llama: update vendored code to commit 46e3556 (#8308)

llama/ggml-cpu-quants.c

 /**
- * llama.cpp - commit ba1cb19cdd0d92e012e0f6e009e0620f854b6afd - do not edit this file
+ * llama.cpp - commit 46e3556e01b824e52395fb050b29804b6cff2a7c - do not edit this file
  *
  * MIT License
  *
@@ -129,10 +129,14 @@ static inline __m256 sum_i16_pairs_float(const __m256i x) {
 }
 
 static inline __m256 mul_sum_us8_pairs_float(const __m256i ax, const __m256i sy) {
-#if defined(__AVXVNNI__) || (defined(__AVX512VNNI__) && defined(__AVX512VL__))
+#if defined(__AVX512VNNI__) && defined(__AVX512VL__)
     const __m256i zero = _mm256_setzero_si256();
     const __m256i summed_pairs = _mm256_dpbusd_epi32(zero, ax, sy);
     return _mm256_cvtepi32_ps(summed_pairs);
+#elif defined(__AVXVNNI__)
+    const __m256i zero = _mm256_setzero_si256();
+    const __m256i summed_pairs = _mm256_dpbusd_avx_epi32(zero, ax, sy);
+    return _mm256_cvtepi32_ps(summed_pairs);
 #else
     // Perform multiplication and create 16-bit values
     const __m256i dot = _mm256_maddubs_epi16(ax, sy);

llama/ggml-cpu-quants.h

 /**
- * llama.cpp - commit ba1cb19cdd0d92e012e0f6e009e0620f854b6afd - do not edit this file
+ * llama.cpp - commit 46e3556e01b824e52395fb050b29804b6cff2a7c - do not edit this file
  *
  * MIT License
  *

llama/ggml-cpu-traits.cpp

 /**
- * llama.cpp - commit ba1cb19cdd0d92e012e0f6e009e0620f854b6afd - do not edit this file
+ * llama.cpp - commit 46e3556e01b824e52395fb050b29804b6cff2a7c - do not edit this file
  *
  * MIT License
  *

llama/ggml-cpu-traits.h

 /**
- * llama.cpp - commit ba1cb19cdd0d92e012e0f6e009e0620f854b6afd - do not edit this file
+ * llama.cpp - commit 46e3556e01b824e52395fb050b29804b6cff2a7c - do not edit this file
  *
  * MIT License
  *

llama/ggml-cpu.c

 /**
- * llama.cpp - commit ba1cb19cdd0d92e012e0f6e009e0620f854b6afd - do not edit this file
+ * llama.cpp - commit 46e3556e01b824e52395fb050b29804b6cff2a7c - do not edit this file
  *
  * MIT License
  *
@@ -1012,7 +1012,7 @@ inline static void __wasm_f16x4_store(ggml_fp16_t * p, v128_t x) {
 #define GGML_F16_STEP 32
 #define GGML_F16_EPR  4
 
-static inline __m128 __sse_f16x4_load(ggml_fp16_t *x) {
+static inline __m128 __sse_f16x4_load(const ggml_fp16_t * x) {
     float tmp[4];
 
     tmp[0] = GGML_FP16_TO_FP32(x[0]);
@@ -1023,7 +1023,7 @@ static inline __m128 __sse_f16x4_load(ggml_fp16_t *x) {
     return _mm_loadu_ps(tmp);
 }
 
-static inline void __sse_f16x4_store(ggml_fp16_t *x, __m128 y) {
+static inline void __sse_f16x4_store(ggml_fp16_t * x, __m128 y) {
     float arr[4];
 
     _mm_storeu_ps(arr, y);
@@ -7445,14 +7445,14 @@ static void ggml_compute_forward_mul_mat(
     if (src1_cont) {
         for (int64_t i13 = 0; i13 < ne13; i13++)
             for (int64_t i12 = 0; i12 < ne12; i12++)
-                if (!llamafile_sgemm(ne01, ne11, ne00/ggml_blck_size(src0->type),
+                if (!llamafile_sgemm(params,
+                                     ne01, ne11, ne00/ggml_blck_size(src0->type),
                                      (const char *)src0->data + i12/r2*nb02 + i13/r3*nb03,
                                      nb01/ggml_type_size(src0->type),
                                      (const char *)src1->data + i12*nb12 + i13*nb13,
                                      nb11/ggml_type_size(src1->type),
                                      (char *)dst->data + i12*nb2 + i13*nb3,
                                      nb1/ggml_type_size(dst->type),
-                                     ith, nth,
                                      src0->type,
                                      src1->type,
                                      dst->type))
@@ -7497,14 +7497,14 @@ UseGgmlGemm1:;
 
         for (int64_t i13 = 0; i13 < ne13; i13++)
             for (int64_t i12 = 0; i12 < ne12; i12++)
-                if (!llamafile_sgemm(ne01, ne11, ne00/ggml_blck_size(src0->type),
+                if (!llamafile_sgemm(params,
+                                     ne01, ne11, ne00/ggml_blck_size(src0->type),
                                      (const char *)src0->data + i12/r2*nb02 + i13/r3*nb03,
                                      nb01/ggml_type_size(src0->type),
                                      (const char *)wdata + (i12*ne11 + i13*ne12*ne11)*row_size,
                                      row_size/ggml_type_size(vec_dot_type),
                                      (char *)dst->data + i12*nb2 + i13*nb3,
                                      nb1/ggml_type_size(dst->type),
-                                     ith, nth,
                                      src0->type,
                                      vec_dot_type,
                                      dst->type))

llama/ggml-cpu.cpp

 /**
- * llama.cpp - commit ba1cb19cdd0d92e012e0f6e009e0620f854b6afd - do not edit this file
+ * llama.cpp - commit 46e3556e01b824e52395fb050b29804b6cff2a7c - do not edit this file
  *
  * MIT License
  *
@@ -419,8 +419,11 @@ static bool ggml_backend_cpu_device_supports_op(ggml_backend_dev_t dev, const st
     switch (op->op) {
         case GGML_OP_CPY:
             return
+                op->type != GGML_TYPE_IQ3_XXS &&
+                op->type != GGML_TYPE_IQ3_S   &&
                 op->type != GGML_TYPE_IQ2_XXS &&
                 op->type != GGML_TYPE_IQ2_XS  &&
+                op->type != GGML_TYPE_IQ2_S   &&
                 op->type != GGML_TYPE_IQ1_S   &&
                 op->type != GGML_TYPE_IQ1_M; // missing type_traits.from_float
         case GGML_OP_MUL_MAT:
@@ -544,6 +547,12 @@ static ggml_backend_feature * ggml_backend_cpu_get_features(ggml_backend_reg_t r
         if (ggml_cpu_has_sve()) {
             features.push_back({ "SVE", "1" });
         }
+        if (ggml_cpu_has_dotprod()) {
+            features.push_back({ "DOTPROD", "1" });
+        }
+        if (ggml_cpu_has_matmul_int8()) {
+            features.push_back({ "MATMUL_INT8", "1" });
+        }
         if (ggml_cpu_get_sve_cnt() > 0) {
             static std::string sve_cnt = std::to_string(ggml_cpu_get_sve_cnt());
             features.push_back({ "SVE_CNT", sve_cnt.c_str() });

llama/ggml-cpu.h

 /**
- * llama.cpp - commit ba1cb19cdd0d92e012e0f6e009e0620f854b6afd - do not edit this file
+ * llama.cpp - commit 46e3556e01b824e52395fb050b29804b6cff2a7c - do not edit this file
  *
  * MIT License
  *

llama/ggml-cuda.h

 /**
- * llama.cpp - commit ba1cb19cdd0d92e012e0f6e009e0620f854b6afd - do not edit this file
+ * llama.cpp - commit 46e3556e01b824e52395fb050b29804b6cff2a7c - do not edit this file
  *
  * MIT License
  *

llama/ggml-cuda/acc.cu

 /**
- * llama.cpp - commit ba1cb19cdd0d92e012e0f6e009e0620f854b6afd - do not edit this file
+ * llama.cpp - commit 46e3556e01b824e52395fb050b29804b6cff2a7c - do not edit this file
  *
  * MIT License
  *

llama/ggml-cuda/acc.cuh

 /**
- * llama.cpp - commit ba1cb19cdd0d92e012e0f6e009e0620f854b6afd - do not edit this file
+ * llama.cpp - commit 46e3556e01b824e52395fb050b29804b6cff2a7c - do not edit this file
  *
  * MIT License
  *

llama/ggml-cuda/arange.cu

 /**
- * llama.cpp - commit ba1cb19cdd0d92e012e0f6e009e0620f854b6afd - do not edit this file
+ * llama.cpp - commit 46e3556e01b824e52395fb050b29804b6cff2a7c - do not edit this file
  *
  * MIT License
  *

llama/ggml-cuda/arange.cuh

 /**
- * llama.cpp - commit ba1cb19cdd0d92e012e0f6e009e0620f854b6afd - do not edit this file
+ * llama.cpp - commit 46e3556e01b824e52395fb050b29804b6cff2a7c - do not edit this file
  *
  * MIT License
  *

llama/ggml-cuda/argmax.cu

 /**
- * llama.cpp - commit ba1cb19cdd0d92e012e0f6e009e0620f854b6afd - do not edit this file
+ * llama.cpp - commit 46e3556e01b824e52395fb050b29804b6cff2a7c - do not edit this file
  *
  * MIT License
  *

llama/ggml-cuda/argmax.cuh

 /**
- * llama.cpp - commit ba1cb19cdd0d92e012e0f6e009e0620f854b6afd - do not edit this file
+ * llama.cpp - commit 46e3556e01b824e52395fb050b29804b6cff2a7c - do not edit this file
  *
  * MIT License
  *

llama/ggml-cuda/argsort.cu

 /**
- * llama.cpp - commit ba1cb19cdd0d92e012e0f6e009e0620f854b6afd - do not edit this file
+ * llama.cpp - commit 46e3556e01b824e52395fb050b29804b6cff2a7c - do not edit this file
  *
  * MIT License
  *

llama/ggml-cuda/argsort.cuh

 /**
- * llama.cpp - commit ba1cb19cdd0d92e012e0f6e009e0620f854b6afd - do not edit this file
+ * llama.cpp - commit 46e3556e01b824e52395fb050b29804b6cff2a7c - do not edit this file
  *
  * MIT License
  *

llama/ggml-cuda/binbcast.cu

 /**
- * llama.cpp - commit ba1cb19cdd0d92e012e0f6e009e0620f854b6afd - do not edit this file
+ * llama.cpp - commit 46e3556e01b824e52395fb050b29804b6cff2a7c - do not edit this file
  *
  * MIT License
  *

llama/ggml-cuda/binbcast.cuh

 /**
- * llama.cpp - commit ba1cb19cdd0d92e012e0f6e009e0620f854b6afd - do not edit this file
+ * llama.cpp - commit 46e3556e01b824e52395fb050b29804b6cff2a7c - do not edit this file
  *
  * MIT License
  *

llama/ggml-cuda/clamp.cu

 /**
- * llama.cpp - commit ba1cb19cdd0d92e012e0f6e009e0620f854b6afd - do not edit this file
+ * llama.cpp - commit 46e3556e01b824e52395fb050b29804b6cff2a7c - do not edit this file
  *
  * MIT License
  *

llama/ggml-cuda/clamp.cuh

 /**
- * llama.cpp - commit ba1cb19cdd0d92e012e0f6e009e0620f854b6afd - do not edit this file
+ * llama.cpp - commit 46e3556e01b824e52395fb050b29804b6cff2a7c - do not edit this file
  *
  * MIT License
  *