From 3cf0f5468fb5dcea17aff4d97cc097267bf7a4ce Mon Sep 17 00:00:00 2001
From: Kawrakow <iwankawrakow@gmail.com>
Date: Fri, 19 Jun 2026 15:24:24 +0000
Subject: [PATCH] Also these

---
 ggml/src/ggml.c      |  4 ----
 src/llama-context.h  |  3 ---
 src/llama-dflash.cpp |  5 -----
 src/llama.cpp        | 24 ------------------------
 4 files changed, 36 deletions(-)

diff --git a/ggml/src/ggml.c b/ggml/src/ggml.c
index 2c1e3667..a865aef6 100644
--- a/ggml/src/ggml.c
+++ b/ggml/src/ggml.c
@@ -30051,11 +30051,7 @@ int ggml_cpu_has_wasm_simd(void) {
 }
 
 int ggml_cpu_has_blas(void) {
-#if defined(GGML_USE_BLAS) || defined(GGML_USE_CUDA) || defined(GGML_USE_VULKAN) || defined(GGML_USE_SYCL)
-    return 1;
-#else
     return 0;
-#endif
 }
 
 int ggml_cpu_has_cuda(void) {
diff --git a/src/llama-context.h b/src/llama-context.h
index b8a2b4cc..5d38a12a 100644
--- a/src/llama-context.h
+++ b/src/llama-context.h
@@ -218,9 +218,6 @@ struct llama_context {
     std::vector<ggml_backend_t> backends;
 #ifdef GGML_USE_METAL
     ggml_backend_t backend_metal = nullptr;
-#endif
-#ifdef GGML_USE_BLAS
-    ggml_backend_t backend_blas = nullptr;
 #endif
     ggml_backend_t backend_cpu = nullptr;
 
diff --git a/src/llama-dflash.cpp b/src/llama-dflash.cpp
index aa912a96..549a8dd1 100644
--- a/src/llama-dflash.cpp
+++ b/src/llama-dflash.cpp
@@ -248,11 +248,6 @@ static void llama_graph_compute_sched(
         ggml_backend_cpu_set_n_threads(lctx.backend_cpu, n_threads);
         ggml_backend_cpu_set_abort_callback(lctx.backend_cpu, lctx.abort_callback, lctx.abort_callback_data);
     }
-#ifdef GGML_USE_BLAS
-    if (lctx.backend_blas != nullptr) {
-        ggml_backend_blas_set_n_threads(lctx.backend_blas, n_threads);
-    }
-#endif
 
     ggml_backend_sched_graph_compute_async(sched, gf);
 }
diff --git a/src/llama.cpp b/src/llama.cpp
index 0a20563f..198d4e45 100644
--- a/src/llama.cpp
+++ b/src/llama.cpp
@@ -51,10 +51,6 @@ void llama_set_mtp_target_context(struct llama_context * ctx, struct llama_conte
 #   include "ggml-cann.h"
 #endif
 
-#ifdef GGML_USE_BLAS
-#  include "ggml-blas.h"
-#endif
-
 #ifdef GGML_USE_METAL
 #  include "ggml-metal.h"
 #endif
@@ -5010,11 +5006,6 @@ static void llama_graph_compute(
         ggml_backend_cpu_set_n_threads(lctx.backend_cpu, n_threads);
         ggml_backend_cpu_set_abort_callback(lctx.backend_cpu, lctx.abort_callback, lctx.abort_callback_data);
     }
-#ifdef GGML_USE_BLAS
-    if (lctx.backend_blas != nullptr) {
-        ggml_backend_blas_set_n_threads(lctx.backend_blas, n_threads);
-    }
-#endif
 
     ggml_backend_sched_graph_compute_async(lctx.sched, gf);
 
@@ -5036,11 +5027,6 @@ static void llama_graph_compute_sched(
         ggml_backend_cpu_set_n_threads(lctx.backend_cpu, n_threads);
         ggml_backend_cpu_set_abort_callback(lctx.backend_cpu, lctx.abort_callback, lctx.abort_callback_data);
     }
-#ifdef GGML_USE_BLAS
-    if (lctx.backend_blas != nullptr) {
-        ggml_backend_blas_set_n_threads(lctx.backend_blas, n_threads);
-    }
-#endif
 
     ggml_backend_sched_graph_compute_async(sched, gf);
 }
@@ -7223,15 +7209,6 @@ struct llama_context * llama_init_from_model(
     }
 #endif
 
-#ifdef GGML_USE_BLAS
-        ctx->backend_blas = ggml_backend_blas_init();
-        if (ctx->backend_blas == nullptr) {
-            LLAMA_LOG_WARN("%s: failed to initialize BLAS backend\n", __func__);
-        } else {
-            ggml_backend_add_from_device(ctx, ctx->backend_blas);
-        }
-#endif
-
 #if defined(GGML_USE_RPC)
         if (model->n_gpu_layers > 0) {
             for (const auto & device : model->rpc_servers) {
@@ -10988,7 +10965,6 @@ const char * llama_print_system_info(void) {
     s += "F16C = "        + std::to_string(ggml_cpu_has_f16c())        + " | ";
     s += "FP16_VA = "     + std::to_string(ggml_cpu_has_fp16_va())     + " | ";
     s += "WASM_SIMD = "   + std::to_string(ggml_cpu_has_wasm_simd())   + " | ";
-    s += "BLAS = "        + std::to_string(ggml_cpu_has_blas())        + " | ";
     s += "SSE3 = "        + std::to_string(ggml_cpu_has_sse3())        + " | ";
     s += "SSSE3 = "       + std::to_string(ggml_cpu_has_ssse3())       + " | ";
     s += "VSX = "         + std::to_string(ggml_cpu_has_vsx())         + " | ";