Fix duplication of pk_add_f16 symbols (#1858)

11e4082d · Bartłomiej Kocot · GitHub · 9ee69dd2 · 11e4082d
Unverified Commit 11e4082d authored Feb 04, 2025 by Bartłomiej Kocot Committed by GitHub Feb 04, 2025
Hide whitespace changes
Inline Side-by-side

Showing with 2 additions and 2 deletions

include/ck_tile/core/numeric/vector_type.hpp include/ck_tile/core/numeric/vector_type.hpp +2 -2

No files found.
--- a/include/ck_tile/core/numeric/vector_type.hpp
+++ b/include/ck_tile/core/numeric/vector_type.hpp
@@ -200,7 +200,7 @@ using bf8x32_t = bf8_t __attribute((ext_vector_type(32)));
 using bf8x64_t = bf8_t __attribute((ext_vector_type(64)));
 #endif

-__host__ fp16x2_t pk_add_f16(const fp16x2_t& x, const fp16x2_t& y)
+CK_TILE_HOST fp16x2_t pk_add_f16(const fp16x2_t& x, const fp16x2_t& y)
 {
    fp16x2_t vector_res;

@@ -210,7 +210,7 @@ __host__ fp16x2_t pk_add_f16(const fp16x2_t& x, const fp16x2_t& y)
    return vector_res;
 }

-__device__ fp16x2_t pk_add_f16(const fp16x2_t& x, const fp16x2_t& y)
+CK_TILE_DEVICE fp16x2_t pk_add_f16(const fp16x2_t& x, const fp16x2_t& y)
 {
    fp16x2_t c;
    asm volatile("v_pk_add_f16 %0, %1, %2" : "=v"(c) : "v"(x), "v"(y));