From 9e1bda931599db4cc6ca0deb54a5a304ab96ed21 Mon Sep 17 00:00:00 2001
From: Iwan Kawrakow <iwan.kawrakow@gmail.com>
Date: Wed, 20 Mar 2024 19:39:15 +0100
Subject: [PATCH] k_cache: add newly added supported types to llama-bench and
 CUDA supports_op

---
 examples/llama-bench/llama-bench.cpp | 3 +++
 ggml-cuda.cu                         | 9 +++++++++
 2 files changed, 12 insertions(+)

diff --git a/examples/llama-bench/llama-bench.cpp b/examples/llama-bench/llama-bench.cpp
index 4cb230804..82413b79d 100644
--- a/examples/llama-bench/llama-bench.cpp
+++ b/examples/llama-bench/llama-bench.cpp
@@ -249,6 +249,9 @@ static ggml_type ggml_type_from_name(const std::string & s) {
     if (s == "q5_1") {
         return GGML_TYPE_Q5_1;
     }
+    if (s == "iq4_nl") {
+        return GGML_TYPE_IQ4_NL;
+    }
 
     return GGML_TYPE_COUNT;
 }
diff --git a/ggml-cuda.cu b/ggml-cuda.cu
index 14d021ac8..a2d4fae91 100644
--- a/ggml-cuda.cu
+++ b/ggml-cuda.cu
@@ -11465,6 +11465,15 @@ GGML_CALL static bool ggml_backend_cuda_supports_op(ggml_backend_t backend, cons
                 if (src0_type == GGML_TYPE_F32 && src1_type == GGML_TYPE_Q4_1) {
                     return true;
                 }
+                if (src0_type == GGML_TYPE_F32 && src1_type == GGML_TYPE_Q5_0) {
+                    return true;
+                }
+                if (src0_type == GGML_TYPE_F32 && src1_type == GGML_TYPE_Q5_1) {
+                    return true;
+                }
+                if (src0_type == GGML_TYPE_F32 && src1_type == GGML_TYPE_IQ4_NL) {
+                    return true;
+                }
                 if (src0_type == GGML_TYPE_F16 && src1_type == GGML_TYPE_F16) {
                     return true;
                 }