diff --git a/ggml/src/ggml-metal/ggml-metal-context.m b/ggml/src/ggml-metal/ggml-metal-context.m
index 5d3a8ce412ac..48b32b7cbf97 100644
--- a/ggml/src/ggml-metal/ggml-metal-context.m
+++ b/ggml/src/ggml-metal/ggml-metal-context.m
@@ -130,7 +130,8 @@ ggml_metal_t ggml_metal_init(ggml_metal_device_t dev) {
     res->d_queue = dispatch_queue_create("ggml-metal", DISPATCH_QUEUE_CONCURRENT);
 
     res->use_fusion      = getenv("GGML_METAL_FUSION_DISABLE") == nil;
-    res->use_concurrency = getenv("GGML_METAL_CONCURRENCY_DISABLE") == nil;
+    // non-UMA GPUs (AMD discrete) produce incorrect results with concurrent encoders
+    res->use_concurrency = props_dev->has_unified_memory && getenv("GGML_METAL_CONCURRENCY_DISABLE") == nil;
 
     {
         const char * val = getenv("GGML_METAL_GRAPH_DEBUG");
diff --git a/ggml/src/ggml-metal/ggml-metal-device.m b/ggml/src/ggml-metal/ggml-metal-device.m
index 4ea0bfb94da5..370d824765ab 100644
--- a/ggml/src/ggml-metal/ggml-metal-device.m
+++ b/ggml/src/ggml-metal/ggml-metal-device.m
@@ -1595,7 +1595,7 @@ void ggml_metal_buffer_memset_tensor(ggml_metal_buffer_t buf, struct ggml_tensor
             id<MTLBlitCommandEncoder> encoder = [cmd_buf blitCommandEncoder];
 
             [encoder fillBuffer:bid_dst.metal
-                          range:NSMakeRange(bid_dst.offs, bid_dst.offs + size)
+                          range:NSMakeRange(bid_dst.offs, size)
                           value:value];
 
             [encoder endEncoding];