vulkan: implement deferred_memset on UMA

Signed-off-by: Giuseppe Scrivano <[email protected]>

Author: giuseppe

ggml/src/ggml-vulkan/ggml-vulkan.cpp

@@ -1185,6 +1185,14 @@ struct vk_staging_memcpy {
     size_t n;
 };
 
+struct vk_staging_memset {
+    vk_staging_memset(void * _dst, uint32_t _val, size_t _n) : dst(_dst), val(_val), n(_n) {}
+
+    void * dst;
+    uint32_t val;
+    size_t n;
+};
+
 struct vk_context_struct {
     vk_submission * s;
     std::vector<vk_sequence> seqs;
@@ -1193,6 +1201,7 @@ struct vk_context_struct {
 
     std::vector<vk_staging_memcpy> in_memcpys;
     std::vector<vk_staging_memcpy> out_memcpys;
+    std::vector<vk_staging_memset> memsets;
 
     vk_command_pool * p {};
 };
@@ -5194,6 +5203,14 @@ static void deferred_memcpy(void * dst, const void * src, size_t size, std::vect
     }
 }
 
+static void deferred_memset(void * dst, uint32_t val, size_t size, std::vector<vk_staging_memset>* memsets = nullptr) {
+    if (memsets == nullptr) {
+        memset(dst, val, size);
+    } else {
+        memsets->emplace_back(dst, val, size);
+    }
+}
+
 static void ggml_vk_ensure_sync_staging_buffer(vk_device& device, size_t size) {
     if (device->sync_staging == nullptr || device->sync_staging->size < size) {
         VK_LOG_MEMORY("ggml_vk_ensure_sync_staging_buffer(" << size << ")");
@@ -5389,6 +5406,10 @@ static void ggml_vk_buffer_write_2d(vk_buffer& dst, size_t offset, const void *
             memcpy(cpy.dst, cpy.src, cpy.n);
         }
 
+        for (auto& mset : subctx->memsets) {
+            memset(mset.dst, mset.val, mset.n);
+        }
+
         ggml_vk_submit(subctx, dst->device->fence);
         VK_CHECK(dst->device->device.waitForFences({ dst->device->fence }, true, UINT64_MAX), "vk_buffer_write_2d waitForFences");
         dst->device->device.resetFences({ dst->device->fence });
@@ -5528,6 +5549,13 @@ static void ggml_vk_buffer_copy(vk_buffer& dst, size_t dst_offset, vk_buffer& sr
 static void ggml_vk_buffer_memset_async(vk_context& ctx, vk_buffer& dst, size_t offset, uint32_t c, size_t size) {
     VK_LOG_DEBUG("ggml_vk_buffer_memset_async(" << offset << ", " << c << ", " << size << ")");
 
+    if (dst->memory_property_flags & vk::MemoryPropertyFlagBits::eHostVisible &&
+        dst->device->uma) {
+        deferred_memset((uint8_t*)dst->ptr + offset, c, size, &ctx->memsets);
+        return;
+    }
+
+    // Fall back to GPU fillBuffer for non-UMA or non-host-visible buffers
     ctx->s->buffer.fillBuffer(dst->buffer, offset, size, c);
 }
 
@@ -11174,6 +11202,10 @@ static bool ggml_vk_compute_forward(ggml_backend_vk_context * ctx, ggml_cgraph *
             memcpy(cpy.dst, cpy.src, cpy.n);
         }
 
+        for (auto& mset : subctx->memsets) {
+            memset(mset.dst, mset.val, mset.n);
+        }
+
         if (almost_ready && !ctx->almost_ready_fence_pending && !use_fence) {
             ggml_vk_submit(subctx, ctx->almost_ready_fence);
             ctx->almost_ready_fence_pending = true;
@@ -11196,6 +11228,7 @@ static bool ggml_vk_compute_forward(ggml_backend_vk_context * ctx, ggml_cgraph *
         }
         subctx->in_memcpys.clear();
         subctx->out_memcpys.clear();
+        subctx->memsets.clear();
     }
 
     return true;