X86_AVX512VNNI: test _mm256_dpbusd_epi32 too

On RHEL9 the GCC is new enough to support AVX512-VNNI, but its assembler
(binutils) is not and errors with

```
Error: unsupported instruction vpdpbusd
```

This was already addressed earlier in
https://github.com/zlib-ng/zlib-ng/pull/1562 to some extent, except that
a check for `_mm256_dpbusd_epi32` was not added, which is what the
assembler errors over.

diff --git a/cmake/detect-intrinsics.cmake b/cmake/detect-intrinsics.cmake
index b96ac0a4428a9fdc6f153a715b7c0671a1da682c..e953346f3fb4b828dd716a8ff2567e7884086ad5 100644 (file)
--- a/cmake/detect-intrinsics.cmake
+++ b/cmake/detect-intrinsics.cmake
@@ -137,11 +137,15 @@ macro(check_avx512vnni_intrinsics)
     set(CMAKE_REQUIRED_FLAGS "${AVX512VNNIFLAG} ${NATIVEFLAG} ${ZNOLTOFLAG}")
     check_c_source_compiles(
         "#include <immintrin.h>
-        __m512i f(__m512i x, __m512i y) {
-            __m512i z = _mm512_setzero_epi32();
-            return _mm512_dpbusd_epi32(z, x, y);
-        }
-        int main(void) { return 0; }"
+        int main(void) {
+            const __m512i z512 = _mm512_setzero_si512();
+            const __m256i z256 = _mm256_setzero_si256();
+            volatile __m512i r512 = _mm512_dpbusd_epi32(z512, z512, z512);
+            volatile __m256i r256 = _mm256_dpbusd_epi32(z256, z256, z256);
+            (void)r512;
+            (void)r256;
+            return 0;
+        }"
         HAVE_AVX512VNNI_INTRIN
     )
     set(CMAKE_REQUIRED_FLAGS)

diff --git a/configure b/configure
index 6eda42e7c88a75e794b5d611cc4a50bc363b9d1f..832b42ac102835d46ace868165c017e5b2bc6f7b 100755 (executable)
--- a/configure
+++ b/configure
@@ -1140,11 +1140,15 @@ check_avx512vnni_intrinsics() {
     # Check whether compiler supports AVX512-VNNI intrinsics
     cat > $test.c << EOF
 #include <immintrin.h>
-__m512i f(__m512i x, __m512i y) {
-    __m512i z = _mm512_setzero_epi32();
-    return _mm512_dpbusd_epi32(z, x, y);
+int main(void) {
+    const __m512i z512 = _mm512_setzero_si512();
+    const __m256i z256 = _mm256_setzero_si256();
+    volatile __m512i r512 = _mm512_dpbusd_epi32(z512, z512, z512);
+    volatile __m256i r256 = _mm256_dpbusd_epi32(z256, z256, z256);
+    (void)r512;
+    (void)r256;
+    return 0;
 }
-int main(void) { return 0; }
 EOF
     if try ${CC} ${CFLAGS} ${avx512vnniflag} $test.c; then
         echo "Checking for AVX512VNNI intrinsics ... Yes." | tee -a configure.log