From 906aef3da84894d4b59e4f94d5fe69bc6fd0f01b Mon Sep 17 00:00:00 2001
From: Kevin Hannon <kehannon@redhat.com>
Date: Tue, 17 Mar 2026 13:16:49 -0400
Subject: [PATCH] ggml-blas: set mkl threads from thread context (llama/20602)

* ggml blas: set mkl threads from thread context

* add code to run blas locally
---
 ggml/src/ggml-blas/ggml-blas.cpp | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/ggml/src/ggml-blas/ggml-blas.cpp b/ggml/src/ggml-blas/ggml-blas.cpp
index 5de64b81..e7a1763b 100644
--- a/ggml/src/ggml-blas/ggml-blas.cpp
+++ b/ggml/src/ggml-blas/ggml-blas.cpp
@@ -121,6 +121,8 @@ static void ggml_backend_blas_mul_mat(ggml_backend_blas_context * ctx, struct gg
     bli_thread_set_num_threads(ctx->n_threads);
 #elif defined(GGML_BLAS_USE_NVPL)
     nvpl_blas_set_num_threads(ctx->n_threads);
+#elif defined(GGML_BLAS_USE_MKL)
+    mkl_set_num_threads(ctx->n_threads);
 #endif
 
     for (int64_t i13 = 0; i13 < ne13; i13++) {