTION(x.s2, ptr + 2), \
FUNCTION(x.s3, ptr + 3), FUNCTION(x.s4, ptr + 4), \
FUNCTION(x.s5, ptr + 5), FUNCTION(x.s6, ptr + 6), \
FUNCTION(x.s7, ptr + 7)); \
} \
\
DECLSPEC __CLC_XCONCAT(RET_TYPE, 16) \
FUNCTION(__CLC_XCONCAT(ARG1_TYPE, 16) x, \
ADDR_SPACE __CLC_XCONCAT(ARG2_TYPE, 16) * y) { \
ADDR_SPACE ARG2_TYPE *ptr = (ADDR_SPACE ARG2_TYPE *)y; \
return (__CLC_XCONCAT(RET_TYPE, 16))( \
FUNCTION(x.lo, (ADDR_SPACE __CLC_XCONCAT(ARG2_TYPE, 8) *)y), \
FUNCTION(x.hi, (ADDR_SPACE __CLC_XCONCAT( \
ARG2_TYPE, 8) *)((ADDR_SPACE ARG2_TYPE *)y + 8))); \
FUNCTION(x.s0, ptr), FUNCTION(x.s1, ptr + 1), FUNCTION(x.s2, ptr + 2), \
FUNCTION(x.s3, ptr + 3), FUNCTION(x.s4, ptr + 4), \
FUNCTION(x.s5, ptr + 5), FUNCTION(x.s6, ptr + 6), \
FUNCTION(x.s7, ptr + 7), FUNCTION(x.s8, ptr + 8), \
FUNCTION(x.s9, ptr + 9), FUNCTION(x.sa, ptr + 10), \
FUNCTION(x.sb, ptr + 11), FUNCTION(x.sc, ptr + 12), \
FUNCTION(x.sd, ptr + 13), FUNCTION(x.se, ptr + 14), \
FUNCTION(x.sf, ptr + 15)); \
}

#define _CLC_DEFINE_BINARY_BUILTIN(RET_TYPE, FUNCTION, BUILTIN, ARG1_TYPE, \
Expand Down
Loading
Loading