From 66ec3a0717e767c6773f051fcf60b876d9c253e6 Mon Sep 17 00:00:00 2001
From: Aaron Teo <aaron.teo1@ibm.com>
Date: Sun, 26 Oct 2025 13:13:54 +0800
Subject: [PATCH] ggml: initial s390 feat detection

Signed-off-by: Aaron Teo <aaron.teo1@ibm.com>
---
 ggml/src/CMakeLists.txt                   |  6 +--
 ggml/src/ggml-cpu/CMakeLists.txt          |  2 +
 ggml/src/ggml-cpu/arch/s390/cpu-feats.cpp | 59 +++++++++++++++++++++++
 3 files changed, 64 insertions(+), 3 deletions(-)
 create mode 100644 ggml/src/ggml-cpu/arch/s390/cpu-feats.cpp

diff --git a/ggml/src/CMakeLists.txt b/ggml/src/CMakeLists.txt
index 3356ef550d..01c6032806 100644
--- a/ggml/src/CMakeLists.txt
+++ b/ggml/src/CMakeLists.txt
@@ -377,9 +377,9 @@ if (GGML_CPU_ALL_VARIANTS)
         endif()
     elseif (GGML_SYSTEM_ARCH STREQUAL "s390x")
         if (CMAKE_SYSTEM_NAME MATCHES "Linux")
-            ggml_add_cpu_backend_variant(s390x_z15  Z15 VXE)
-            # ggml_add_cpu_backend_variant(s390x_z16  Z16 VXE)
-            # ggml_add_cpu_backend_variant(s390x_z17  Z17 VXE)
+            ggml_add_cpu_backend_variant(s390x)
+            ggml_add_cpu_backend_variant(z15    Z15 VXE)
+            ggml_add_cpu_backend_variant(z16    Z16 VXE)
         else()
             message(FATAL_ERROR "Unsupported s390x target OS: ${CMAKE_SYSTEM_NAME}")
         endif()
diff --git a/ggml/src/ggml-cpu/CMakeLists.txt b/ggml/src/ggml-cpu/CMakeLists.txt
index 34323afa07..8af62a4d1e 100644
--- a/ggml/src/ggml-cpu/CMakeLists.txt
+++ b/ggml/src/ggml-cpu/CMakeLists.txt
@@ -509,6 +509,8 @@ function(ggml_add_cpu_backend_variant_impl tag_name)
             list(APPEND ARCH_FLAGS -mvx -mzvector)
             list(APPEND ARCH_DEFINITIONS GGML_VXE)
         endif()
+
+        ggml_add_cpu_backend_features(${GGML_CPU_NAME} arm ${ARCH_DEFINITIONS})
     elseif (CMAKE_SYSTEM_PROCESSOR MATCHES "wasm")
         message(STATUS "Wasm detected")
         list (APPEND GGML_CPU_SOURCES ggml-cpu/arch/wasm/quants.c)
diff --git a/ggml/src/ggml-cpu/arch/s390/cpu-feats.cpp b/ggml/src/ggml-cpu/arch/s390/cpu-feats.cpp
new file mode 100644
index 0000000000..1fd925d039
--- /dev/null
+++ b/ggml/src/ggml-cpu/arch/s390/cpu-feats.cpp
@@ -0,0 +1,59 @@
+#include "ggml-backend-impl.h"
+
+#if defined(__s390x__)
+#include <sys/auxv.h>
+
+// find hwcap bits in asm/elf.h
+#ifndef HWCAP_VXRS_EXT
+#define HWCAP_VXRS_EXT (1 << 13)
+#endif
+
+#ifndef HWCAP_VXRS_EXT2
+#define HWCAP_VXRS_EXT2 (1 << 15)
+#endif
+
+#ifndef HWCAP_NNPA
+#define HWCAP_NNPA (1 << 20)
+#endif
+
+struct s390x_features {
+    bool has_vxe  = false;
+    bool has_vxe2 = false;
+    bool has_nnpa = false;
+
+    s390x_features() {
+        uint32_t hwcap  = getauxval(AT_HWCAP);
+        uint32_t hwcap2 = getauxval(AT_HWCAP2);
+
+        has_vxe  = !!(hwcap & HWCAP_VXRS_EXT);
+        has_vxe2 = !!(hwcap & HWCAP_VXRS_EXT2);
+        has_nnpa = !!(hwcap & HWCAP_NNPA);
+
+        GGML_LOG_INFO("s390x features detected: VXE=%d, VXE2=%d, NNPA=%d",
+                      has_vxe, has_vxe2, has_nnpa);
+    }
+};
+
+static int ggml_backend_cpu_s390x_score() {
+    int score = 1;
+    s390x_features sf;
+
+#ifdef GGML_USE_VXE
+    if (!sf.has_vxe) { return 0; }
+    score += 1 << 1;
+#endif
+#ifdef GGML_USE_VXE2
+    if (!sf.has_vxe2) { return 0; }
+    score += 1 << 2;
+#endif
+#ifdef GGML_USE_NNPA
+    if (!sf.has_nnpa) { return 0; }
+    score += 1 << 3;
+#endif
+
+    return score;
+}
+
+GGML_BACKEND_DL_SCORE_IMPL(ggml_backend_cpu_s390x_score)
+
+#endif  // __s390x__