Dev/simd for interp

.github/workflows/compilation_on_android_ubuntu.yml

@@ -158,6 +158,7 @@ jobs:
             "-DWAMR_BUILD_PERF_PROFILING=1",
             "-DWAMR_BUILD_REF_TYPES=1",
             "-DWAMR_BUILD_SIMD=1",
+            "-DWAMR_BUILD_LIB_SIMDE=1",
             "-DWAMR_BUILD_TAIL_CALL=1",
             "-DWAMR_DISABLE_HW_BOUND_CHECK=1",
             "-DWAMR_BUILD_MEMORY64=1",
@@ -178,11 +179,9 @@ jobs:
             make_options_feature: "-DWAMR_BUILD_MULTI_MODULE=1"
           - make_options_run_mode: $MULTI_TIER_JIT_BUILD_OPTIONS
             make_options_feature: "-DWAMR_BUILD_MULTI_MODULE=1"
-          # SIMD only on JIT/AOT mode
+          # SIMD only on JIT/AOT/fast interpreter mode
           - make_options_run_mode: $CLASSIC_INTERP_BUILD_OPTIONS
             make_options_feature: "-DWAMR_BUILD_SIMD=1"
-          - make_options_run_mode: $FAST_INTERP_BUILD_OPTIONS
-            make_options_feature: "-DWAMR_BUILD_SIMD=1"
           # DEBUG_INTERP only on CLASSIC INTERP mode
           - make_options_run_mode: $AOT_BUILD_OPTIONS
             make_options_feature: "-DWAMR_BUILD_DEBUG_INTERP=1"
@@ -649,11 +648,9 @@ jobs:
             test_option: $WAMR_COMPILER_TEST_OPTIONS
         exclude:
           # incompatible modes and features
-          # classic-interp and fast-interp don't support simd
+          # classic-interp doesn't support simd
           - running_mode: "classic-interp"
             test_option: $SIMD_TEST_OPTIONS
-          - running_mode: "fast-interp"
-            test_option: $SIMD_TEST_OPTIONS
           # llvm jit doesn't support multi module
           - running_mode: "jit"
             test_option: $MULTI_MODULES_TEST_OPTIONS

.github/workflows/compilation_on_sgx.yml

@@ -49,7 +49,7 @@ env:
   # ref types enabled in wamrc by default, so we need to enable it for iwasm in AOT mode
   AOT_BUILD_OPTIONS: "-DWAMR_BUILD_AOT=1 -DWAMR_BUILD_FAST_INTERP=0 -DWAMR_BUILD_INTERP=0 -DWAMR_BUILD_JIT=0 -DWAMR_BUILD_LAZY_JIT=0 -DWAMR_BUILD_REF_TYPES=1"
   CLASSIC_INTERP_BUILD_OPTIONS: "-DWAMR_BUILD_AOT=0 -DWAMR_BUILD_FAST_INTERP=0 -DWAMR_BUILD_INTERP=1 -DWAMR_BUILD_JIT=0 -DWAMR_BUILD_LAZY_JIT=0"
-  FAST_INTERP_BUILD_OPTIONS: "-DWAMR_BUILD_AOT=0 -DWAMR_BUILD_FAST_INTERP=1 -DWAMR_BUILD_INTERP=1 -DWAMR_BUILD_JIT=0 -DWAMR_BUILD_LAZY_JIT=0"
+  FAST_INTERP_BUILD_OPTIONS: "-DWAMR_BUILD_AOT=0 -DWAMR_BUILD_FAST_INTERP=1 -DWAMR_BUILD_INTERP=1 -DWAMR_BUILD_JIT=0 -DWAMR_BUILD_LAZY_JIT=0 -DWAMR_BUILD_SIMD=0"
   FAST_JIT_BUILD_OPTIONS: "-DWAMR_BUILD_AOT=1 -DWAMR_BUILD_FAST_INTERP=0 -DWAMR_BUILD_INTERP=1 -DWAMR_BUILD_FAST_JIT=1 -DWAMR_BUILD_JIT=0 -DWAMR_BUILD_LAZY_JIT=1"
   LLVM_LAZY_JIT_BUILD_OPTIONS: "-DWAMR_BUILD_AOT=1 -DWAMR_BUILD_FAST_INTERP=0 -DWAMR_BUILD_INTERP=0 -DWAMR_BUILD_JIT=1 -DWAMR_BUILD_LAZY_JIT=1"
   LLVM_EAGER_JIT_BUILD_OPTIONS: "-DWAMR_BUILD_AOT=1 -DWAMR_BUILD_FAST_INTERP=0 -DWAMR_BUILD_INTERP=0 -DWAMR_BUILD_JIT=1 -DWAMR_BUILD_LAZY_JIT=0"
@@ -97,7 +97,7 @@ jobs:
             "-DWAMR_BUILD_PERF_PROFILING=1",
             "-DWAMR_BUILD_REF_TYPES=1",
             # doesn't support
-            # "-DWAMR_BUILD_SIMD=1",
+            "-DWAMR_BUILD_SIMD=0",
             "-DWAMR_BUILD_TAIL_CALL=1",
             "-DWAMR_DISABLE_HW_BOUND_CHECK=1",
             "-DWAMR_BUILD_SGX_IPFS=1",

build-scripts/config_common.cmake

@@ -300,6 +300,9 @@ endif ()
 if (WAMR_BUILD_LIB_RATS EQUAL 1)
   message ("     Lib rats enabled")
 endif()
+if ((WAMR_BUILD_LIB_SIMDE EQUAL 1))
+  message ("     Lib simde enabled")
+endif()
 ################## WAMR features ##################
 if (WAMR_BUILD_MULTI_MODULE EQUAL 1)
   add_definitions (-DWASM_ENABLE_MULTI_MODULE=1)
@@ -363,11 +366,17 @@ else ()
   message ("     Wakeup of blocking operations enabled")
 endif ()
 if (WAMR_BUILD_SIMD EQUAL 1)
-  if (NOT WAMR_BUILD_TARGET MATCHES "RISCV64.*")
-    add_definitions (-DWASM_ENABLE_SIMD=1)
-  else ()
+  if (WAMR_BUILD_FAST_INTERP EQUAL 1 AND WAMR_BUILD_SIMDE EQUAL 0)
+    set(SIMD_ENABLED 0)
+    message("     SIMD disabled for fast-interp as simde is not being built")
+  elseif (WAMR_BUILD_TARGET MATCHES "RISCV64.*")
+    set(SIMD_ENABLED 0)
     message ("     SIMD disabled due to not supported on target RISCV64")
+  else()
+    set(SIMD_ENABLED 1)
+    message ("     SIMD enabled")
   endif ()
+  add_definitions(-DWASM_ENABLE_SIMD=${SIMD_ENABLED})
 endif ()
 if (WAMR_BUILD_AOT_STACK_FRAME EQUAL 1)
   add_definitions (-DWASM_ENABLE_AOT_STACK_FRAME=1)

build-scripts/runtime_lib.cmake

@@ -155,6 +155,16 @@ if (WAMR_BUILD_LIB_RATS EQUAL 1)
     include (${IWASM_DIR}/libraries/lib-rats/lib_rats.cmake)
 endif ()
 
+if (WAMR_BUILD_SIMD EQUAL 1 AND WAMR_BUILD_FAST_INTERP EQUAL 1)
+    if (WAMR_BUILD_PLATFORM STREQUAL "windows")
+        message(STATUS "SIMDe doesnt support platform " ${WAMR_BUILD_PLATFORM})
+        set(WAMR_BUILD_SIMDE 0)
+    else()
+        include (${IWASM_DIR}/libraries/simde/simde.cmake)
+        set (WAMR_BUILD_SIMDE 1)
+    endif()
+endif ()
+
 if (WAMR_BUILD_WASM_CACHE EQUAL 1)
     include (${WAMR_ROOT_DIR}/build-scripts/involve_boringssl.cmake)
 endif ()

core/config.h

@@ -318,6 +318,12 @@
 #define WASM_ENABLE_SIMD 0
 #endif
 
+/* Disable SIMDe (used in the fast interpreter for SIMD opcodes)
+unless used elsewhere */
+#ifndef WASM_ENABLE_SIMDE
+#define WASM_ENABLE_SIMDE 0
+#endif
+
 /* GC performance profiling */
 #ifndef WASM_ENABLE_GC_PERF_PROFILING
 #define WASM_ENABLE_GC_PERF_PROFILING 0

core/iwasm/common/wasm_loader_common.c

@@ -151,7 +151,8 @@ is_valid_value_type(uint8 type)
 bool
 is_valid_value_type_for_interpreter(uint8 value_type)
 {
-#if (WASM_ENABLE_WAMR_COMPILER == 0) && (WASM_ENABLE_JIT == 0)
+#if (WASM_ENABLE_WAMR_COMPILER == 0) && (WASM_ENABLE_JIT == 0) \
+    && (WASM_ENABLE_FAST_INTERP == 0)
     /*
      * Note: regardless of WASM_ENABLE_SIMD, our interpreters don't have
      * SIMD implemented. It's safer to reject v128, especially for the

core/iwasm/common/wasm_runtime_common.h

@@ -37,6 +37,10 @@ extern "C" {
     do {                                   \
         *(int64 *)(addr) = (int64)(value); \
     } while (0)
+#define PUT_V128_TO_ADDR(addr, value) \
+    do {                              \
+        *(V128 *)(addr) = (value);    \
+    } while (0)
 #define PUT_F64_TO_ADDR(addr, value)           \
     do {                                       \
         *(float64 *)(addr) = (float64)(value); \
@@ -49,6 +53,7 @@ extern "C" {
 #define GET_I64_FROM_ADDR(addr) (*(int64 *)(addr))
 #define GET_F64_FROM_ADDR(addr) (*(float64 *)(addr))
 #define GET_REF_FROM_ADDR(addr) (*(void **)(addr))
+#define GET_V128_FROM_ADDR(addr) (*(V128 *)(addr))
 
 /* For STORE opcodes */
 #define STORE_I64 PUT_I64_TO_ADDR
@@ -68,13 +73,20 @@ STORE_U8(void *addr, uint8_t value)
     *(uint8 *)addr = value;
 }
 
+static inline void
+STORE_V128(void *addr, V128 value)
+{
+    *(V128 *)addr = value;
+}
+
 /* For LOAD opcodes */
 #define LOAD_I64(addr) (*(int64 *)(addr))
 #define LOAD_F64(addr) (*(float64 *)(addr))
 #define LOAD_I32(addr) (*(int32 *)(addr))
 #define LOAD_U32(addr) (*(uint32 *)(addr))
 #define LOAD_I16(addr) (*(int16 *)(addr))
 #define LOAD_U16(addr) (*(uint16 *)(addr))
+#define LOAD_V128(addr) (*(V128 *)(addr))
 
 #define STORE_PTR(addr, ptr)          \
     do {                              \
@@ -83,6 +95,15 @@ STORE_U8(void *addr, uint8_t value)
 
 #else /* WASM_CPU_SUPPORTS_UNALIGNED_ADDR_ACCESS != 0 */
 
+#define PUT_V128_TO_ADDR(addr, value)        \
+    do {                                     \
+        uint32 *addr_u32 = (uint32 *)(addr); \
+        addr_u32[0] = (value).i32x4[0];      \
+        addr_u32[1] = (value).i32x4[1];      \
+        addr_u32[2] = (value).i32x4[2];      \
+        addr_u32[3] = (value).i32x4[3];      \
+    } while (0)
+
 #define PUT_I64_TO_ADDR(addr, value)         \
     do {                                     \
         uint32 *addr_u32 = (uint32 *)(addr); \
@@ -124,6 +145,17 @@ STORE_U8(void *addr, uint8_t value)
     } while (0)
 #endif
 
+static inline V128
+GET_V128_FROM_ADDR(uint32 *addr)
+{
+    V128 ret;
+    ret.i32x4[0] = addr[0];
+    ret.i32x4[1] = addr[1];
+    ret.i32x4[2] = addr[2];
+    ret.i32x4[3] = addr[3];
+    return ret;
+}
+
 static inline int64
 GET_I64_FROM_ADDR(uint32 *addr)
 {
@@ -239,7 +271,94 @@ STORE_U16(void *addr, uint16_t value)
     ((uint8_t *)(addr))[0] = u.u8[0];
     ((uint8_t *)(addr))[1] = u.u8[1];
 }
+
+static inline void
+STORE_V128(void *addr, V128 value)
+{
+    uintptr_t addr_ = (uintptr_t)(addr);
+    union {
+        V128 val;
+        uint64 u64[2];
+        uint32 u32[4];
+        uint16 u16[8];
+        uint8 u8[16];
+    } u;
+
+    if ((addr_ & (uintptr_t)15) == 0) {
+        *(V128 *)addr = value;
+    }
+    else if ((addr_ & (uintptr_t)7) == 0) {
+        u.val = value;
+        ((uint64 *)(addr))[0] = u.u64[0];
+        ((uint64 *)(addr))[1] = u.u64[1];
+    }
+    else if ((addr_ & (uintptr_t)3) == 0) {
+        u.val = value;
+        ((uint32 *)addr)[0] = u.u32[0];
+        ((uint32 *)addr)[1] = u.u32[1];
+        ((uint32 *)addr)[2] = u.u32[2];
+        ((uint32 *)addr)[3] = u.u32[3];
+    }
+    else if ((addr_ & (uintptr_t)1) == 0) {
+        u.val = value;
+        ((uint16 *)addr)[0] = u.u16[0];
+        ((uint16 *)addr)[1] = u.u16[1];
+        ((uint16 *)addr)[2] = u.u16[2];
+        ((uint16 *)addr)[3] = u.u16[3];
+        ((uint16 *)addr)[4] = u.u16[4];
+        ((uint16 *)addr)[5] = u.u16[5];
+        ((uint16 *)addr)[6] = u.u16[6];
+        ((uint16 *)addr)[7] = u.u16[7];
+    }
+    else {
+        u.val = value;
+        for (int i = 0; i < 16; i++)
+            ((uint8 *)addr)[i] = u.u8[i];
+    }
+}
+
 /* For LOAD opcodes */
+static inline V128
+LOAD_V128(void *addr)
+{
+    uintptr_t addr1 = (uintptr_t)addr;
+    union {
+        V128 val;
+        uint64 u64[2];
+        uint32 u32[4];
+        uint16 u16[8];
+        uint8 u8[16];
+    } u;
+    if ((addr1 & (uintptr_t)15) == 0)
+        return *(V128 *)addr;
+
+    if ((addr1 & (uintptr_t)7) == 0) {
+        u.u64[0] = ((uint64 *)addr)[0];
+        u.u64[1] = ((uint64 *)addr)[1];
+    }
+    else if ((addr1 & (uintptr_t)3) == 0) {
+        u.u32[0] = ((uint32 *)addr)[0];
+        u.u32[1] = ((uint32 *)addr)[1];
+        u.u32[2] = ((uint32 *)addr)[2];
+        u.u32[3] = ((uint32 *)addr)[3];
+    }
+    else if ((addr1 & (uintptr_t)1) == 0) {
+        u.u16[0] = ((uint16 *)addr)[0];
+        u.u16[1] = ((uint16 *)addr)[1];
+        u.u16[2] = ((uint16 *)addr)[2];
+        u.u16[3] = ((uint16 *)addr)[3];
+        u.u16[4] = ((uint16 *)addr)[4];
+        u.u16[5] = ((uint16 *)addr)[5];
+        u.u16[6] = ((uint16 *)addr)[6];
+        u.u16[7] = ((uint16 *)addr)[7];
+    }
+    else {
+        for (int i = 0; i < 16; i++)
+            u.u8[i] = ((uint8 *)addr)[i];
+    }
+    return u.val;
+}
+
 static inline int64
 LOAD_I64(void *addr)
 {