diff --git a/include/libyuv/cpu_id.h b/include/libyuv/cpu_id.h
index 61a934ce2..c6983fb32 100644
--- a/include/libyuv/cpu_id.h
+++ b/include/libyuv/cpu_id.h
@@ -60,6 +60,7 @@ static const int kCpuHasAVX10_2 = 0x2000000;
 static const int kCpuHasAVXVNNI = 0x4000000;
 static const int kCpuHasAVXVNNIINT8 = 0x8000000;
 static const int kCpuHasAMXINT8 = 0x10000000;
+static const int kCpuHasAVX512BMM = 0x20000000;
 
 // These flags are only valid on LOONGARCH processors.
 static const int kCpuHasLOONGARCH = 0x20;
diff --git a/source/cpu_id.cc b/source/cpu_id.cc
index 0d7ea9a95..0cc46b10a 100644
--- a/source/cpu_id.cc
+++ b/source/cpu_id.cc
@@ -397,6 +397,7 @@ static SAFEBUFFERS int GetCpuFlags(void) {
   int cpu_info7[4] = {0, 0, 0, 0};
   int cpu_einfo7[4] = {0, 0, 0, 0};
   int cpu_info24[4] = {0, 0, 0, 0};
+  int cpu_info21[4] = {0, 0, 0, 0};
   int cpu_amdinfo21[4] = {0, 0, 0, 0};
   CpuId(0, 0, cpu_info0);
   CpuId(1, 0, cpu_info1);
@@ -405,6 +406,9 @@ static SAFEBUFFERS int GetCpuFlags(void) {
     CpuId(7, 1, cpu_einfo7);
     CpuId(0x80000021, 0, cpu_amdinfo21);
   }
+  if (cpu_info0[0] >= 0x21) {
+    CpuId(0x21, 0, cpu_info21);
+  }
   if (cpu_info0[0] >= 0x24) {
     CpuId(0x24, 0, cpu_info24);
   }
@@ -435,7 +439,8 @@ static SAFEBUFFERS int GetCpuFlags(void) {
                   ((cpu_info7[2] & 0x00000800) ? kCpuHasAVX512VNNI : 0) |
                   ((cpu_info7[2] & 0x00001000) ? kCpuHasAVX512VBITALG : 0) |
                   ((cpu_einfo7[3] & 0x00080000) ? kCpuHasAVX10 : 0) |
-                  ((cpu_info7[3] & 0x02000000) ? kCpuHasAMXINT8 : 0);
+                  ((cpu_info7[3] & 0x02000000) ? kCpuHasAMXINT8 : 0) |
+                  ((cpu_info21[0] & 0x00800000) ? kCpuHasAVX512BMM : 0);
       if (cpu_info0[0] >= 0x24 && (cpu_einfo7[3] & 0x00080000)) {
         cpu_info |= ((cpu_info24[1] & 0xFF) >= 2) ? kCpuHasAVX10_2 : 0;
       }
diff --git a/unit_test/cpu_test.cc b/unit_test/cpu_test.cc
index b24f3e250..b249e70df 100644
--- a/unit_test/cpu_test.cc
+++ b/unit_test/cpu_test.cc
@@ -189,6 +189,7 @@ TEST_F(LibYUVBaseTest, TestCpuHas) {
     int has_avxvnni = TestCpuFlag(kCpuHasAVXVNNI);
     int has_avxvnniint8 = TestCpuFlag(kCpuHasAVXVNNIINT8);
     int has_amxint8 = TestCpuFlag(kCpuHasAMXINT8);
+    int has_avx512bmm = TestCpuFlag(kCpuHasAVX512BMM);
     printf("Has X86 0x%x\n", has_x86);
     printf("Has SSE2 0x%x\n", has_sse2);
     printf("Has SSSE3 0x%x\n", has_ssse3);
@@ -211,6 +212,7 @@ TEST_F(LibYUVBaseTest, TestCpuHas) {
     printf("HAS AVXVNNI 0x%x\n", has_avxvnni);
     printf("Has AVXVNNIINT8 0x%x\n", has_avxvnniint8);
     printf("Has AMXINT8 0x%x\n", has_amxint8);
+    printf("Has AVX512BMM 0x%x\n", has_avx512bmm);
   }
 #endif  // defined(__i386__) || defined(__x86_64__) || defined(_M_IX86) ||
         // defined(_M_X64)
diff --git a/util/cpuid.c b/util/cpuid.c
index 38b2c0e9d..bbaea8398 100644
--- a/util/cpuid.c
+++ b/util/cpuid.c
@@ -15,6 +15,8 @@
 #ifdef __linux__
 #include <ctype.h>
 #include <sys/utsname.h>
+#include <signal.h>
+#include <setjmp.h>
 #endif
 
 #include "libyuv/cpu_id.h"
@@ -40,6 +42,14 @@ static void KernelVersion(int* version) {
 }
 #endif
 
+#ifdef __linux__
+static sigjmp_buf vdpphps_jmpbuf;
+static void vdpphps_sigill_handler(int sig) {
+  (void)sig;
+  siglongjmp(vdpphps_jmpbuf, 1);
+}
+#endif
+
 int main(int argc, const char* argv[]) {
   (void)argc;
   (void)argv;
@@ -182,6 +192,7 @@ int main(int argc, const char* argv[]) {
     int has_avxvnni = TestCpuFlag(kCpuHasAVXVNNI);
     int has_avxvnniint8 = TestCpuFlag(kCpuHasAVXVNNIINT8);
     int has_amxint8 = TestCpuFlag(kCpuHasAMXINT8);
+    int has_avx512bmm = TestCpuFlag(kCpuHasAVX512BMM);
     printf("Has X86 0x%x\n", has_x86);
     printf("Has SSE2 0x%x\n", has_sse2);
     printf("Has SSSE3 0x%x\n", has_ssse3);
@@ -204,6 +215,30 @@ int main(int argc, const char* argv[]) {
     printf("HAS AVXVNNI 0x%x\n", has_avxvnni);
     printf("Has AVXVNNIINT8 0x%x\n", has_avxvnniint8);
     printf("Has AMXINT8 0x%x\n", has_amxint8);
+    printf("Has AVX512BMM 0x%x\n", has_avx512bmm);
+
+#ifdef __linux__
+    // Test VDPPHPS instruction
+    {
+      struct sigaction act, oldact;
+      memset(&act, 0, sizeof(act));
+      act.sa_handler = vdpphps_sigill_handler;
+      sigaction(SIGILL, &act, &oldact);
+
+      printf("Testing VDPPHPS instruction... ");
+      fflush(stdout);
+
+      if (sigsetjmp(vdpphps_jmpbuf, 1) == 0) {
+        // VDPPHPS xmm0, xmm0, xmm0
+        __asm__ volatile("vdpphps %%xmm0, %%xmm0, %%xmm0" : : : "xmm0");
+        printf("Works!\n");
+      } else {
+        printf("Crashed (SIGILL)!\n");
+      }
+
+      sigaction(SIGILL, &oldact, NULL);
+    }
+#endif
   }
 #endif  // defined(__i386__) || defined(__x86_64__) || defined(_M_IX86) ||
         // defined(_M_X64)