wasm simd128: more powerpc fixes

simd-everywhere · May 20, 2023 · 7cb5691 · 7cb5691
1 parent 4de999a
commit 7cb5691
Show file tree

Hide file tree

Showing 10 changed files with 16 additions and 22 deletions.
diff --git a/.github/workflows/ci.yml b/.github/workflows/ci.yml
@@ -336,11 +336,11 @@ jobs:
           arch_deb: s390x
         # power: 150 failures: x86/{avx512/range,sse,sse3}, arm/neon/*, wasm_simd128/, wasm_relexad_simd/*
         # https://github.com/simd-everywhere/simde/issues/903
-        # - version: 12
-        #   distro: ubuntu-22.04
-        #   cross: power9
-        #   arch_gnu: powerpc64le
-        #   arch_deb: ppc64el
+        - version: 12
+          distro: ubuntu-22.04
+          cross: power9
+          arch_gnu: powerpc64le
+          arch_deb: ppc64el
     runs-on: ${{ matrix.distro }}
     steps:
     - uses: actions/checkout@v3

diff --git a/docker/cross-files/power9-gcc-12-ccache.cross b/docker/cross-files/power9-gcc-12-ccache.cross
@@ -13,6 +13,6 @@ cpp_args = ['-mcpu=power9', '-Wextra', '-Werror']
 
 [host_machine]
 system = 'linux'
-cpu_family = 'powerpc64'
+cpu_family = 'ppc64'
 cpu = 'ppc64el'
 endian = 'little'
diff --git a/docker/cross-files/power9-gcc-12.cross b/docker/cross-files/power9-gcc-12.cross
@@ -13,6 +13,6 @@ cpp_args = ['-mcpu=power9', '-Wextra', '-Werror']
 
 [host_machine]
 system = 'linux'
-cpu_family = 'powerpc64'
+cpu_family = 'ppc64'
 cpu = 'ppc64el'
 endian = 'little'
diff --git a/docker/cross-files/ppc64el-gcc-10-debflags.cross b/docker/cross-files/ppc64el-gcc-10-debflags.cross
@@ -15,6 +15,6 @@ cpp_link_args = ['-Wl,-z,relro', '-Wl,-z,now']
 
 [host_machine]
 system = 'linux'
-cpu_family = 'powerpc64'
+cpu_family = 'ppc64'
 cpu = 'ppc64el'
 endian = 'little'
diff --git a/docker/cross-files/ppc64el-gcc-10.cross b/docker/cross-files/ppc64el-gcc-10.cross
@@ -13,6 +13,6 @@ cpp_args = ['-Wextra', '-Werror', '-g',]
 
 [host_machine]
 system = 'linux'
-cpu_family = 'powerpc64'
+cpu_family = 'ppc64'
 cpu = 'ppc64el'
 endian = 'little'
diff --git a/docker/cross-files/ppc64el-gcc-11-debflags.cross b/docker/cross-files/ppc64el-gcc-11-debflags.cross
@@ -15,6 +15,6 @@ cpp_link_args = ['-Wl,-z,relro', '-Wl,-z,now']
 
 [host_machine]
 system = 'linux'
-cpu_family = 'powerpc64'
+cpu_family = 'ppc64'
 cpu = 'ppc64el'
 endian = 'little'
diff --git a/docker/cross-files/ppc64el-gcc-11.cross b/docker/cross-files/ppc64el-gcc-11.cross
@@ -13,6 +13,6 @@ cpp_args = ['-Wextra', '-Werror', '-g',]
 
 [host_machine]
 system = 'linux'
-cpu_family = 'powerpc64'
+cpu_family = 'ppc64'
 cpu = 'ppc64el'
 endian = 'little'
diff --git a/docker/cross-files/ppc64el-gcc-12-debflags.cross b/docker/cross-files/ppc64el-gcc-12-debflags.cross
@@ -15,6 +15,6 @@ cpp_link_args = ['-Wl,-z,relro', '-Wl,-z,now']
 
 [host_machine]
 system = 'linux'
-cpu_family = 'powerpc64'
+cpu_family = 'ppc64'
 cpu = 'ppc64el'
 endian = 'little'
diff --git a/simde/arm/neon/dup_n.h b/simde/arm/neon/dup_n.h
@@ -338,7 +338,7 @@ simde_vdupq_n_f16(simde_float16 value) {
     return simde_float16x8_from_private(r_);
   #endif
 }
-#define simde_vmovq_n_f32 simde_vdupq_n_f32
+#define simde_vmovq_n_f16 simde_vdupq_n_f16
 #if defined(SIMDE_ARM_NEON_A32V7_ENABLE_NATIVE_ALIASES)
   #undef vdupq_n_f16
   #define vdupq_n_f16(value) simde_vdupq_n_f16((value))

diff --git a/simde/wasm/simd128.h b/simde/wasm/simd128.h
@@ -4367,7 +4367,7 @@ simde_wasm_u16x8_shr (simde_v128_t a, uint32_t count) {
     #elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
       r_.neon_u16 = vshlq_u16(a_.neon_u16, vdupq_n_s16(-HEDLEY_STATIC_CAST(int16_t, count & 15)));
     #elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
-      r_.altivec_i16 = vec_sra(a_.altivec_i16, vec_splats(HEDLEY_STATIC_CAST(unsigned short, count & 15)));
+      r_.altivec_u16 = vec_sr(a_.altivec_u16, vec_splats(HEDLEY_STATIC_CAST(unsigned short, count & 15)));
     #elif defined(SIMDE_VECTOR_SUBSCRIPT) && defined(SIMDE_VECTOR_SCALAR)
       r_.u16 = a_.u16 >> (count & 15);
     #else
@@ -4399,7 +4399,7 @@ simde_wasm_u32x4_shr (simde_v128_t a, uint32_t count) {
     #elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
       r_.neon_u32 = vshlq_u32(a_.neon_u32, vdupq_n_s32(-HEDLEY_STATIC_CAST(int32_t, count & 31)));
     #elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
-      r_.altivec_i32 = vec_sra(a_.altivec_i32, vec_splats(HEDLEY_STATIC_CAST(unsigned int, count & 31)));
+      r_.altivec_u32 = vec_sr(a_.altivec_u32, vec_splats(HEDLEY_STATIC_CAST(unsigned int, count & 31)));
     #elif defined(SIMDE_VECTOR_SUBSCRIPT) && defined(SIMDE_VECTOR_SCALAR)
       r_.u32 = a_.u32 >> (count & 31);
     #else
@@ -4434,7 +4434,7 @@ simde_wasm_u64x2_shr (simde_v128_t a, uint32_t count) {
     #elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
       r_.neon_u64 = vshlq_u64(a_.neon_u64, vdupq_n_s64(-HEDLEY_STATIC_CAST(int64_t, count & 63)));
     #elif defined(SIMDE_POWER_ALTIVEC_P8_NATIVE)
-      r_.altivec_i64 = vec_sra(a_.altivec_i64, vec_splats(HEDLEY_STATIC_CAST(unsigned long long, count & 63)));
+      r_.altivec_u64 = vec_sr(a_.altivec_u64, vec_splats(HEDLEY_STATIC_CAST(unsigned long long, count & 63)));
     #elif defined(SIMDE_VECTOR_SUBSCRIPT) && defined(SIMDE_VECTOR_SCALAR)
       r_.u64 = a_.u64 >> (count & 63);
     #else
@@ -4820,12 +4820,6 @@ simde_wasm_i16x8_mul (simde_v128_t a, simde_v128_t b) {
       r_.sse_m128i = _mm_mullo_epi16(a_.sse_m128i, b_.sse_m128i);
     #elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
       r_.neon_i16 = vmulq_s16(a_.neon_i16, b_.neon_i16);
-    #elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
-      r_.altivec_i16 =
-        vec_pack(
-          vec_mule(a_.altivec_i16, b_.altivec_i16),
-          vec_mulo(a_.altivec_i16, b_.altivec_i16)
-        );
     #elif defined(SIMDE_VECTOR_SUBSCRIPT)
       r_.i16 = a_.i16 * b_.i16;
     #else